Key finding: PV A-fragment layout is IDENTICAL for (128,128)/(128,32)/(128,16) PV. Bug is NOT TMEM alias. cta_tile_shape_mnk wrong for non-(128,128) PV. V SMEM and O C-fragment sizes look correct. Debugging V/epilogue paths.

tests/test_128_32_ctafix.py

@@ -0,0 +1,386 @@
+"""
+Minimal PV-only test: Load P from GMEM to TMEM via QK-style MMA, then PV from TMEM.
+Step 1: QK MMA writes FP32 S to TMEM (we know this works)
+Step 2: Softmax packing writes BF16 P to TMEM (test this)
+Step 3: PV MMA reads BF16 P from TMEM and V from SMEM, produces O
+
+But to isolate the bug, let me test just the PV MMA in isolation.
+I'll write known BF16 values to TMEM using the softmax packing path,
+then immediately read them back using the PV A-fragment path,
+and compare.
+
+Actually, the simplest isolation test:
+1. Do QK MMA to get S in TMEM (cosine 0.999999 verified)
+2. Do softmax packing: S → P in TMEM (at offset 32)
+3. Skip PV entirely — read P from TMEM using the C-fragment composition LOAD path
+4. Output P to GMEM and compare against S.to(BF16)
+
+This tests whether the softmax packing writes P correctly to the same TMEM
+that the PV would read from.
+
+But we can't easily read P from TMEM using the standard epilogue path
+because the epilogue expects FP32 accumulator data.
+
+Alternative: Use the PV MMA with V=I (identity). If P is correct,
+then P @ I = P. But V needs to be MN-major and (128, 128), not (128, 64).
+The output would be (128, 128) which doesn't match our (128, 64) c tensor.
+
+Let me use V that selects the first 64 columns: V[k, n] = delta(k, n) for k in [0,63].
+This gives P @ V = P[:, :64], and the output is (128, 64).
+But V is (128, 128) in the MMA K,N dims. V[k, n] for k in [0,127], n in [0,63].
+Hmm, this is getting complicated. Let me just do the identity approach with a (128, 128) output.
+"""
+import torch, cutlass, cutlass.cute as cute, cutlass.utils as utils, cutlass.pipeline as pipeline
+from cutlass.cute.nvgpu import cpasync, tcgen05
+from cutlass import Float32, BFloat16, Int32, Boolean, const_expr
+from cutlass.utils import LayoutEnum
+from cutlass.utils.tmem_allocator import find_tmem_tensor_col_offset
+import cuda.bindings.driver as cuda
+import cutlass.torch as ct
+
+
+class Native32Kernel:
+    """QK + softmax packing + PV with V=I to isolate PV MMA correctness.
+    Output should be P = S.to(BF16), i.e. (Q@K^T).bfloat16()
+    With V=I, O = P @ I = P.
+    But V is (K=128, N=128) in the MMA. We need a 128x128 identity in MN-major.
+    Output tensor is (128, 128).
+    """
+    def __init__(self, mma_tiler_mn):
+        self.acc_dtype = Float32; self.qk_acc_dtype = Float32
+        self.q_dtype = BFloat16; self.o_dtype = BFloat16; self.c_dtype = BFloat16
+        self.mma_tiler_mn = mma_tiler_mn; self.mma_tiler = (*mma_tiler_mn, 1)
+        self.use_2cta_instrs = False  # needed by epilogue_tma_store
+        self.epilog_sync_bar_id = 1  # needed by epilogue_tma_store
+        self.cluster_shape_mn = (1, 1)
+        self.cta_group = tcgen05.CtaGroup.ONE
+        self.epilogue_warp_id = (0, 1, 2, 3)
+        self.mma_warp_id = 4; self.tma_warp_id = 5
+        self.threads_per_cta = 192
+        self.num_c_stage = 2
+
+    def _setup(self, qk_mma, pv_mma):
+        qk_inst_k = int(cute.size(qk_mma.shape_mnk, mode=[2]))
+        self.qk_mma_tiler = (*self.mma_tiler_mn, qk_inst_k * 4)
+        # PV with V=I: output is (128, 128), same as QK
+        self.pv_mma_tiler = (self.qk_mma_tiler[0], qk_inst_k, self.qk_mma_tiler[1])
+        # pv_mma_tiler = (128, 128, 128) since V is 128x128
+        self.mma_tiler = self.qk_mma_tiler
+
+        self.cluster_layout_vmnk = cute.tiled_divide(cute.make_layout((1,1,1)), (qk_mma.thr_id.shape,))
+        self.cta_tile_shape_mnk = (
+            self.qk_mma_tiler[0] // cute.size(qk_mma.thr_id.shape),
+            self.qk_mma_tiler[1], self.qk_mma_tiler[2])
+        self.c_layout = LayoutEnum.ROW_MAJOR
+        self.epi_tile = utils.sm100.compute_epilogue_tile_shape(
+            (self.pv_mma_tiler[0], self.pv_mma_tiler[1], self.pv_mma_tiler[2]), False, self.c_layout, self.o_dtype)
+        self.num_ab_stage = 1; self.num_acc_stage = 1
+
+        self.a_smem_s = utils.sm100.make_smem_layout_a(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.b_smem_s = utils.sm100.make_smem_layout_b(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.v_smem_s = utils.sm100.make_smem_layout_b(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.p_tmem_s = utils.sm100.make_smem_layout_a(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.c_smem_s = utils.sm100.make_smem_layout_epi(self.o_dtype, self.c_layout, self.epi_tile, 2)
+
+        qk_thr = qk_mma.get_slice(0)
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        s_cols = find_tmem_tensor_col_offset(tStS)
+        pv_thr = pv_mma.get_slice(0)
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        o_cols = find_tmem_tensor_col_offset(tOtO)
+
+        self.tilePlikeFP32 = self.qk_mma_tiler[1] // Float32.width * self.o_dtype.width
+        self.tmem_s0_offset = 0
+        self.tmem_p0_offset = 32
+        self.tmem_o0_offset = s_cols
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+        self.num_tmem_alloc_cols = utils.get_num_tmem_alloc_cols([tCtS_fake, tCtO_fake], arch="sm_100")
+
+        # ⛔⛔⛔ CRITICAL: num_tma_load_bytes MUST include ALL TMA-loaded tensors (Q + K + V). Missing V → DEADLOCK. See FOOTGUN #0 in README.
+        a_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        b_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        self.num_tma_load_bytes = (
+            cute.size_in_bytes(self.q_dtype, a_smem) + cute.size_in_bytes(self.q_dtype, b_smem) +
+            cute.size_in_bytes(self.q_dtype, v_smem)
+        ) * cute.size(qk_mma.thr_id.shape)
+
+    @cute.jit
+    def __call__(self, q, k, v, c, stream):
+        self.q_dtype = q.element_type; self.o_dtype = c.element_type; self.c_dtype = self.o_dtype
+        self.a_major = LayoutEnum.from_tensor(q).mma_major_mode()
+        self.b_major = LayoutEnum.from_tensor(k).mma_major_mode()
+        self.v_major = LayoutEnum.from_tensor(v).mma_major_mode()
+        self.c_layout = LayoutEnum.from_tensor(c)
+
+        qk_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, self.a_major, self.b_major,
+            self.qk_acc_dtype, self.cta_group, self.mma_tiler_mn, tcgen05.OperandSource.SMEM)
+        # PV with 128x128 output (V=I)
+        pv_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, cute.nvgpu.OperandMajorMode.K, self.v_major,
+            self.qk_acc_dtype, self.cta_group, (128, 32), tcgen05.OperandSource.TMEM)
+        self._setup(qk_mma, pv_mma)
+
+        q_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        k_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        tma_q, tma_tq = cute.nvgpu.make_tiled_tma_atom_A(
+            utils.sm100.cluster_shape_to_tma_atom_A(self.cluster_shape_mn, qk_mma.thr_id),
+            q, q_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_k, tma_tk = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, qk_mma.thr_id),
+            k, k_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_v, tma_tv = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, pv_mma.thr_id),
+            v, v_smem, self.pv_mma_tiler, pv_mma, self.cluster_layout_vmnk.shape)
+        epi_smem = cute.select(self.c_smem_s, mode=[0, 1])
+        tma_c, tma_tc = cpasync.make_tiled_tma_atom(cpasync.CopyBulkTensorTileS2GOp(), c, epi_smem, self.epi_tile)
+
+        self._kernel(qk_mma, pv_mma, tma_q, tma_tq, tma_k, tma_tk, tma_v, tma_tv,
+            tma_c, tma_tc, self.cluster_layout_vmnk,
+            self.a_smem_s, self.b_smem_s, self.v_smem_s, self.p_tmem_s, self.c_smem_s, self.epi_tile
+        ).launch(grid=(1,1,1), block=[self.threads_per_cta,1,1], stream=stream)
+
+    @cute.kernel
+    def _kernel(self, qk_mma, pv_mma, tma_q, mQ, tma_k, mK, tma_v, mV,
+                tma_c, mC, cl_vmnk, a_smem_s, b_smem_s, v_smem_s, p_tmem_s, c_smem_s, epi_tile):
+        warp_idx = cute.arch.make_warp_uniform(cute.arch.warp_idx())
+        tidx, _, _ = cute.arch.thread_idx()
+        use_2cta = cute.size(qk_mma.thr_id.shape) == 2
+
+        if warp_idx == self.tma_warp_id:
+            cpasync.prefetch_descriptor(tma_q); cpasync.prefetch_descriptor(tma_k)
+            cpasync.prefetch_descriptor(tma_v); cpasync.prefetch_descriptor(tma_c)
+
+        @cute.struct
+        class SS:
+            ab_bar: cute.struct.MemRange[cutlass.Int64, self.num_ab_stage * 2]
+            mma_si_bar: cute.struct.MemRange[cutlass.Int64, 2]
+            acc_bar: cute.struct.MemRange[cutlass.Int64, self.num_acc_stage * 2]
+            tmem_dealloc: cutlass.Int64
+            holding: cutlass.Int32
+
+        smem = utils.SmemAllocator(); st = smem.allocate(SS)
+
+        ab_p, ab_c = pipeline.PipelineTmaUmma.create(
+            barrier_storage=st.ab_bar.data_ptr(), num_stages=self.num_ab_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 1),
+            tx_count=self.num_tma_load_bytes, cta_layout_vmnk=cl_vmnk, defer_sync=True
+        ).make_participants()
+
+        mma_si_prod, mma_si_cons = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.mma_si_bar.data_ptr(), num_stages=1,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id)),
+        ).make_participants()
+
+        acc_pipe = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.acc_bar.data_ptr(), num_stages=self.num_acc_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(
+                pipeline.Agent.Thread, len(self.epilogue_warp_id) * (2 if use_2cta else 1)),
+            cta_layout_vmnk=cl_vmnk, defer_sync=True)
+
+        tmem_bar = pipeline.NamedBarrier(barrier_id=2,
+            num_threads=32 * len((self.mma_warp_id, *self.epilogue_warp_id)))
+        tmem = utils.TmemAllocator(st.holding.ptr, barrier_for_retrieve=tmem_bar,
+            allocator_warp_id=self.epilogue_warp_id[0], is_two_cta=use_2cta,
+            two_cta_tmem_dealloc_mbar_ptr=st.tmem_dealloc.ptr)
+
+        pipeline.pipeline_init_arrive(cluster_shape_mn=cl_vmnk, is_relaxed=True)
+
+        sQ = smem.allocate_tensor(element_type=self.q_dtype, layout=a_smem_s.outer, byte_alignment=128, swizzle=a_smem_s.inner)
+        sK = smem.allocate_tensor(element_type=self.q_dtype, layout=b_smem_s.outer, byte_alignment=128, swizzle=b_smem_s.inner)
+        sV = smem.allocate_tensor(element_type=self.q_dtype, layout=v_smem_s.outer, byte_alignment=128, swizzle=v_smem_s.inner)
+        sC = smem.allocate_tensor(element_type=self.o_dtype, layout=c_smem_s.outer, byte_alignment=128, swizzle=c_smem_s.inner)
+
+        gQ = cute.local_tile(mQ, cute.slice_(self.qk_mma_tiler, (None,0,None)), (None,None,None))
+        gK = cute.local_tile(mK, cute.slice_(self.qk_mma_tiler, (0,None,None)), (None,None,None))
+        gC = cute.local_tile(mC, cute.slice_(self.qk_mma_tiler, (None,None,0)), (None,None,None))
+        k_cnt = cute.size(gQ, mode=[3])
+
+        qk_thr = qk_mma.get_slice(0)
+        pv_thr = pv_mma.get_slice(0)
+        tCgQ = qk_thr.partition_A(gQ); tCgK = qk_thr.partition_B(gK); tCgC = qk_thr.partition_C(gC)
+        a_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,0,None,0)).shape)
+        tAsQ, tAgQ = cpasync.tma_partition(tma_q, 0, a_lay, cute.group_modes(sQ,0,3), cute.group_modes(tCgQ,0,3))
+        b_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,None,0,0)).shape)
+        tBsK, tBgK = cpasync.tma_partition(tma_k, 0, b_lay, cute.group_modes(sK,0,3), cute.group_modes(tCgK,0,3))
+        tAgQ = tAgQ[(None,0,None,0)]; tBgK = tBgK[(None,0,None,0)]
+
+        gV = cute.local_tile(mV, cute.slice_(self.pv_mma_tiler, (0,None,None)), (None,None,None))
+        tCgV = pv_thr.partition_B(gV)
+        tVsV, tVgV = cpasync.tma_partition(tma_v, 0, b_lay, cute.group_modes(sV,0,3), cute.group_modes(tCgV,0,3))
+        tVgV = tVgV[(None,0,None,0)]
+
+        tCrQ = qk_mma.make_fragment_A(sQ); tCrK = qk_mma.make_fragment_B(sK)
+        tCrV = pv_mma.make_fragment_B(sV)
+
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        tStS0 = cute.make_tensor(tStS.iterator + self.tmem_s0_offset, tStS.layout)
+
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        tOtO0 = cute.make_tensor(tOtO.iterator + self.tmem_o0_offset, tOtO.layout)
+
+        tP = cute.make_tensor(tStS.iterator, p_tmem_s.outer)
+        tOrP_base = pv_thr.make_fragment_A(tP)
+        tOrP = tOrP_base[(None, None, None, 0)]
+        tOrP0 = cute.make_tensor(
+            tOrP.iterator + self.qk_acc_dtype.width // self.q_dtype.width * self.tmem_p0_offset,
+            tOrP.layout)
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+
+        pipeline.pipeline_init_wait(cluster_shape_mn=cl_vmnk)
+
+        # ═══ TMA LOAD WARP ═══
+        if warp_idx == self.tma_warp_id:
+            ab_p.reset(); peek = ab_p.try_acquire()
+            for kt in cutlass.range(k_cnt, unroll=1):
+                h = ab_p.acquire_and_advance(peek)
+                cute.copy(tma_q, tAgQ[(None,h.count)], tAsQ[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_k, tBgK[(None,h.count)], tBsK[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_v, tVgV[(None,h.count)], tVsV[(None,h.index)], tma_bar_ptr=h.barrier)
+                peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_p.try_acquire()
+            ab_p.tail()
+
+        # ═══ MMA WARP ═══
+        if warp_idx == self.mma_warp_id:
+            tmem.wait_for_alloc()
+            ab_c.reset(); peek = ab_c.try_wait()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            acc_prod_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Producer, self.num_acc_stage)
+            acc_pipe.producer_acquire(acc_prod_st)
+            qk_mma.set(tcgen05.Field.ACCUMULATE, False)
+            for kt in range(k_cnt):
+                h = ab_c.wait_and_advance(peek)
+                nblk = cute.size(tCrQ, mode=[2])
+                for kb in cutlass.range(nblk, unroll_full=True):
+                    cute.gemm(qk_mma, tStS0, tCrQ[(None,None,kb,h.index)], tCrK[(None,None,kb,h.index)], tStS0)
+                    qk_mma.set(tcgen05.Field.ACCUMULATE, True)
+                h.release(); peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_c.try_wait()
+
+            cute.arch.fence_view_async_tmem_store()
+            s0_handle.commit()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            # PV MMA: P @ V where V=I → O = P
+            pv_mma.set(tcgen05.Field.ACCUMULATE, False)
+            tCrV_s = tCrV[(None, None, None, 0)]
+            nblk_pv = cute.size(tOrP0, mode=[2])
+            for kb in cutlass.range(nblk_pv, unroll_full=True):
+                cute.gemm(pv_mma, tOtO0, tOrP0[(None,None,kb)], tCrV_s[(None,None,kb)], tOtO0)
+                pv_mma.set(tcgen05.Field.ACCUMULATE, True)
+
+            acc_pipe.producer_commit(acc_prod_st)
+            acc_prod_st.advance()
+            acc_pipe.producer_tail(acc_prod_st)
+
+        # ═══ EPILOGUE WARPS ═══
+        if warp_idx < self.mma_warp_id:
+            tmem.allocate(self.num_tmem_alloc_cols)
+            tmem.wait_for_alloc()
+            tmem_ptr = tmem.retrieve_ptr(self.qk_acc_dtype)
+            sfw_idx = tidx % (32 * len(self.epilogue_warp_id))
+
+            tmem_load_atom = cute.make_copy_atom(
+                tcgen05.copy.Ld32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_load = tcgen05.make_tmem_copy(tmem_load_atom, tStS0)
+            thr_load = tiled_tmem_load.get_slice(sfw_idx)
+            tTMEM_LOADtS = thr_load.partition_S(tStS0)
+            cS = cute.make_identity_tensor((self.qk_mma_tiler[0], self.qk_mma_tiler[1]))
+            tScS = qk_thr.partition_C(cS)
+            tTMEM_LOADcS = thr_load.partition_D(tScS)
+
+            tStS_P_layout = cute.composition(tStS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tStS_P = cute.make_tensor(tStS.iterator + self.tmem_p0_offset, tStS_P_layout)
+            tmem_store_atom = cute.make_copy_atom(
+                tcgen05.copy.St32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_store = tcgen05.make_tmem_copy(tmem_store_atom, tStS_P)
+            thr_store = tiled_tmem_store.get_slice(sfw_idx)
+            tTMEM_STOREtS_x4 = thr_store.partition_D(tStS_P)
+            tScS_P_layout = cute.composition(tScS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tScS_P = cute.make_tensor(tScS.iterator, tScS_P_layout)
+            tTMEM_STOREcS = thr_store.partition_S(tScS_P)
+
+            si_handle = mma_si_cons.wait_and_advance()
+
+            tTMEM_LOADrS = cute.make_rmem_tensor(tTMEM_LOADcS.shape, self.qk_acc_dtype)
+            cute.copy(tiled_tmem_load, tTMEM_LOADtS, tTMEM_LOADrS)
+            tTMEM_STORErS_x4 = cute.make_rmem_tensor(tTMEM_STOREcS.shape, self.qk_acc_dtype)
+            tTMEM_STORErS_x4_e = cute.make_tensor(
+                cute.recast_ptr(tTMEM_STORErS_x4.iterator, dtype=self.q_dtype),
+                tTMEM_LOADrS.layout)
+            frg_cnt = 4
+            frg_tile = cute.size(tTMEM_LOADrS) // frg_cnt
+            tTMEM_LOADrS_frg = cute.logical_divide(tTMEM_LOADrS, cute.make_layout(frg_tile))
+            tTMEM_STORErS_x4_e_frg = cute.logical_divide(
+                tTMEM_STORErS_x4_e, cute.make_layout(frg_tile))
+            for j in range(frg_cnt):
+                s_vec = tTMEM_LOADrS_frg[None, j].load()
+                tTMEM_STORErS_x4_e_frg[None, j].store(s_vec.to(self.q_dtype))
+            cute.copy(tiled_tmem_store, tTMEM_STORErS_x4, tTMEM_STOREtS_x4)
+            cute.arch.fence_view_async_tmem_store()
+            si_handle.release()
+
+            # Output epilogue
+            tCtO_base = cute.make_tensor(tmem_ptr + self.tmem_o0_offset, tCtO_fake.layout)
+            acc_cons_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Consumer, self.num_acc_stage)
+            c_grp = pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id))
+            c_pipe = pipeline.PipelineTmaStore.create(num_stages=self.num_c_stage, producer_group=c_grp)
+            # Swap cta_tile to PV for epilogue
+            saved_cta = self.cta_tile_shape_mnk
+            self.cta_tile_shape_mnk = (int(self.pv_mma_tiler[0]), int(self.pv_mma_tiler[1]), int(self.pv_mma_tiler[2]))
+            acc_cons_st = utils.gemm.sm100.epilogue_tma_store(
+                self, tidx, warp_idx, tma_c, tCtO_base, sC, tCgC,
+                epi_tile, 0, const_expr(lambda x: x), (0,0,0), acc_cons_st, acc_pipe, c_pipe)
+            self.cta_tile_shape_mnk = saved_cta
+            c_pipe.producer_tail()
+            tmem.relinquish_alloc_permit()
+            tmem.free(tmem_ptr)
+
+
+def test():
+    torch.manual_seed(42)
+    m, n, head_dim = 128, 128, 64
+    q = torch.randn(m, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    k = torch.randn(n, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    # V = I[:32,:] MN-major: (32,128) with strides (1,32)
+    v_data = torch.eye(128, dtype=torch.bfloat16, device='cuda')
+    v = v_data[:32, :].as_strided((32, 128), (1, 32)).unsqueeze(-1)
+    c = torch.zeros(m, 32, 1, dtype=torch.bfloat16, device='cuda')
+
+    qf = q[:,:,0].float(); kf = k[:,:,0].float()
+    # V=I[:32,:] MN-major -> O = P[:,:32] = (Q@K^T).bf16()[:,:32]
+    ref = (qf @ kf.T).bfloat16().float()[:, :32]
+
+    mQ = ct.from_dlpack(q).mark_layout_dynamic(leading_dim=ct.get_leading_dim(q))
+    mK = ct.from_dlpack(k).mark_layout_dynamic(leading_dim=ct.get_leading_dim(k))
+    mV = ct.from_dlpack(v).mark_layout_dynamic(leading_dim=ct.get_leading_dim(v))
+    mC = ct.from_dlpack(c).mark_layout_dynamic(leading_dim=ct.get_leading_dim(c))
+    stream = cuda.CUstream(torch.cuda.current_stream().cuda_stream)
+    kernel = Native32Kernel(mma_tiler_mn=(128, 128))
+    print('Compiling...', flush=True)
+    compiled = cute.compile(kernel, mQ, mK, mV, mC, stream)
+    print('Running...', flush=True)
+    compiled(mQ, mK, mV, mC, stream)
+    torch.cuda.synchronize()
+    out = c[:,:,0].float()
+    cos = torch.nn.functional.cosine_similarity(out.flatten().unsqueeze(0), ref.flatten().unsqueeze(0)).item()
+    print('PV(128,32) ctafix: cosine {:.6f} {}'.format(cos, 'PASS' if cos >= 0.99 else 'FAIL'))
+
+
+if __name__ == '__main__':
+    test()

tests/test_128_32_ctafix2.py

@@ -0,0 +1,384 @@
+"""
+Minimal PV-only test: Load P from GMEM to TMEM via QK-style MMA, then PV from TMEM.
+Step 1: QK MMA writes FP32 S to TMEM (we know this works)
+Step 2: Softmax packing writes BF16 P to TMEM (test this)
+Step 3: PV MMA reads BF16 P from TMEM and V from SMEM, produces O
+
+But to isolate the bug, let me test just the PV MMA in isolation.
+I'll write known BF16 values to TMEM using the softmax packing path,
+then immediately read them back using the PV A-fragment path,
+and compare.
+
+Actually, the simplest isolation test:
+1. Do QK MMA to get S in TMEM (cosine 0.999999 verified)
+2. Do softmax packing: S → P in TMEM (at offset 32)
+3. Skip PV entirely — read P from TMEM using the C-fragment composition LOAD path
+4. Output P to GMEM and compare against S.to(BF16)
+
+This tests whether the softmax packing writes P correctly to the same TMEM
+that the PV would read from.
+
+But we can't easily read P from TMEM using the standard epilogue path
+because the epilogue expects FP32 accumulator data.
+
+Alternative: Use the PV MMA with V=I (identity). If P is correct,
+then P @ I = P. But V needs to be MN-major and (128, 128), not (128, 64).
+The output would be (128, 128) which doesn't match our (128, 64) c tensor.
+
+Let me use V that selects the first 64 columns: V[k, n] = delta(k, n) for k in [0,63].
+This gives P @ V = P[:, :64], and the output is (128, 64).
+But V is (128, 128) in the MMA K,N dims. V[k, n] for k in [0,127], n in [0,63].
+Hmm, this is getting complicated. Let me just do the identity approach with a (128, 128) output.
+"""
+import torch, cutlass, cutlass.cute as cute, cutlass.utils as utils, cutlass.pipeline as pipeline
+from cutlass.cute.nvgpu import cpasync, tcgen05
+from cutlass import Float32, BFloat16, Int32, Boolean, const_expr
+from cutlass.utils import LayoutEnum
+from cutlass.utils.tmem_allocator import find_tmem_tensor_col_offset
+import cuda.bindings.driver as cuda
+import cutlass.torch as ct
+
+
+class Native32Kernel:
+    """QK + softmax packing + PV with V=I to isolate PV MMA correctness.
+    Output should be P = S.to(BF16), i.e. (Q@K^T).bfloat16()
+    With V=I, O = P @ I = P.
+    But V is (K=128, N=128) in the MMA. We need a 128x128 identity in MN-major.
+    Output tensor is (128, 128).
+    """
+    def __init__(self, mma_tiler_mn):
+        self.acc_dtype = Float32; self.qk_acc_dtype = Float32
+        self.q_dtype = BFloat16; self.o_dtype = BFloat16; self.c_dtype = BFloat16
+        self.mma_tiler_mn = mma_tiler_mn; self.mma_tiler = (*mma_tiler_mn, 1)
+        self.use_2cta_instrs = False  # needed by epilogue_tma_store
+        self.epilog_sync_bar_id = 1  # needed by epilogue_tma_store
+        self.cluster_shape_mn = (1, 1)
+        self.cta_group = tcgen05.CtaGroup.ONE
+        self.epilogue_warp_id = (0, 1, 2, 3)
+        self.mma_warp_id = 4; self.tma_warp_id = 5
+        self.threads_per_cta = 192
+        self.num_c_stage = 2
+        self.pv_cta_tile = (128, 16, 128)  # Will be updated in _setup with int values
+
+    def _setup(self, qk_mma, pv_mma):
+        qk_inst_k = int(cute.size(qk_mma.shape_mnk, mode=[2]))
+        self.qk_mma_tiler = (*self.mma_tiler_mn, qk_inst_k * 4)
+        # PV with V=I: output is (128, 128), same as QK
+        self.pv_mma_tiler = (self.qk_mma_tiler[0], qk_inst_k, self.qk_mma_tiler[1])
+        self.pv_cta_tile = (int(self.pv_mma_tiler[0]), int(self.pv_mma_tiler[1]), int(self.pv_mma_tiler[2]))
+        # pv_mma_tiler = (128, 128, 128) since V is 128x128
+        self.mma_tiler = self.qk_mma_tiler
+
+        self.cluster_layout_vmnk = cute.tiled_divide(cute.make_layout((1,1,1)), (qk_mma.thr_id.shape,))
+        self.cta_tile_shape_mnk = (
+            self.qk_mma_tiler[0] // cute.size(qk_mma.thr_id.shape),
+            self.qk_mma_tiler[1], self.qk_mma_tiler[2])
+        self.c_layout = LayoutEnum.ROW_MAJOR
+        self.epi_tile = utils.sm100.compute_epilogue_tile_shape(
+            (self.pv_mma_tiler[0], self.pv_mma_tiler[1], self.pv_mma_tiler[2]), False, self.c_layout, self.o_dtype)
+        self.num_ab_stage = 1; self.num_acc_stage = 1
+
+        self.a_smem_s = utils.sm100.make_smem_layout_a(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.b_smem_s = utils.sm100.make_smem_layout_b(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.v_smem_s = utils.sm100.make_smem_layout_b(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.p_tmem_s = utils.sm100.make_smem_layout_a(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.c_smem_s = utils.sm100.make_smem_layout_epi(self.o_dtype, self.c_layout, self.epi_tile, 2)
+
+        qk_thr = qk_mma.get_slice(0)
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        s_cols = find_tmem_tensor_col_offset(tStS)
+        pv_thr = pv_mma.get_slice(0)
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        o_cols = find_tmem_tensor_col_offset(tOtO)
+
+        self.tilePlikeFP32 = self.qk_mma_tiler[1] // Float32.width * self.o_dtype.width
+        self.tmem_s0_offset = 0
+        self.tmem_p0_offset = 32
+        self.tmem_o0_offset = s_cols
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+        self.num_tmem_alloc_cols = utils.get_num_tmem_alloc_cols([tCtS_fake, tCtO_fake], arch="sm_100")
+
+        # ⛔⛔⛔ CRITICAL: num_tma_load_bytes MUST include ALL TMA-loaded tensors (Q + K + V). Missing V → DEADLOCK. See FOOTGUN #0 in README.
+        a_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        b_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        self.num_tma_load_bytes = (
+            cute.size_in_bytes(self.q_dtype, a_smem) + cute.size_in_bytes(self.q_dtype, b_smem) +
+            cute.size_in_bytes(self.q_dtype, v_smem)
+        ) * cute.size(qk_mma.thr_id.shape)
+
+    @cute.jit
+    def __call__(self, q, k, v, c, stream):
+        self.q_dtype = q.element_type; self.o_dtype = c.element_type; self.c_dtype = self.o_dtype
+        self.a_major = LayoutEnum.from_tensor(q).mma_major_mode()
+        self.b_major = LayoutEnum.from_tensor(k).mma_major_mode()
+        self.v_major = LayoutEnum.from_tensor(v).mma_major_mode()
+        self.c_layout = LayoutEnum.from_tensor(c)
+
+        qk_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, self.a_major, self.b_major,
+            self.qk_acc_dtype, self.cta_group, self.mma_tiler_mn, tcgen05.OperandSource.SMEM)
+        # PV with 128x128 output (V=I)
+        pv_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, cute.nvgpu.OperandMajorMode.K, self.v_major,
+            self.qk_acc_dtype, self.cta_group, (128, 32), tcgen05.OperandSource.TMEM)
+        self._setup(qk_mma, pv_mma)
+
+        q_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        k_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        tma_q, tma_tq = cute.nvgpu.make_tiled_tma_atom_A(
+            utils.sm100.cluster_shape_to_tma_atom_A(self.cluster_shape_mn, qk_mma.thr_id),
+            q, q_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_k, tma_tk = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, qk_mma.thr_id),
+            k, k_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_v, tma_tv = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, pv_mma.thr_id),
+            v, v_smem, self.pv_mma_tiler, pv_mma, self.cluster_layout_vmnk.shape)
+        epi_smem = cute.select(self.c_smem_s, mode=[0, 1])
+        tma_c, tma_tc = cpasync.make_tiled_tma_atom(cpasync.CopyBulkTensorTileS2GOp(), c, epi_smem, self.epi_tile)
+
+        self._kernel(qk_mma, pv_mma, tma_q, tma_tq, tma_k, tma_tk, tma_v, tma_tv,
+            tma_c, tma_tc, self.cluster_layout_vmnk,
+            self.a_smem_s, self.b_smem_s, self.v_smem_s, self.p_tmem_s, self.c_smem_s, self.epi_tile
+        ).launch(grid=(1,1,1), block=[self.threads_per_cta,1,1], stream=stream)
+
+    @cute.kernel
+    def _kernel(self, qk_mma, pv_mma, tma_q, mQ, tma_k, mK, tma_v, mV,
+                tma_c, mC, cl_vmnk, a_smem_s, b_smem_s, v_smem_s, p_tmem_s, c_smem_s, epi_tile):
+        warp_idx = cute.arch.make_warp_uniform(cute.arch.warp_idx())
+        tidx, _, _ = cute.arch.thread_idx()
+        use_2cta = cute.size(qk_mma.thr_id.shape) == 2
+
+        if warp_idx == self.tma_warp_id:
+            cpasync.prefetch_descriptor(tma_q); cpasync.prefetch_descriptor(tma_k)
+            cpasync.prefetch_descriptor(tma_v); cpasync.prefetch_descriptor(tma_c)
+
+        @cute.struct
+        class SS:
+            ab_bar: cute.struct.MemRange[cutlass.Int64, self.num_ab_stage * 2]
+            mma_si_bar: cute.struct.MemRange[cutlass.Int64, 2]
+            acc_bar: cute.struct.MemRange[cutlass.Int64, self.num_acc_stage * 2]
+            tmem_dealloc: cutlass.Int64
+            holding: cutlass.Int32
+
+        smem = utils.SmemAllocator(); st = smem.allocate(SS)
+
+        ab_p, ab_c = pipeline.PipelineTmaUmma.create(
+            barrier_storage=st.ab_bar.data_ptr(), num_stages=self.num_ab_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 1),
+            tx_count=self.num_tma_load_bytes, cta_layout_vmnk=cl_vmnk, defer_sync=True
+        ).make_participants()
+
+        mma_si_prod, mma_si_cons = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.mma_si_bar.data_ptr(), num_stages=1,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id)),
+        ).make_participants()
+
+        acc_pipe = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.acc_bar.data_ptr(), num_stages=self.num_acc_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(
+                pipeline.Agent.Thread, len(self.epilogue_warp_id) * (2 if use_2cta else 1)),
+            cta_layout_vmnk=cl_vmnk, defer_sync=True)
+
+        tmem_bar = pipeline.NamedBarrier(barrier_id=2,
+            num_threads=32 * len((self.mma_warp_id, *self.epilogue_warp_id)))
+        tmem = utils.TmemAllocator(st.holding.ptr, barrier_for_retrieve=tmem_bar,
+            allocator_warp_id=self.epilogue_warp_id[0], is_two_cta=use_2cta,
+            two_cta_tmem_dealloc_mbar_ptr=st.tmem_dealloc.ptr)
+
+        pipeline.pipeline_init_arrive(cluster_shape_mn=cl_vmnk, is_relaxed=True)
+
+        sQ = smem.allocate_tensor(element_type=self.q_dtype, layout=a_smem_s.outer, byte_alignment=128, swizzle=a_smem_s.inner)
+        sK = smem.allocate_tensor(element_type=self.q_dtype, layout=b_smem_s.outer, byte_alignment=128, swizzle=b_smem_s.inner)
+        sV = smem.allocate_tensor(element_type=self.q_dtype, layout=v_smem_s.outer, byte_alignment=128, swizzle=v_smem_s.inner)
+        sC = smem.allocate_tensor(element_type=self.o_dtype, layout=c_smem_s.outer, byte_alignment=128, swizzle=c_smem_s.inner)
+
+        gQ = cute.local_tile(mQ, cute.slice_(self.qk_mma_tiler, (None,0,None)), (None,None,None))
+        gK = cute.local_tile(mK, cute.slice_(self.qk_mma_tiler, (0,None,None)), (None,None,None))
+        gC = cute.local_tile(mC, cute.slice_(self.qk_mma_tiler, (None,None,0)), (None,None,None))
+        k_cnt = cute.size(gQ, mode=[3])
+
+        qk_thr = qk_mma.get_slice(0)
+        pv_thr = pv_mma.get_slice(0)
+        tCgQ = qk_thr.partition_A(gQ); tCgK = qk_thr.partition_B(gK); tCgC = qk_thr.partition_C(gC)
+        a_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,0,None,0)).shape)
+        tAsQ, tAgQ = cpasync.tma_partition(tma_q, 0, a_lay, cute.group_modes(sQ,0,3), cute.group_modes(tCgQ,0,3))
+        b_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,None,0,0)).shape)
+        tBsK, tBgK = cpasync.tma_partition(tma_k, 0, b_lay, cute.group_modes(sK,0,3), cute.group_modes(tCgK,0,3))
+        tAgQ = tAgQ[(None,0,None,0)]; tBgK = tBgK[(None,0,None,0)]
+
+        gV = cute.local_tile(mV, cute.slice_(self.pv_mma_tiler, (0,None,None)), (None,None,None))
+        tCgV = pv_thr.partition_B(gV)
+        tVsV, tVgV = cpasync.tma_partition(tma_v, 0, b_lay, cute.group_modes(sV,0,3), cute.group_modes(tCgV,0,3))
+        tVgV = tVgV[(None,0,None,0)]
+
+        tCrQ = qk_mma.make_fragment_A(sQ); tCrK = qk_mma.make_fragment_B(sK)
+        tCrV = pv_mma.make_fragment_B(sV)
+
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        tStS0 = cute.make_tensor(tStS.iterator + self.tmem_s0_offset, tStS.layout)
+
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        tOtO0 = cute.make_tensor(tOtO.iterator + self.tmem_o0_offset, tOtO.layout)
+
+        tP = cute.make_tensor(tStS.iterator, p_tmem_s.outer)
+        tOrP_base = pv_thr.make_fragment_A(tP)
+        tOrP = tOrP_base[(None, None, None, 0)]
+        tOrP0 = cute.make_tensor(
+            tOrP.iterator + self.qk_acc_dtype.width // self.q_dtype.width * self.tmem_p0_offset,
+            tOrP.layout)
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+
+        pipeline.pipeline_init_wait(cluster_shape_mn=cl_vmnk)
+
+        # ═══ TMA LOAD WARP ═══
+        if warp_idx == self.tma_warp_id:
+            ab_p.reset(); peek = ab_p.try_acquire()
+            for kt in cutlass.range(k_cnt, unroll=1):
+                h = ab_p.acquire_and_advance(peek)
+                cute.copy(tma_q, tAgQ[(None,h.count)], tAsQ[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_k, tBgK[(None,h.count)], tBsK[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_v, tVgV[(None,h.count)], tVsV[(None,h.index)], tma_bar_ptr=h.barrier)
+                peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_p.try_acquire()
+            ab_p.tail()
+
+        # ═══ MMA WARP ═══
+        if warp_idx == self.mma_warp_id:
+            tmem.wait_for_alloc()
+            ab_c.reset(); peek = ab_c.try_wait()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            acc_prod_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Producer, self.num_acc_stage)
+            acc_pipe.producer_acquire(acc_prod_st)
+            qk_mma.set(tcgen05.Field.ACCUMULATE, False)
+            for kt in range(k_cnt):
+                h = ab_c.wait_and_advance(peek)
+                nblk = cute.size(tCrQ, mode=[2])
+                for kb in cutlass.range(nblk, unroll_full=True):
+                    cute.gemm(qk_mma, tStS0, tCrQ[(None,None,kb,h.index)], tCrK[(None,None,kb,h.index)], tStS0)
+                    qk_mma.set(tcgen05.Field.ACCUMULATE, True)
+                h.release(); peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_c.try_wait()
+
+            cute.arch.fence_view_async_tmem_store()
+            s0_handle.commit()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            # PV MMA: P @ V where V=I → O = P
+            pv_mma.set(tcgen05.Field.ACCUMULATE, False)
+            tCrV_s = tCrV[(None, None, None, 0)]
+            nblk_pv = cute.size(tOrP0, mode=[2])
+            for kb in cutlass.range(nblk_pv, unroll_full=True):
+                cute.gemm(pv_mma, tOtO0, tOrP0[(None,None,kb)], tCrV_s[(None,None,kb)], tOtO0)
+                pv_mma.set(tcgen05.Field.ACCUMULATE, True)
+
+            acc_pipe.producer_commit(acc_prod_st)
+            acc_prod_st.advance()
+            acc_pipe.producer_tail(acc_prod_st)
+
+        # ═══ EPILOGUE WARPS ═══
+        if warp_idx < self.mma_warp_id:
+            tmem.allocate(self.num_tmem_alloc_cols)
+            tmem.wait_for_alloc()
+            tmem_ptr = tmem.retrieve_ptr(self.qk_acc_dtype)
+            sfw_idx = tidx % (32 * len(self.epilogue_warp_id))
+
+            tmem_load_atom = cute.make_copy_atom(
+                tcgen05.copy.Ld32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_load = tcgen05.make_tmem_copy(tmem_load_atom, tStS0)
+            thr_load = tiled_tmem_load.get_slice(sfw_idx)
+            tTMEM_LOADtS = thr_load.partition_S(tStS0)
+            cS = cute.make_identity_tensor((self.qk_mma_tiler[0], self.qk_mma_tiler[1]))
+            tScS = qk_thr.partition_C(cS)
+            tTMEM_LOADcS = thr_load.partition_D(tScS)
+
+            tStS_P_layout = cute.composition(tStS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tStS_P = cute.make_tensor(tStS.iterator + self.tmem_p0_offset, tStS_P_layout)
+            tmem_store_atom = cute.make_copy_atom(
+                tcgen05.copy.St32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_store = tcgen05.make_tmem_copy(tmem_store_atom, tStS_P)
+            thr_store = tiled_tmem_store.get_slice(sfw_idx)
+            tTMEM_STOREtS_x4 = thr_store.partition_D(tStS_P)
+            tScS_P_layout = cute.composition(tScS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tScS_P = cute.make_tensor(tScS.iterator, tScS_P_layout)
+            tTMEM_STOREcS = thr_store.partition_S(tScS_P)
+
+            si_handle = mma_si_cons.wait_and_advance()
+
+            tTMEM_LOADrS = cute.make_rmem_tensor(tTMEM_LOADcS.shape, self.qk_acc_dtype)
+            cute.copy(tiled_tmem_load, tTMEM_LOADtS, tTMEM_LOADrS)
+            tTMEM_STORErS_x4 = cute.make_rmem_tensor(tTMEM_STOREcS.shape, self.qk_acc_dtype)
+            tTMEM_STORErS_x4_e = cute.make_tensor(
+                cute.recast_ptr(tTMEM_STORErS_x4.iterator, dtype=self.q_dtype),
+                tTMEM_LOADrS.layout)
+            frg_cnt = 4
+            frg_tile = cute.size(tTMEM_LOADrS) // frg_cnt
+            tTMEM_LOADrS_frg = cute.logical_divide(tTMEM_LOADrS, cute.make_layout(frg_tile))
+            tTMEM_STORErS_x4_e_frg = cute.logical_divide(
+                tTMEM_STORErS_x4_e, cute.make_layout(frg_tile))
+            for j in range(frg_cnt):
+                s_vec = tTMEM_LOADrS_frg[None, j].load()
+                tTMEM_STORErS_x4_e_frg[None, j].store(s_vec.to(self.q_dtype))
+            cute.copy(tiled_tmem_store, tTMEM_STORErS_x4, tTMEM_STOREtS_x4)
+            cute.arch.fence_view_async_tmem_store()
+            si_handle.release()
+
+            # Output epilogue
+            tCtO_base = cute.make_tensor(tmem_ptr + self.tmem_o0_offset, tCtO_fake.layout)
+            acc_cons_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Consumer, self.num_acc_stage)
+            c_grp = pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id))
+            c_pipe = pipeline.PipelineTmaStore.create(num_stages=self.num_c_stage, producer_group=c_grp)
+            acc_cons_st = utils.gemm.sm100.epilogue_tma_store(
+                self, tidx, warp_idx, tma_c, tCtO_base, sC, tCgC,
+                epi_tile, 0, const_expr(lambda x: x), (0,0,0), acc_cons_st, acc_pipe, c_pipe)
+            c_pipe.producer_tail()
+            tmem.relinquish_alloc_permit()
+            tmem.free(tmem_ptr)
+
+
+def test():
+    torch.manual_seed(42)
+    m, n, head_dim = 128, 128, 64
+    q = torch.randn(m, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    k = torch.randn(n, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    # V = I[:32,:] MN-major: (32,128) with strides (1,32)
+    v_data = torch.eye(128, dtype=torch.bfloat16, device='cuda')
+    v = v_data[:32, :].as_strided((32, 128), (1, 32)).unsqueeze(-1)
+    c = torch.zeros(m, 32, 1, dtype=torch.bfloat16, device='cuda')
+
+    qf = q[:,:,0].float(); kf = k[:,:,0].float()
+    # V=I[:32,:] MN-major -> O = P[:,:32] = (Q@K^T).bf16()[:,:32]
+    ref = (qf @ kf.T).bfloat16().float()[:, :32]
+
+    mQ = ct.from_dlpack(q).mark_layout_dynamic(leading_dim=ct.get_leading_dim(q))
+    mK = ct.from_dlpack(k).mark_layout_dynamic(leading_dim=ct.get_leading_dim(k))
+    mV = ct.from_dlpack(v).mark_layout_dynamic(leading_dim=ct.get_leading_dim(v))
+    mC = ct.from_dlpack(c).mark_layout_dynamic(leading_dim=ct.get_leading_dim(c))
+    stream = cuda.CUstream(torch.cuda.current_stream().cuda_stream)
+    kernel = Native32Kernel(mma_tiler_mn=(128, 128))
+    print('Compiling...', flush=True)
+    compiled = cute.compile(kernel, mQ, mK, mV, mC, stream)
+    print('Running...', flush=True)
+    compiled(mQ, mK, mV, mC, stream)
+    torch.cuda.synchronize()
+    out = c[:,:,0].float()
+    cos = torch.nn.functional.cosine_similarity(out.flatten().unsqueeze(0), ref.flatten().unsqueeze(0)).item()
+    print('PV(128,32) ctafix2: cosine {:.6f} {}'.format(cos, 'PASS' if cos >= 0.99 else 'FAIL'))
+
+
+if __name__ == '__main__':
+    test()

tests/test_128_32_native.py

@@ -0,0 +1,382 @@
+"""
+Minimal PV-only test: Load P from GMEM to TMEM via QK-style MMA, then PV from TMEM.
+Step 1: QK MMA writes FP32 S to TMEM (we know this works)
+Step 2: Softmax packing writes BF16 P to TMEM (test this)
+Step 3: PV MMA reads BF16 P from TMEM and V from SMEM, produces O
+
+But to isolate the bug, let me test just the PV MMA in isolation.
+I'll write known BF16 values to TMEM using the softmax packing path,
+then immediately read them back using the PV A-fragment path,
+and compare.
+
+Actually, the simplest isolation test:
+1. Do QK MMA to get S in TMEM (cosine 0.999999 verified)
+2. Do softmax packing: S → P in TMEM (at offset 32)
+3. Skip PV entirely — read P from TMEM using the C-fragment composition LOAD path
+4. Output P to GMEM and compare against S.to(BF16)
+
+This tests whether the softmax packing writes P correctly to the same TMEM
+that the PV would read from.
+
+But we can't easily read P from TMEM using the standard epilogue path
+because the epilogue expects FP32 accumulator data.
+
+Alternative: Use the PV MMA with V=I (identity). If P is correct,
+then P @ I = P. But V needs to be MN-major and (128, 128), not (128, 64).
+The output would be (128, 128) which doesn't match our (128, 64) c tensor.
+
+Let me use V that selects the first 64 columns: V[k, n] = delta(k, n) for k in [0,63].
+This gives P @ V = P[:, :64], and the output is (128, 64).
+But V is (128, 128) in the MMA K,N dims. V[k, n] for k in [0,127], n in [0,63].
+Hmm, this is getting complicated. Let me just do the identity approach with a (128, 128) output.
+"""
+import torch, cutlass, cutlass.cute as cute, cutlass.utils as utils, cutlass.pipeline as pipeline
+from cutlass.cute.nvgpu import cpasync, tcgen05
+from cutlass import Float32, BFloat16, Int32, Boolean, const_expr
+from cutlass.utils import LayoutEnum
+from cutlass.utils.tmem_allocator import find_tmem_tensor_col_offset
+import cuda.bindings.driver as cuda
+import cutlass.torch as ct
+
+
+class Native32Kernel:
+    """QK + softmax packing + PV with V=I to isolate PV MMA correctness.
+    Output should be P = S.to(BF16), i.e. (Q@K^T).bfloat16()
+    With V=I, O = P @ I = P.
+    But V is (K=128, N=128) in the MMA. We need a 128x128 identity in MN-major.
+    Output tensor is (128, 128).
+    """
+    def __init__(self, mma_tiler_mn):
+        self.acc_dtype = Float32; self.qk_acc_dtype = Float32
+        self.q_dtype = BFloat16; self.o_dtype = BFloat16; self.c_dtype = BFloat16
+        self.mma_tiler_mn = mma_tiler_mn; self.mma_tiler = (*mma_tiler_mn, 1)
+        self.use_2cta_instrs = False  # needed by epilogue_tma_store
+        self.epilog_sync_bar_id = 1  # needed by epilogue_tma_store
+        self.cluster_shape_mn = (1, 1)
+        self.cta_group = tcgen05.CtaGroup.ONE
+        self.epilogue_warp_id = (0, 1, 2, 3)
+        self.mma_warp_id = 4; self.tma_warp_id = 5
+        self.threads_per_cta = 192
+        self.num_c_stage = 2
+
+    def _setup(self, qk_mma, pv_mma):
+        qk_inst_k = int(cute.size(qk_mma.shape_mnk, mode=[2]))
+        self.qk_mma_tiler = (*self.mma_tiler_mn, qk_inst_k * 4)
+        # PV with V=I: output is (128, 128), same as QK
+        self.pv_mma_tiler = (self.qk_mma_tiler[0], qk_inst_k, self.qk_mma_tiler[1])
+        # pv_mma_tiler = (128, 128, 128) since V is 128x128
+        self.mma_tiler = self.qk_mma_tiler
+
+        self.cluster_layout_vmnk = cute.tiled_divide(cute.make_layout((1,1,1)), (qk_mma.thr_id.shape,))
+        self.cta_tile_shape_mnk = (
+            self.qk_mma_tiler[0] // cute.size(qk_mma.thr_id.shape),
+            self.qk_mma_tiler[1], self.qk_mma_tiler[2])
+        self.c_layout = LayoutEnum.ROW_MAJOR
+        self.epi_tile = utils.sm100.compute_epilogue_tile_shape(
+            (self.pv_mma_tiler[0], self.pv_mma_tiler[1], self.pv_mma_tiler[2]), False, self.c_layout, self.o_dtype)
+        self.num_ab_stage = 1; self.num_acc_stage = 1
+
+        self.a_smem_s = utils.sm100.make_smem_layout_a(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.b_smem_s = utils.sm100.make_smem_layout_b(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.v_smem_s = utils.sm100.make_smem_layout_b(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.p_tmem_s = utils.sm100.make_smem_layout_a(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.c_smem_s = utils.sm100.make_smem_layout_epi(self.o_dtype, self.c_layout, self.epi_tile, 2)
+
+        qk_thr = qk_mma.get_slice(0)
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        s_cols = find_tmem_tensor_col_offset(tStS)
+        pv_thr = pv_mma.get_slice(0)
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        o_cols = find_tmem_tensor_col_offset(tOtO)
+
+        self.tilePlikeFP32 = self.qk_mma_tiler[1] // Float32.width * self.o_dtype.width
+        self.tmem_s0_offset = 0
+        self.tmem_p0_offset = 32
+        self.tmem_o0_offset = s_cols
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+        self.num_tmem_alloc_cols = utils.get_num_tmem_alloc_cols([tCtS_fake, tCtO_fake], arch="sm_100")
+
+        # ⛔⛔⛔ CRITICAL: num_tma_load_bytes MUST include ALL TMA-loaded tensors (Q + K + V). Missing V → DEADLOCK. See FOOTGUN #0 in README.
+        a_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        b_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        self.num_tma_load_bytes = (
+            cute.size_in_bytes(self.q_dtype, a_smem) + cute.size_in_bytes(self.q_dtype, b_smem) +
+            cute.size_in_bytes(self.q_dtype, v_smem)
+        ) * cute.size(qk_mma.thr_id.shape)
+
+    @cute.jit
+    def __call__(self, q, k, v, c, stream):
+        self.q_dtype = q.element_type; self.o_dtype = c.element_type; self.c_dtype = self.o_dtype
+        self.a_major = LayoutEnum.from_tensor(q).mma_major_mode()
+        self.b_major = LayoutEnum.from_tensor(k).mma_major_mode()
+        self.v_major = LayoutEnum.from_tensor(v).mma_major_mode()
+        self.c_layout = LayoutEnum.from_tensor(c)
+
+        qk_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, self.a_major, self.b_major,
+            self.qk_acc_dtype, self.cta_group, self.mma_tiler_mn, tcgen05.OperandSource.SMEM)
+        # PV with 128x128 output (V=I)
+        pv_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, cute.nvgpu.OperandMajorMode.K, self.v_major,
+            self.qk_acc_dtype, self.cta_group, (128, 32), tcgen05.OperandSource.TMEM)
+        self._setup(qk_mma, pv_mma)
+
+        q_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        k_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        tma_q, tma_tq = cute.nvgpu.make_tiled_tma_atom_A(
+            utils.sm100.cluster_shape_to_tma_atom_A(self.cluster_shape_mn, qk_mma.thr_id),
+            q, q_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_k, tma_tk = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, qk_mma.thr_id),
+            k, k_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_v, tma_tv = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, pv_mma.thr_id),
+            v, v_smem, self.pv_mma_tiler, pv_mma, self.cluster_layout_vmnk.shape)
+        epi_smem = cute.select(self.c_smem_s, mode=[0, 1])
+        tma_c, tma_tc = cpasync.make_tiled_tma_atom(cpasync.CopyBulkTensorTileS2GOp(), c, epi_smem, self.epi_tile)
+
+        self._kernel(qk_mma, pv_mma, tma_q, tma_tq, tma_k, tma_tk, tma_v, tma_tv,
+            tma_c, tma_tc, self.cluster_layout_vmnk,
+            self.a_smem_s, self.b_smem_s, self.v_smem_s, self.p_tmem_s, self.c_smem_s, self.epi_tile
+        ).launch(grid=(1,1,1), block=[self.threads_per_cta,1,1], stream=stream)
+
+    @cute.kernel
+    def _kernel(self, qk_mma, pv_mma, tma_q, mQ, tma_k, mK, tma_v, mV,
+                tma_c, mC, cl_vmnk, a_smem_s, b_smem_s, v_smem_s, p_tmem_s, c_smem_s, epi_tile):
+        warp_idx = cute.arch.make_warp_uniform(cute.arch.warp_idx())
+        tidx, _, _ = cute.arch.thread_idx()
+        use_2cta = cute.size(qk_mma.thr_id.shape) == 2
+
+        if warp_idx == self.tma_warp_id:
+            cpasync.prefetch_descriptor(tma_q); cpasync.prefetch_descriptor(tma_k)
+            cpasync.prefetch_descriptor(tma_v); cpasync.prefetch_descriptor(tma_c)
+
+        @cute.struct
+        class SS:
+            ab_bar: cute.struct.MemRange[cutlass.Int64, self.num_ab_stage * 2]
+            mma_si_bar: cute.struct.MemRange[cutlass.Int64, 2]
+            acc_bar: cute.struct.MemRange[cutlass.Int64, self.num_acc_stage * 2]
+            tmem_dealloc: cutlass.Int64
+            holding: cutlass.Int32
+
+        smem = utils.SmemAllocator(); st = smem.allocate(SS)
+
+        ab_p, ab_c = pipeline.PipelineTmaUmma.create(
+            barrier_storage=st.ab_bar.data_ptr(), num_stages=self.num_ab_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 1),
+            tx_count=self.num_tma_load_bytes, cta_layout_vmnk=cl_vmnk, defer_sync=True
+        ).make_participants()
+
+        mma_si_prod, mma_si_cons = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.mma_si_bar.data_ptr(), num_stages=1,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id)),
+        ).make_participants()
+
+        acc_pipe = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.acc_bar.data_ptr(), num_stages=self.num_acc_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(
+                pipeline.Agent.Thread, len(self.epilogue_warp_id) * (2 if use_2cta else 1)),
+            cta_layout_vmnk=cl_vmnk, defer_sync=True)
+
+        tmem_bar = pipeline.NamedBarrier(barrier_id=2,
+            num_threads=32 * len((self.mma_warp_id, *self.epilogue_warp_id)))
+        tmem = utils.TmemAllocator(st.holding.ptr, barrier_for_retrieve=tmem_bar,
+            allocator_warp_id=self.epilogue_warp_id[0], is_two_cta=use_2cta,
+            two_cta_tmem_dealloc_mbar_ptr=st.tmem_dealloc.ptr)
+
+        pipeline.pipeline_init_arrive(cluster_shape_mn=cl_vmnk, is_relaxed=True)
+
+        sQ = smem.allocate_tensor(element_type=self.q_dtype, layout=a_smem_s.outer, byte_alignment=128, swizzle=a_smem_s.inner)
+        sK = smem.allocate_tensor(element_type=self.q_dtype, layout=b_smem_s.outer, byte_alignment=128, swizzle=b_smem_s.inner)
+        sV = smem.allocate_tensor(element_type=self.q_dtype, layout=v_smem_s.outer, byte_alignment=128, swizzle=v_smem_s.inner)
+        sC = smem.allocate_tensor(element_type=self.o_dtype, layout=c_smem_s.outer, byte_alignment=128, swizzle=c_smem_s.inner)
+
+        gQ = cute.local_tile(mQ, cute.slice_(self.qk_mma_tiler, (None,0,None)), (None,None,None))
+        gK = cute.local_tile(mK, cute.slice_(self.qk_mma_tiler, (0,None,None)), (None,None,None))
+        gC = cute.local_tile(mC, cute.slice_(self.qk_mma_tiler, (None,None,0)), (None,None,None))
+        k_cnt = cute.size(gQ, mode=[3])
+
+        qk_thr = qk_mma.get_slice(0)
+        pv_thr = pv_mma.get_slice(0)
+        tCgQ = qk_thr.partition_A(gQ); tCgK = qk_thr.partition_B(gK); tCgC = qk_thr.partition_C(gC)
+        a_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,0,None,0)).shape)
+        tAsQ, tAgQ = cpasync.tma_partition(tma_q, 0, a_lay, cute.group_modes(sQ,0,3), cute.group_modes(tCgQ,0,3))
+        b_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,None,0,0)).shape)
+        tBsK, tBgK = cpasync.tma_partition(tma_k, 0, b_lay, cute.group_modes(sK,0,3), cute.group_modes(tCgK,0,3))
+        tAgQ = tAgQ[(None,0,None,0)]; tBgK = tBgK[(None,0,None,0)]
+
+        gV = cute.local_tile(mV, cute.slice_(self.pv_mma_tiler, (0,None,None)), (None,None,None))
+        tCgV = pv_thr.partition_B(gV)
+        tVsV, tVgV = cpasync.tma_partition(tma_v, 0, b_lay, cute.group_modes(sV,0,3), cute.group_modes(tCgV,0,3))
+        tVgV = tVgV[(None,0,None,0)]
+
+        tCrQ = qk_mma.make_fragment_A(sQ); tCrK = qk_mma.make_fragment_B(sK)
+        tCrV = pv_mma.make_fragment_B(sV)
+
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        tStS0 = cute.make_tensor(tStS.iterator + self.tmem_s0_offset, tStS.layout)
+
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        tOtO0 = cute.make_tensor(tOtO.iterator + self.tmem_o0_offset, tOtO.layout)
+
+        tP = cute.make_tensor(tStS.iterator, p_tmem_s.outer)
+        tOrP_base = pv_thr.make_fragment_A(tP)
+        tOrP = tOrP_base[(None, None, None, 0)]
+        tOrP0 = cute.make_tensor(
+            tOrP.iterator + self.qk_acc_dtype.width // self.q_dtype.width * self.tmem_p0_offset,
+            tOrP.layout)
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+
+        pipeline.pipeline_init_wait(cluster_shape_mn=cl_vmnk)
+
+        # ═══ TMA LOAD WARP ═══
+        if warp_idx == self.tma_warp_id:
+            ab_p.reset(); peek = ab_p.try_acquire()
+            for kt in cutlass.range(k_cnt, unroll=1):
+                h = ab_p.acquire_and_advance(peek)
+                cute.copy(tma_q, tAgQ[(None,h.count)], tAsQ[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_k, tBgK[(None,h.count)], tBsK[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_v, tVgV[(None,h.count)], tVsV[(None,h.index)], tma_bar_ptr=h.barrier)
+                peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_p.try_acquire()
+            ab_p.tail()
+
+        # ═══ MMA WARP ═══
+        if warp_idx == self.mma_warp_id:
+            tmem.wait_for_alloc()
+            ab_c.reset(); peek = ab_c.try_wait()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            acc_prod_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Producer, self.num_acc_stage)
+            acc_pipe.producer_acquire(acc_prod_st)
+            qk_mma.set(tcgen05.Field.ACCUMULATE, False)
+            for kt in range(k_cnt):
+                h = ab_c.wait_and_advance(peek)
+                nblk = cute.size(tCrQ, mode=[2])
+                for kb in cutlass.range(nblk, unroll_full=True):
+                    cute.gemm(qk_mma, tStS0, tCrQ[(None,None,kb,h.index)], tCrK[(None,None,kb,h.index)], tStS0)
+                    qk_mma.set(tcgen05.Field.ACCUMULATE, True)
+                h.release(); peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_c.try_wait()
+
+            cute.arch.fence_view_async_tmem_store()
+            s0_handle.commit()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            # PV MMA: P @ V where V=I → O = P
+            pv_mma.set(tcgen05.Field.ACCUMULATE, False)
+            tCrV_s = tCrV[(None, None, None, 0)]
+            nblk_pv = cute.size(tOrP0, mode=[2])
+            for kb in cutlass.range(nblk_pv, unroll_full=True):
+                cute.gemm(pv_mma, tOtO0, tOrP0[(None,None,kb)], tCrV_s[(None,None,kb)], tOtO0)
+                pv_mma.set(tcgen05.Field.ACCUMULATE, True)
+
+            acc_pipe.producer_commit(acc_prod_st)
+            acc_prod_st.advance()
+            acc_pipe.producer_tail(acc_prod_st)
+
+        # ═══ EPILOGUE WARPS ═══
+        if warp_idx < self.mma_warp_id:
+            tmem.allocate(self.num_tmem_alloc_cols)
+            tmem.wait_for_alloc()
+            tmem_ptr = tmem.retrieve_ptr(self.qk_acc_dtype)
+            sfw_idx = tidx % (32 * len(self.epilogue_warp_id))
+
+            tmem_load_atom = cute.make_copy_atom(
+                tcgen05.copy.Ld32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_load = tcgen05.make_tmem_copy(tmem_load_atom, tStS0)
+            thr_load = tiled_tmem_load.get_slice(sfw_idx)
+            tTMEM_LOADtS = thr_load.partition_S(tStS0)
+            cS = cute.make_identity_tensor((self.qk_mma_tiler[0], self.qk_mma_tiler[1]))
+            tScS = qk_thr.partition_C(cS)
+            tTMEM_LOADcS = thr_load.partition_D(tScS)
+
+            tStS_P_layout = cute.composition(tStS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tStS_P = cute.make_tensor(tStS.iterator + self.tmem_p0_offset, tStS_P_layout)
+            tmem_store_atom = cute.make_copy_atom(
+                tcgen05.copy.St32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_store = tcgen05.make_tmem_copy(tmem_store_atom, tStS_P)
+            thr_store = tiled_tmem_store.get_slice(sfw_idx)
+            tTMEM_STOREtS_x4 = thr_store.partition_D(tStS_P)
+            tScS_P_layout = cute.composition(tScS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tScS_P = cute.make_tensor(tScS.iterator, tScS_P_layout)
+            tTMEM_STOREcS = thr_store.partition_S(tScS_P)
+
+            si_handle = mma_si_cons.wait_and_advance()
+
+            tTMEM_LOADrS = cute.make_rmem_tensor(tTMEM_LOADcS.shape, self.qk_acc_dtype)
+            cute.copy(tiled_tmem_load, tTMEM_LOADtS, tTMEM_LOADrS)
+            tTMEM_STORErS_x4 = cute.make_rmem_tensor(tTMEM_STOREcS.shape, self.qk_acc_dtype)
+            tTMEM_STORErS_x4_e = cute.make_tensor(
+                cute.recast_ptr(tTMEM_STORErS_x4.iterator, dtype=self.q_dtype),
+                tTMEM_LOADrS.layout)
+            frg_cnt = 4
+            frg_tile = cute.size(tTMEM_LOADrS) // frg_cnt
+            tTMEM_LOADrS_frg = cute.logical_divide(tTMEM_LOADrS, cute.make_layout(frg_tile))
+            tTMEM_STORErS_x4_e_frg = cute.logical_divide(
+                tTMEM_STORErS_x4_e, cute.make_layout(frg_tile))
+            for j in range(frg_cnt):
+                s_vec = tTMEM_LOADrS_frg[None, j].load()
+                tTMEM_STORErS_x4_e_frg[None, j].store(s_vec.to(self.q_dtype))
+            cute.copy(tiled_tmem_store, tTMEM_STORErS_x4, tTMEM_STOREtS_x4)
+            cute.arch.fence_view_async_tmem_store()
+            si_handle.release()
+
+            # Output epilogue
+            tCtO_base = cute.make_tensor(tmem_ptr + self.tmem_o0_offset, tCtO_fake.layout)
+            acc_cons_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Consumer, self.num_acc_stage)
+            c_grp = pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id))
+            c_pipe = pipeline.PipelineTmaStore.create(num_stages=self.num_c_stage, producer_group=c_grp)
+            acc_cons_st = utils.gemm.sm100.epilogue_tma_store(
+                self, tidx, warp_idx, tma_c, tCtO_base, sC, tCgC,
+                epi_tile, 0, const_expr(lambda x: x), (0,0,0), acc_cons_st, acc_pipe, c_pipe)
+            c_pipe.producer_tail()
+            tmem.relinquish_alloc_permit()
+            tmem.free(tmem_ptr)
+
+
+def test():
+    torch.manual_seed(42)
+    m, n, head_dim = 128, 128, 64
+    q = torch.randn(m, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    k = torch.randn(n, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    # V = I[:32,:] MN-major: (32,128) with strides (1,32)
+    v_data = torch.eye(128, dtype=torch.bfloat16, device='cuda')
+    v = v_data[:32, :].as_strided((32, 128), (1, 32)).unsqueeze(-1)
+    c = torch.zeros(m, 32, 1, dtype=torch.bfloat16, device='cuda')
+
+    qf = q[:,:,0].float(); kf = k[:,:,0].float()
+    # V=I[:32,:] MN-major -> O = P[:,:32] = (Q@K^T).bf16()[:,:32]
+    ref = (qf @ kf.T).bfloat16().float()[:, :32]
+
+    mQ = ct.from_dlpack(q).mark_layout_dynamic(leading_dim=ct.get_leading_dim(q))
+    mK = ct.from_dlpack(k).mark_layout_dynamic(leading_dim=ct.get_leading_dim(k))
+    mV = ct.from_dlpack(v).mark_layout_dynamic(leading_dim=ct.get_leading_dim(v))
+    mC = ct.from_dlpack(c).mark_layout_dynamic(leading_dim=ct.get_leading_dim(c))
+    stream = cuda.CUstream(torch.cuda.current_stream().cuda_stream)
+    kernel = Native32Kernel(mma_tiler_mn=(128, 128))
+    print('Compiling...', flush=True)
+    compiled = cute.compile(kernel, mQ, mK, mV, mC, stream)
+    print('Running...', flush=True)
+    compiled(mQ, mK, mV, mC, stream)
+    torch.cuda.synchronize()
+    out = c[:,:,0].float()
+    cos = torch.nn.functional.cosine_similarity(out.flatten().unsqueeze(0), ref.flatten().unsqueeze(0)).item()
+    print('PV(128,32) native: cosine {:.6f} {}'.format(cos, 'PASS' if cos >= 0.99 else 'FAIL'))
+
+
+if __name__ == '__main__':
+    test()

tests/test_128_32_zeropad.py

@@ -0,0 +1,385 @@
+"""
+Minimal PV-only test: Load P from GMEM to TMEM via QK-style MMA, then PV from TMEM.
+Step 1: QK MMA writes FP32 S to TMEM (we know this works)
+Step 2: Softmax packing writes BF16 P to TMEM (test this)
+Step 3: PV MMA reads BF16 P from TMEM and V from SMEM, produces O
+
+But to isolate the bug, let me test just the PV MMA in isolation.
+I'll write known BF16 values to TMEM using the softmax packing path,
+then immediately read them back using the PV A-fragment path,
+and compare.
+
+Actually, the simplest isolation test:
+1. Do QK MMA to get S in TMEM (cosine 0.999999 verified)
+2. Do softmax packing: S → P in TMEM (at offset 32)
+3. Skip PV entirely — read P from TMEM using the C-fragment composition LOAD path
+4. Output P to GMEM and compare against S.to(BF16)
+
+This tests whether the softmax packing writes P correctly to the same TMEM
+that the PV would read from.
+
+But we can't easily read P from TMEM using the standard epilogue path
+because the epilogue expects FP32 accumulator data.
+
+Alternative: Use the PV MMA with V=I (identity). If P is correct,
+then P @ I = P. But V needs to be MN-major and (128, 128), not (128, 64).
+The output would be (128, 128) which doesn't match our (128, 64) c tensor.
+
+Let me use V that selects the first 64 columns: V[k, n] = delta(k, n) for k in [0,63].
+This gives P @ V = P[:, :64], and the output is (128, 64).
+But V is (128, 128) in the MMA K,N dims. V[k, n] for k in [0,127], n in [0,63].
+Hmm, this is getting complicated. Let me just do the identity approach with a (128, 128) output.
+"""
+import torch, cutlass, cutlass.cute as cute, cutlass.utils as utils, cutlass.pipeline as pipeline
+from cutlass.cute.nvgpu import cpasync, tcgen05
+from cutlass import Float32, BFloat16, Int32, Boolean, const_expr
+from cutlass.utils import LayoutEnum
+from cutlass.utils.tmem_allocator import find_tmem_tensor_col_offset
+import cuda.bindings.driver as cuda
+import cutlass.torch as ct
+
+
+class ZeroPad32Kernel:
+    """QK + softmax packing + PV with V=I to isolate PV MMA correctness.
+    Output should be P = S.to(BF16), i.e. (Q@K^T).bfloat16()
+    With V=I, O = P @ I = P.
+    But V is (K=128, N=128) in the MMA. We need a 128x128 identity in MN-major.
+    Output tensor is (128, 128).
+    """
+    def __init__(self, mma_tiler_mn):
+        self.acc_dtype = Float32; self.qk_acc_dtype = Float32
+        self.q_dtype = BFloat16; self.o_dtype = BFloat16; self.c_dtype = BFloat16
+        self.mma_tiler_mn = mma_tiler_mn; self.mma_tiler = (*mma_tiler_mn, 1)
+        self.use_2cta_instrs = False  # needed by epilogue_tma_store
+        self.epilog_sync_bar_id = 1  # needed by epilogue_tma_store
+        self.cluster_shape_mn = (1, 1)
+        self.cta_group = tcgen05.CtaGroup.ONE
+        self.epilogue_warp_id = (0, 1, 2, 3)
+        self.mma_warp_id = 4; self.tma_warp_id = 5
+        self.threads_per_cta = 192
+        self.num_c_stage = 2
+
+    def _setup(self, qk_mma, pv_mma):
+        qk_inst_k = cute.size(qk_mma.shape_mnk, mode=[2])
+        self.qk_mma_tiler = (*self.mma_tiler_mn, qk_inst_k * 4)
+        # PV with V=I: output is (128, 128), same as QK
+        self.pv_mma_tiler = (self.qk_mma_tiler[0], self.qk_mma_tiler[1], self.qk_mma_tiler[1])
+        # pv_mma_tiler = (128, 128, 128) since V is 128x128
+        self.mma_tiler = self.qk_mma_tiler
+
+        self.cluster_layout_vmnk = cute.tiled_divide(cute.make_layout((1,1,1)), (qk_mma.thr_id.shape,))
+        self.cta_tile_shape_mnk = (
+            self.qk_mma_tiler[0] // cute.size(qk_mma.thr_id.shape),
+            self.qk_mma_tiler[1], self.qk_mma_tiler[2])
+        self.c_layout = LayoutEnum.ROW_MAJOR
+        self.epi_tile = utils.sm100.compute_epilogue_tile_shape(
+            self.cta_tile_shape_mnk, False, self.c_layout, self.o_dtype)
+        self.num_ab_stage = 1; self.num_acc_stage = 1
+
+        self.a_smem_s = utils.sm100.make_smem_layout_a(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.b_smem_s = utils.sm100.make_smem_layout_b(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.v_smem_s = utils.sm100.make_smem_layout_b(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.p_tmem_s = utils.sm100.make_smem_layout_a(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.c_smem_s = utils.sm100.make_smem_layout_epi(self.o_dtype, self.c_layout, self.epi_tile, 2)
+
+        qk_thr = qk_mma.get_slice(0)
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        s_cols = find_tmem_tensor_col_offset(tStS)
+        pv_thr = pv_mma.get_slice(0)
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        o_cols = find_tmem_tensor_col_offset(tOtO)
+
+        self.tilePlikeFP32 = self.qk_mma_tiler[1] // Float32.width * self.o_dtype.width
+        self.tmem_s0_offset = 0
+        self.tmem_p0_offset = 32
+        self.tmem_o0_offset = s_cols
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+        self.num_tmem_alloc_cols = utils.get_num_tmem_alloc_cols([tCtS_fake, tCtO_fake], arch="sm_100")
+
+        # ⛔⛔⛔ CRITICAL: num_tma_load_bytes MUST include ALL TMA-loaded tensors (Q + K + V). Missing V → DEADLOCK. See FOOTGUN #0 in README.
+        a_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        b_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        self.num_tma_load_bytes = (
+            cute.size_in_bytes(self.q_dtype, a_smem) + cute.size_in_bytes(self.q_dtype, b_smem) +
+            cute.size_in_bytes(self.q_dtype, v_smem)
+        ) * cute.size(qk_mma.thr_id.shape)
+
+    @cute.jit
+    def __call__(self, q, k, v, c, stream):
+        self.q_dtype = q.element_type; self.o_dtype = c.element_type; self.c_dtype = self.o_dtype
+        self.a_major = LayoutEnum.from_tensor(q).mma_major_mode()
+        self.b_major = LayoutEnum.from_tensor(k).mma_major_mode()
+        self.v_major = LayoutEnum.from_tensor(v).mma_major_mode()
+        self.c_layout = LayoutEnum.from_tensor(c)
+
+        qk_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, self.a_major, self.b_major,
+            self.qk_acc_dtype, self.cta_group, self.mma_tiler_mn, tcgen05.OperandSource.SMEM)
+        # PV with 128x128 output (V=I)
+        pv_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, cute.nvgpu.OperandMajorMode.K, self.v_major,
+            self.qk_acc_dtype, self.cta_group, self.mma_tiler_mn, tcgen05.OperandSource.TMEM)
+        self._setup(qk_mma, pv_mma)
+
+        q_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        k_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        tma_q, tma_tq = cute.nvgpu.make_tiled_tma_atom_A(
+            utils.sm100.cluster_shape_to_tma_atom_A(self.cluster_shape_mn, qk_mma.thr_id),
+            q, q_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_k, tma_tk = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, qk_mma.thr_id),
+            k, k_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_v, tma_tv = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, pv_mma.thr_id),
+            v, v_smem, self.pv_mma_tiler, pv_mma, self.cluster_layout_vmnk.shape)
+        epi_smem = cute.select(self.c_smem_s, mode=[0, 1])
+        tma_c, tma_tc = cpasync.make_tiled_tma_atom(cpasync.CopyBulkTensorTileS2GOp(), c, epi_smem, self.epi_tile)
+
+        self._kernel(qk_mma, pv_mma, tma_q, tma_tq, tma_k, tma_tk, tma_v, tma_tv,
+            tma_c, tma_tc, self.cluster_layout_vmnk,
+            self.a_smem_s, self.b_smem_s, self.v_smem_s, self.p_tmem_s, self.c_smem_s, self.epi_tile
+        ).launch(grid=(1,1,1), block=[self.threads_per_cta,1,1], stream=stream)
+
+    @cute.kernel
+    def _kernel(self, qk_mma, pv_mma, tma_q, mQ, tma_k, mK, tma_v, mV,
+                tma_c, mC, cl_vmnk, a_smem_s, b_smem_s, v_smem_s, p_tmem_s, c_smem_s, epi_tile):
+        warp_idx = cute.arch.make_warp_uniform(cute.arch.warp_idx())
+        tidx, _, _ = cute.arch.thread_idx()
+        use_2cta = cute.size(qk_mma.thr_id.shape) == 2
+
+        if warp_idx == self.tma_warp_id:
+            cpasync.prefetch_descriptor(tma_q); cpasync.prefetch_descriptor(tma_k)
+            cpasync.prefetch_descriptor(tma_v); cpasync.prefetch_descriptor(tma_c)
+
+        @cute.struct
+        class SS:
+            ab_bar: cute.struct.MemRange[cutlass.Int64, self.num_ab_stage * 2]
+            mma_si_bar: cute.struct.MemRange[cutlass.Int64, 2]
+            acc_bar: cute.struct.MemRange[cutlass.Int64, self.num_acc_stage * 2]
+            tmem_dealloc: cutlass.Int64
+            holding: cutlass.Int32
+
+        smem = utils.SmemAllocator(); st = smem.allocate(SS)
+
+        ab_p, ab_c = pipeline.PipelineTmaUmma.create(
+            barrier_storage=st.ab_bar.data_ptr(), num_stages=self.num_ab_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 1),
+            tx_count=self.num_tma_load_bytes, cta_layout_vmnk=cl_vmnk, defer_sync=True
+        ).make_participants()
+
+        mma_si_prod, mma_si_cons = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.mma_si_bar.data_ptr(), num_stages=1,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id)),
+        ).make_participants()
+
+        acc_pipe = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.acc_bar.data_ptr(), num_stages=self.num_acc_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(
+                pipeline.Agent.Thread, len(self.epilogue_warp_id) * (2 if use_2cta else 1)),
+            cta_layout_vmnk=cl_vmnk, defer_sync=True)
+
+        tmem_bar = pipeline.NamedBarrier(barrier_id=2,
+            num_threads=32 * len((self.mma_warp_id, *self.epilogue_warp_id)))
+        tmem = utils.TmemAllocator(st.holding.ptr, barrier_for_retrieve=tmem_bar,
+            allocator_warp_id=self.epilogue_warp_id[0], is_two_cta=use_2cta,
+            two_cta_tmem_dealloc_mbar_ptr=st.tmem_dealloc.ptr)
+
+        pipeline.pipeline_init_arrive(cluster_shape_mn=cl_vmnk, is_relaxed=True)
+
+        sQ = smem.allocate_tensor(element_type=self.q_dtype, layout=a_smem_s.outer, byte_alignment=128, swizzle=a_smem_s.inner)
+        sK = smem.allocate_tensor(element_type=self.q_dtype, layout=b_smem_s.outer, byte_alignment=128, swizzle=b_smem_s.inner)
+        sV = smem.allocate_tensor(element_type=self.q_dtype, layout=v_smem_s.outer, byte_alignment=128, swizzle=v_smem_s.inner)
+        sC = smem.allocate_tensor(element_type=self.o_dtype, layout=c_smem_s.outer, byte_alignment=128, swizzle=c_smem_s.inner)
+
+        gQ = cute.local_tile(mQ, cute.slice_(self.qk_mma_tiler, (None,0,None)), (None,None,None))
+        gK = cute.local_tile(mK, cute.slice_(self.qk_mma_tiler, (0,None,None)), (None,None,None))
+        gC = cute.local_tile(mC, cute.slice_(self.qk_mma_tiler, (None,None,0)), (None,None,None))
+        k_cnt = cute.size(gQ, mode=[3])
+
+        qk_thr = qk_mma.get_slice(0)
+        pv_thr = pv_mma.get_slice(0)
+        tCgQ = qk_thr.partition_A(gQ); tCgK = qk_thr.partition_B(gK); tCgC = qk_thr.partition_C(gC)
+        a_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,0,None,0)).shape)
+        tAsQ, tAgQ = cpasync.tma_partition(tma_q, 0, a_lay, cute.group_modes(sQ,0,3), cute.group_modes(tCgQ,0,3))
+        b_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,None,0,0)).shape)
+        tBsK, tBgK = cpasync.tma_partition(tma_k, 0, b_lay, cute.group_modes(sK,0,3), cute.group_modes(tCgK,0,3))
+        tAgQ = tAgQ[(None,0,None,0)]; tBgK = tBgK[(None,0,None,0)]
+
+        gV = cute.local_tile(mV, cute.slice_(self.pv_mma_tiler, (0,None,None)), (None,None,None))
+        tCgV = pv_thr.partition_B(gV)
+        tVsV, tVgV = cpasync.tma_partition(tma_v, 0, b_lay, cute.group_modes(sV,0,3), cute.group_modes(tCgV,0,3))
+        tVgV = tVgV[(None,0,None,0)]
+
+        tCrQ = qk_mma.make_fragment_A(sQ); tCrK = qk_mma.make_fragment_B(sK)
+        tCrV = pv_mma.make_fragment_B(sV)
+
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        tStS0 = cute.make_tensor(tStS.iterator + self.tmem_s0_offset, tStS.layout)
+
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        tOtO0 = cute.make_tensor(tOtO.iterator + self.tmem_o0_offset, tOtO.layout)
+
+        tP = cute.make_tensor(tStS.iterator, p_tmem_s.outer)
+        tOrP_base = pv_thr.make_fragment_A(tP)
+        tOrP = tOrP_base[(None, None, None, 0)]
+        tOrP0 = cute.make_tensor(
+            tOrP.iterator + self.qk_acc_dtype.width // self.q_dtype.width * self.tmem_p0_offset,
+            tOrP.layout)
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+
+        pipeline.pipeline_init_wait(cluster_shape_mn=cl_vmnk)
+
+        # ═══ TMA LOAD WARP ═══
+        if warp_idx == self.tma_warp_id:
+            ab_p.reset(); peek = ab_p.try_acquire()
+            for kt in cutlass.range(k_cnt, unroll=1):
+                h = ab_p.acquire_and_advance(peek)
+                cute.copy(tma_q, tAgQ[(None,h.count)], tAsQ[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_k, tBgK[(None,h.count)], tBsK[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_v, tVgV[(None,h.count)], tVsV[(None,h.index)], tma_bar_ptr=h.barrier)
+                peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_p.try_acquire()
+            ab_p.tail()
+
+        # ═══ MMA WARP ═══
+        if warp_idx == self.mma_warp_id:
+            tmem.wait_for_alloc()
+            ab_c.reset(); peek = ab_c.try_wait()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            acc_prod_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Producer, self.num_acc_stage)
+            acc_pipe.producer_acquire(acc_prod_st)
+            qk_mma.set(tcgen05.Field.ACCUMULATE, False)
+            for kt in range(k_cnt):
+                h = ab_c.wait_and_advance(peek)
+                nblk = cute.size(tCrQ, mode=[2])
+                for kb in cutlass.range(nblk, unroll_full=True):
+                    cute.gemm(qk_mma, tStS0, tCrQ[(None,None,kb,h.index)], tCrK[(None,None,kb,h.index)], tStS0)
+                    qk_mma.set(tcgen05.Field.ACCUMULATE, True)
+                h.release(); peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_c.try_wait()
+
+            cute.arch.fence_view_async_tmem_store()
+            s0_handle.commit()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            # PV MMA: P @ V where V=I → O = P
+            pv_mma.set(tcgen05.Field.ACCUMULATE, False)
+            tCrV_s = tCrV[(None, None, None, 0)]
+            nblk_pv = cute.size(tOrP0, mode=[2])
+            for kb in cutlass.range(nblk_pv, unroll_full=True):
+                cute.gemm(pv_mma, tOtO0, tOrP0[(None,None,kb)], tCrV_s[(None,None,kb)], tOtO0)
+                pv_mma.set(tcgen05.Field.ACCUMULATE, True)
+
+            acc_pipe.producer_commit(acc_prod_st)
+            acc_prod_st.advance()
+            acc_pipe.producer_tail(acc_prod_st)
+
+        # ═══ EPILOGUE WARPS ═══
+        if warp_idx < self.mma_warp_id:
+            tmem.allocate(self.num_tmem_alloc_cols)
+            tmem.wait_for_alloc()
+            tmem_ptr = tmem.retrieve_ptr(self.qk_acc_dtype)
+            sfw_idx = tidx % (32 * len(self.epilogue_warp_id))
+
+            tmem_load_atom = cute.make_copy_atom(
+                tcgen05.copy.Ld32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_load = tcgen05.make_tmem_copy(tmem_load_atom, tStS0)
+            thr_load = tiled_tmem_load.get_slice(sfw_idx)
+            tTMEM_LOADtS = thr_load.partition_S(tStS0)
+            cS = cute.make_identity_tensor((self.qk_mma_tiler[0], self.qk_mma_tiler[1]))
+            tScS = qk_thr.partition_C(cS)
+            tTMEM_LOADcS = thr_load.partition_D(tScS)
+
+            tStS_P_layout = cute.composition(tStS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tStS_P = cute.make_tensor(tStS.iterator + self.tmem_p0_offset, tStS_P_layout)
+            tmem_store_atom = cute.make_copy_atom(
+                tcgen05.copy.St32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_store = tcgen05.make_tmem_copy(tmem_store_atom, tStS_P)
+            thr_store = tiled_tmem_store.get_slice(sfw_idx)
+            tTMEM_STOREtS_x4 = thr_store.partition_D(tStS_P)
+            tScS_P_layout = cute.composition(tScS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tScS_P = cute.make_tensor(tScS.iterator, tScS_P_layout)
+            tTMEM_STOREcS = thr_store.partition_S(tScS_P)
+
+            si_handle = mma_si_cons.wait_and_advance()
+
+            tTMEM_LOADrS = cute.make_rmem_tensor(tTMEM_LOADcS.shape, self.qk_acc_dtype)
+            cute.copy(tiled_tmem_load, tTMEM_LOADtS, tTMEM_LOADrS)
+            tTMEM_STORErS_x4 = cute.make_rmem_tensor(tTMEM_STOREcS.shape, self.qk_acc_dtype)
+            tTMEM_STORErS_x4_e = cute.make_tensor(
+                cute.recast_ptr(tTMEM_STORErS_x4.iterator, dtype=self.q_dtype),
+                tTMEM_LOADrS.layout)
+            frg_cnt = 4
+            frg_tile = cute.size(tTMEM_LOADrS) // frg_cnt
+            tTMEM_LOADrS_frg = cute.logical_divide(tTMEM_LOADrS, cute.make_layout(frg_tile))
+            tTMEM_STORErS_x4_e_frg = cute.logical_divide(
+                tTMEM_STORErS_x4_e, cute.make_layout(frg_tile))
+            for j in range(frg_cnt):
+                s_vec = tTMEM_LOADrS_frg[None, j].load()
+                tTMEM_STORErS_x4_e_frg[None, j].store(s_vec.to(self.q_dtype))
+            cute.copy(tiled_tmem_store, tTMEM_STORErS_x4, tTMEM_STOREtS_x4)
+            cute.arch.fence_view_async_tmem_store()
+            si_handle.release()
+
+            # Output epilogue
+            tCtO_base = cute.make_tensor(tmem_ptr + self.tmem_o0_offset, tCtO_fake.layout)
+            acc_cons_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Consumer, self.num_acc_stage)
+            c_grp = pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id))
+            c_pipe = pipeline.PipelineTmaStore.create(num_stages=self.num_c_stage, producer_group=c_grp)
+            acc_cons_st = utils.gemm.sm100.epilogue_tma_store(
+                self, tidx, warp_idx, tma_c, tCtO_base, sC, tCgC,
+                epi_tile, 0, const_expr(lambda x: x), (0,0,0), acc_cons_st, acc_pipe, c_pipe)
+            c_pipe.producer_tail()
+            tmem.relinquish_alloc_permit()
+            tmem.free(tmem_ptr)
+
+
+def test():
+    torch.manual_seed(42)
+    m, n, head_dim = 128, 128, 64
+    q = torch.randn(m, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    k = torch.randn(n, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    # V = zero-padded: (128,128) with first 32 rows = I[:32,:], rest zeros
+    v = torch.zeros(128, 128, dtype=torch.bfloat16, device='cuda')
+    v[:32, :128] = torch.eye(128, dtype=torch.bfloat16, device='cuda')[:32, :]
+    v = v.as_strided((128, 128), (1, 128)).unsqueeze(-1)
+    c = torch.zeros(m, n, 1, dtype=torch.bfloat16, device='cuda')
+
+    qf = q[:,:,0].float(); kf = k[:,:,0].float()
+    # V=zero-padded I[:16,:] -> O = (Q@K^T).bf16()[:,:16], rest zeros
+    ref_full = (qf @ kf.T).bfloat16().float()
+    ref = torch.zeros_like(ref_full)
+    ref[:, :32] = ref_full[:, :32]
+
+    mQ = ct.from_dlpack(q).mark_layout_dynamic(leading_dim=ct.get_leading_dim(q))
+    mK = ct.from_dlpack(k).mark_layout_dynamic(leading_dim=ct.get_leading_dim(k))
+    mV = ct.from_dlpack(v).mark_layout_dynamic(leading_dim=ct.get_leading_dim(v))
+    mC = ct.from_dlpack(c).mark_layout_dynamic(leading_dim=ct.get_leading_dim(c))
+    stream = cuda.CUstream(torch.cuda.current_stream().cuda_stream)
+    kernel = ZeroPad32Kernel(mma_tiler_mn=(128, 128))
+    print('Compiling...', flush=True)
+    compiled = cute.compile(kernel, mQ, mK, mV, mC, stream)
+    print('Running...', flush=True)
+    compiled(mQ, mK, mV, mC, stream)
+    torch.cuda.synchronize()
+    out = c[:,:,0].float()
+    cos = torch.nn.functional.cosine_similarity(out.flatten().unsqueeze(0), ref.flatten().unsqueeze(0)).item()
+    print('PV(128,32) zero-pad: cosine {:.6f} {}'.format(cos, 'PASS' if cos >= 0.99 else 'FAIL'))
+
+
+if __name__ == '__main__':
+    test()

tests/test_tmem_col2.py

@@ -0,0 +1,383 @@
+"""
+Minimal PV-only test: Load P from GMEM to TMEM via QK-style MMA, then PV from TMEM.
+Step 1: QK MMA writes FP32 S to TMEM (we know this works)
+Step 2: Softmax packing writes BF16 P to TMEM (test this)
+Step 3: PV MMA reads BF16 P from TMEM and V from SMEM, produces O
+
+But to isolate the bug, let me test just the PV MMA in isolation.
+I'll write known BF16 values to TMEM using the softmax packing path,
+then immediately read them back using the PV A-fragment path,
+and compare.
+
+Actually, the simplest isolation test:
+1. Do QK MMA to get S in TMEM (cosine 0.999999 verified)
+2. Do softmax packing: S → P in TMEM (at offset 32)
+3. Skip PV entirely — read P from TMEM using the C-fragment composition LOAD path
+4. Output P to GMEM and compare against S.to(BF16)
+
+This tests whether the softmax packing writes P correctly to the same TMEM
+that the PV would read from.
+
+But we can't easily read P from TMEM using the standard epilogue path
+because the epilogue expects FP32 accumulator data.
+
+Alternative: Use the PV MMA with V=I (identity). If P is correct,
+then P @ I = P. But V needs to be MN-major and (128, 128), not (128, 64).
+The output would be (128, 128) which doesn't match our (128, 64) c tensor.
+
+Let me use V that selects the first 64 columns: V[k, n] = delta(k, n) for k in [0,63].
+This gives P @ V = P[:, :64], and the output is (128, 64).
+But V is (128, 128) in the MMA K,N dims. V[k, n] for k in [0,127], n in [0,63].
+Hmm, this is getting complicated. Let me just do the identity approach with a (128, 128) output.
+"""
+import torch, cutlass, cutlass.cute as cute, cutlass.utils as utils, cutlass.pipeline as pipeline
+from cutlass.cute.nvgpu import cpasync, tcgen05
+from cutlass import Float32, BFloat16, Int32, Boolean, const_expr
+from cutlass.utils import LayoutEnum
+from cutlass.utils.tmem_allocator import find_tmem_tensor_col_offset
+import cuda.bindings.driver as cuda
+import cutlass.torch as ct
+
+
+class PvDiagKernel:
+    """QK + softmax packing + PV with V=I to isolate PV MMA correctness.
+    Output should be P = S.to(BF16), i.e. (Q@K^T).bfloat16()
+    With V=I, O = P @ I = P.
+    But V is (K=128, N=128) in the MMA. We need a 128x128 identity in MN-major.
+    Output tensor is (128, 128).
+    """
+    def __init__(self, mma_tiler_mn):
+        self.acc_dtype = Float32; self.qk_acc_dtype = Float32
+        self.q_dtype = BFloat16; self.o_dtype = BFloat16; self.c_dtype = BFloat16
+        self.mma_tiler_mn = mma_tiler_mn; self.mma_tiler = (*mma_tiler_mn, 1)
+        self.use_2cta_instrs = False  # needed by epilogue_tma_store
+        self.epilog_sync_bar_id = 1  # needed by epilogue_tma_store
+        self.cluster_shape_mn = (1, 1)
+        self.cta_group = tcgen05.CtaGroup.ONE
+        self.epilogue_warp_id = (0, 1, 2, 3)
+        self.mma_warp_id = 4; self.tma_warp_id = 5
+        self.threads_per_cta = 192
+        self.num_c_stage = 2
+
+    def _setup(self, qk_mma, pv_mma):
+        qk_inst_k = cute.size(qk_mma.shape_mnk, mode=[2])
+        self.qk_mma_tiler = (*self.mma_tiler_mn, qk_inst_k * 4)
+        # PV with V=I: output is (128, 128), same as QK
+        self.pv_mma_tiler = (self.qk_mma_tiler[0], self.qk_mma_tiler[1], self.qk_mma_tiler[1])
+        # pv_mma_tiler = (128, 128, 128) since V is 128x128
+        self.mma_tiler = self.qk_mma_tiler
+
+        self.cluster_layout_vmnk = cute.tiled_divide(cute.make_layout((1,1,1)), (qk_mma.thr_id.shape,))
+        self.cta_tile_shape_mnk = (
+            self.qk_mma_tiler[0] // cute.size(qk_mma.thr_id.shape),
+            self.qk_mma_tiler[1], self.qk_mma_tiler[2])
+        self.c_layout = LayoutEnum.ROW_MAJOR
+        self.epi_tile = utils.sm100.compute_epilogue_tile_shape(
+            self.cta_tile_shape_mnk, False, self.c_layout, self.o_dtype)
+        self.num_ab_stage = 1; self.num_acc_stage = 1
+
+        self.a_smem_s = utils.sm100.make_smem_layout_a(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.b_smem_s = utils.sm100.make_smem_layout_b(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.v_smem_s = utils.sm100.make_smem_layout_b(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.p_tmem_s = utils.sm100.make_smem_layout_a(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.c_smem_s = utils.sm100.make_smem_layout_epi(self.o_dtype, self.c_layout, self.epi_tile, 2)
+
+        qk_thr = qk_mma.get_slice(0)
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        s_cols = find_tmem_tensor_col_offset(tStS)
+        pv_thr = pv_mma.get_slice(0)
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        o_cols = find_tmem_tensor_col_offset(tOtO)
+
+        self.tilePlikeFP32 = self.qk_mma_tiler[1] // Float32.width * self.o_dtype.width
+        self.tmem_s0_offset = 0
+        self.tmem_p0_offset = 32
+        self.tmem_o0_offset = s_cols
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+        self.num_tmem_alloc_cols = utils.get_num_tmem_alloc_cols([tCtS_fake, tCtO_fake], arch="sm_100")
+
+        # ⛔⛔⛔ CRITICAL: num_tma_load_bytes MUST include ALL TMA-loaded tensors (Q + K + V). Missing V → DEADLOCK. See FOOTGUN #0 in README.
+        a_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        b_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        self.num_tma_load_bytes = (
+            cute.size_in_bytes(self.q_dtype, a_smem) + cute.size_in_bytes(self.q_dtype, b_smem) +
+            cute.size_in_bytes(self.q_dtype, v_smem)
+        ) * cute.size(qk_mma.thr_id.shape)
+
+    @cute.jit
+    def __call__(self, q, k, v, c, stream):
+        self.q_dtype = q.element_type; self.o_dtype = c.element_type; self.c_dtype = self.o_dtype
+        self.a_major = LayoutEnum.from_tensor(q).mma_major_mode()
+        self.b_major = LayoutEnum.from_tensor(k).mma_major_mode()
+        self.v_major = LayoutEnum.from_tensor(v).mma_major_mode()
+        self.c_layout = LayoutEnum.from_tensor(c)
+
+        qk_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, self.a_major, self.b_major,
+            self.qk_acc_dtype, self.cta_group, self.mma_tiler_mn, tcgen05.OperandSource.SMEM)
+        # PV with 128x128 output (V=I)
+        pv_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, cute.nvgpu.OperandMajorMode.K, self.v_major,
+            self.qk_acc_dtype, self.cta_group, self.mma_tiler_mn, tcgen05.OperandSource.TMEM)
+        self._setup(qk_mma, pv_mma)
+
+        q_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        k_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        tma_q, tma_tq = cute.nvgpu.make_tiled_tma_atom_A(
+            utils.sm100.cluster_shape_to_tma_atom_A(self.cluster_shape_mn, qk_mma.thr_id),
+            q, q_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_k, tma_tk = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, qk_mma.thr_id),
+            k, k_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_v, tma_tv = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, pv_mma.thr_id),
+            v, v_smem, self.pv_mma_tiler, pv_mma, self.cluster_layout_vmnk.shape)
+        epi_smem = cute.select(self.c_smem_s, mode=[0, 1])
+        tma_c, tma_tc = cpasync.make_tiled_tma_atom(cpasync.CopyBulkTensorTileS2GOp(), c, epi_smem, self.epi_tile)
+
+        self._kernel(qk_mma, pv_mma, tma_q, tma_tq, tma_k, tma_tk, tma_v, tma_tv,
+            tma_c, tma_tc, self.cluster_layout_vmnk,
+            self.a_smem_s, self.b_smem_s, self.v_smem_s, self.p_tmem_s, self.c_smem_s, self.epi_tile
+        ).launch(grid=(1,1,1), block=[self.threads_per_cta,1,1], stream=stream)
+
+    @cute.kernel
+    def _kernel(self, qk_mma, pv_mma, tma_q, mQ, tma_k, mK, tma_v, mV,
+                tma_c, mC, cl_vmnk, a_smem_s, b_smem_s, v_smem_s, p_tmem_s, c_smem_s, epi_tile):
+        warp_idx = cute.arch.make_warp_uniform(cute.arch.warp_idx())
+        tidx, _, _ = cute.arch.thread_idx()
+        use_2cta = cute.size(qk_mma.thr_id.shape) == 2
+
+        if warp_idx == self.tma_warp_id:
+            cpasync.prefetch_descriptor(tma_q); cpasync.prefetch_descriptor(tma_k)
+            cpasync.prefetch_descriptor(tma_v); cpasync.prefetch_descriptor(tma_c)
+
+        @cute.struct
+        class SS:
+            ab_bar: cute.struct.MemRange[cutlass.Int64, self.num_ab_stage * 2]
+            mma_si_bar: cute.struct.MemRange[cutlass.Int64, 2]
+            acc_bar: cute.struct.MemRange[cutlass.Int64, self.num_acc_stage * 2]
+            tmem_dealloc: cutlass.Int64
+            holding: cutlass.Int32
+
+        smem = utils.SmemAllocator(); st = smem.allocate(SS)
+
+        ab_p, ab_c = pipeline.PipelineTmaUmma.create(
+            barrier_storage=st.ab_bar.data_ptr(), num_stages=self.num_ab_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 1),
+            tx_count=self.num_tma_load_bytes, cta_layout_vmnk=cl_vmnk, defer_sync=True
+        ).make_participants()
+
+        mma_si_prod, mma_si_cons = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.mma_si_bar.data_ptr(), num_stages=1,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id)),
+        ).make_participants()
+
+        acc_pipe = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.acc_bar.data_ptr(), num_stages=self.num_acc_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(
+                pipeline.Agent.Thread, len(self.epilogue_warp_id) * (2 if use_2cta else 1)),
+            cta_layout_vmnk=cl_vmnk, defer_sync=True)
+
+        tmem_bar = pipeline.NamedBarrier(barrier_id=2,
+            num_threads=32 * len((self.mma_warp_id, *self.epilogue_warp_id)))
+        tmem = utils.TmemAllocator(st.holding.ptr, barrier_for_retrieve=tmem_bar,
+            allocator_warp_id=self.epilogue_warp_id[0], is_two_cta=use_2cta,
+            two_cta_tmem_dealloc_mbar_ptr=st.tmem_dealloc.ptr)
+
+        pipeline.pipeline_init_arrive(cluster_shape_mn=cl_vmnk, is_relaxed=True)
+
+        sQ = smem.allocate_tensor(element_type=self.q_dtype, layout=a_smem_s.outer, byte_alignment=128, swizzle=a_smem_s.inner)
+        sK = smem.allocate_tensor(element_type=self.q_dtype, layout=b_smem_s.outer, byte_alignment=128, swizzle=b_smem_s.inner)
+        sV = smem.allocate_tensor(element_type=self.q_dtype, layout=v_smem_s.outer, byte_alignment=128, swizzle=v_smem_s.inner)
+        sC = smem.allocate_tensor(element_type=self.o_dtype, layout=c_smem_s.outer, byte_alignment=128, swizzle=c_smem_s.inner)
+
+        gQ = cute.local_tile(mQ, cute.slice_(self.qk_mma_tiler, (None,0,None)), (None,None,None))
+        gK = cute.local_tile(mK, cute.slice_(self.qk_mma_tiler, (0,None,None)), (None,None,None))
+        gC = cute.local_tile(mC, cute.slice_(self.qk_mma_tiler, (None,None,0)), (None,None,None))
+        k_cnt = cute.size(gQ, mode=[3])
+
+        qk_thr = qk_mma.get_slice(0)
+        pv_thr = pv_mma.get_slice(0)
+        tCgQ = qk_thr.partition_A(gQ); tCgK = qk_thr.partition_B(gK); tCgC = qk_thr.partition_C(gC)
+        a_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,0,None,0)).shape)
+        tAsQ, tAgQ = cpasync.tma_partition(tma_q, 0, a_lay, cute.group_modes(sQ,0,3), cute.group_modes(tCgQ,0,3))
+        b_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,None,0,0)).shape)
+        tBsK, tBgK = cpasync.tma_partition(tma_k, 0, b_lay, cute.group_modes(sK,0,3), cute.group_modes(tCgK,0,3))
+        tAgQ = tAgQ[(None,0,None,0)]; tBgK = tBgK[(None,0,None,0)]
+
+        gV = cute.local_tile(mV, cute.slice_(self.pv_mma_tiler, (0,None,None)), (None,None,None))
+        tCgV = pv_thr.partition_B(gV)
+        tVsV, tVgV = cpasync.tma_partition(tma_v, 0, b_lay, cute.group_modes(sV,0,3), cute.group_modes(tCgV,0,3))
+        tVgV = tVgV[(None,0,None,0)]
+
+        tCrQ = qk_mma.make_fragment_A(sQ); tCrK = qk_mma.make_fragment_B(sK)
+        tCrV = pv_mma.make_fragment_B(sV)
+
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        tStS0 = cute.make_tensor(tStS.iterator + self.tmem_s0_offset, tStS.layout)
+
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        tOtO0 = cute.make_tensor(tOtO.iterator + self.tmem_o0_offset, tOtO.layout)
+
+        tP = cute.make_tensor(tStS.iterator, p_tmem_s.outer)
+        tOrP_base = pv_thr.make_fragment_A(tP)
+        tOrP = tOrP_base[(None, None, None, 0)]
+        tOrP0 = cute.make_tensor(
+            tOrP.iterator + self.qk_acc_dtype.width // self.q_dtype.width * self.tmem_p0_offset,
+            tOrP.layout)
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+
+        pipeline.pipeline_init_wait(cluster_shape_mn=cl_vmnk)
+
+        # ═══ TMA LOAD WARP ═══
+        if warp_idx == self.tma_warp_id:
+            ab_p.reset(); peek = ab_p.try_acquire()
+            for kt in cutlass.range(k_cnt, unroll=1):
+                h = ab_p.acquire_and_advance(peek)
+                cute.copy(tma_q, tAgQ[(None,h.count)], tAsQ[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_k, tBgK[(None,h.count)], tBsK[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_v, tVgV[(None,h.count)], tVsV[(None,h.index)], tma_bar_ptr=h.barrier)
+                peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_p.try_acquire()
+            ab_p.tail()
+
+        # ═══ MMA WARP ═══
+        if warp_idx == self.mma_warp_id:
+            tmem.wait_for_alloc()
+            ab_c.reset(); peek = ab_c.try_wait()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            acc_prod_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Producer, self.num_acc_stage)
+            acc_pipe.producer_acquire(acc_prod_st)
+            qk_mma.set(tcgen05.Field.ACCUMULATE, False)
+            for kt in range(k_cnt):
+                h = ab_c.wait_and_advance(peek)
+                nblk = cute.size(tCrQ, mode=[2])
+                for kb in cutlass.range(nblk, unroll_full=True):
+                    cute.gemm(qk_mma, tStS0, tCrQ[(None,None,kb,h.index)], tCrK[(None,None,kb,h.index)], tStS0)
+                    qk_mma.set(tcgen05.Field.ACCUMULATE, True)
+                h.release(); peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_c.try_wait()
+
+            cute.arch.fence_view_async_tmem_store()
+            s0_handle.commit()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            # PV MMA: P @ V where V=I → O = P
+            pv_mma.set(tcgen05.Field.ACCUMULATE, False)
+            tCrV_s = tCrV[(None, None, None, 0)]
+            nblk_pv = cute.size(tOrP0, mode=[2])
+            for kb in cutlass.range(nblk_pv, unroll_full=True):
+                cute.gemm(pv_mma, tOtO0, tOrP0[(None,None,kb)], tCrV_s[(None,None,kb)], tOtO0)
+                pv_mma.set(tcgen05.Field.ACCUMULATE, True)
+
+            acc_pipe.producer_commit(acc_prod_st)
+            acc_prod_st.advance()
+            acc_pipe.producer_tail(acc_prod_st)
+
+        # ═══ EPILOGUE WARPS ═══
+        if warp_idx < self.mma_warp_id:
+            tmem.allocate(self.num_tmem_alloc_cols)
+            tmem.wait_for_alloc()
+            tmem_ptr = tmem.retrieve_ptr(self.qk_acc_dtype)
+            sfw_idx = tidx % (32 * len(self.epilogue_warp_id))
+
+            tmem_load_atom = cute.make_copy_atom(
+                tcgen05.copy.Ld32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_load = tcgen05.make_tmem_copy(tmem_load_atom, tStS0)
+            thr_load = tiled_tmem_load.get_slice(sfw_idx)
+            tTMEM_LOADtS = thr_load.partition_S(tStS0)
+            cS = cute.make_identity_tensor((self.qk_mma_tiler[0], self.qk_mma_tiler[1]))
+            tScS = qk_thr.partition_C(cS)
+            tTMEM_LOADcS = thr_load.partition_D(tScS)
+
+            tStS_P_layout = cute.composition(tStS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tStS_P = cute.make_tensor(tStS.iterator + self.tmem_p0_offset, tStS_P_layout)
+            tmem_store_atom = cute.make_copy_atom(
+                tcgen05.copy.St32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_store = tcgen05.make_tmem_copy(tmem_store_atom, tStS_P)
+            thr_store = tiled_tmem_store.get_slice(sfw_idx)
+            tTMEM_STOREtS_x4 = thr_store.partition_D(tStS_P)
+            tScS_P_layout = cute.composition(tScS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tScS_P = cute.make_tensor(tScS.iterator, tScS_P_layout)
+            tTMEM_STOREcS = thr_store.partition_S(tScS_P)
+
+            si_handle = mma_si_cons.wait_and_advance()
+
+            tTMEM_LOADrS = cute.make_rmem_tensor(tTMEM_LOADcS.shape, self.qk_acc_dtype)
+            cute.copy(tiled_tmem_load, tTMEM_LOADtS, tTMEM_LOADrS)
+            tTMEM_STORErS_x4 = cute.make_rmem_tensor(tTMEM_STOREcS.shape, self.qk_acc_dtype)
+            tTMEM_STORErS_x4_e = cute.make_tensor(
+                cute.recast_ptr(tTMEM_STORErS_x4.iterator, dtype=self.q_dtype),
+                tTMEM_LOADrS.layout)
+            frg_cnt = 4
+            frg_tile = cute.size(tTMEM_LOADrS) // frg_cnt
+            tTMEM_LOADrS_frg = cute.logical_divide(tTMEM_LOADrS, cute.make_layout(frg_tile))
+            tTMEM_STORErS_x4_e_frg = cute.logical_divide(
+                tTMEM_STORErS_x4_e, cute.make_layout(frg_tile))
+            for j in range(frg_cnt):
+                s_vec = tTMEM_LOADrS_frg[None, j].load()
+                tTMEM_STORErS_x4_e_frg[None, j].store(s_vec.to(self.q_dtype))
+            cute.copy(tiled_tmem_store, tTMEM_STORErS_x4, tTMEM_STOREtS_x4)
+            cute.arch.fence_view_async_tmem_store()
+            si_handle.release()
+
+            # Output epilogue
+            tCtO_base = cute.make_tensor(tmem_ptr + self.tmem_o0_offset, tCtO_fake.layout)
+            acc_cons_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Consumer, self.num_acc_stage)
+            c_grp = pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id))
+            c_pipe = pipeline.PipelineTmaStore.create(num_stages=self.num_c_stage, producer_group=c_grp)
+            acc_cons_st = utils.gemm.sm100.epilogue_tma_store(
+                self, tidx, warp_idx, tma_c, tCtO_base, sC, tCgC,
+                epi_tile, 0, const_expr(lambda x: x), (0,0,0), acc_cons_st, acc_pipe, c_pipe)
+            c_pipe.producer_tail()
+            tmem.relinquish_alloc_permit()
+            tmem.free(tmem_ptr)
+
+
+def test():
+    torch.manual_seed(42)
+    m, n, head_dim = 128, 128, 64
+    q = torch.randn(m, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    k = torch.randn(n, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    # V = identity (128x128) in MN-major: (128,128) with strides (1,128)
+    v = torch.eye(128, dtype=torch.bfloat16, device='cuda')
+    # MN-major: (128,128) with strides (1,128) — row is fast dim
+    v = v.as_strided((128, 128), (1, 128)).unsqueeze(-1)
+    c = torch.zeros(m, n, 1, dtype=torch.bfloat16, device='cuda')
+
+    qf = q[:,:,0].float(); kf = k[:,:,0].float()
+    # With V=I and identity softmax: O = (Q@K^T).bf16() @ I = (Q@K^T).bf16()
+    ref = (qf @ kf.T).bfloat16().float()
+
+    mQ = ct.from_dlpack(q).mark_layout_dynamic(leading_dim=ct.get_leading_dim(q))
+    mK = ct.from_dlpack(k).mark_layout_dynamic(leading_dim=ct.get_leading_dim(k))
+    mV = ct.from_dlpack(v).mark_layout_dynamic(leading_dim=ct.get_leading_dim(v))
+    mC = ct.from_dlpack(c).mark_layout_dynamic(leading_dim=ct.get_leading_dim(c))
+    stream = cuda.CUstream(torch.cuda.current_stream().cuda_stream)
+    kernel = PvDiagKernel(mma_tiler_mn=(128, 128))
+    print('Compiling...', flush=True)
+    compiled = cute.compile(kernel, mQ, mK, mV, mC, stream)
+    print('Running...', flush=True)
+    compiled(mQ, mK, mV, mC, stream)
+    torch.cuda.synchronize()
+    out = c[:,:,0].float()
+    cos = torch.nn.functional.cosine_similarity(out.flatten().unsqueeze(0), ref.flatten().unsqueeze(0)).item()
+    print('PV diag (V=I): cosine {:.6f} {}'.format(cos, 'PASS' if cos >= 0.99 else 'FAIL'))
+
+
+if __name__ == '__main__':
+    test()

tests/test_tmem_col3.py

@@ -0,0 +1,392 @@
+"""
+Minimal PV-only test: Load P from GMEM to TMEM via QK-style MMA, then PV from TMEM.
+Step 1: QK MMA writes FP32 S to TMEM (we know this works)
+Step 2: Softmax packing writes BF16 P to TMEM (test this)
+Step 3: PV MMA reads BF16 P from TMEM and V from SMEM, produces O
+
+But to isolate the bug, let me test just the PV MMA in isolation.
+I'll write known BF16 values to TMEM using the softmax packing path,
+then immediately read them back using the PV A-fragment path,
+and compare.
+
+Actually, the simplest isolation test:
+1. Do QK MMA to get S in TMEM (cosine 0.999999 verified)
+2. Do softmax packing: S → P in TMEM (at offset 32)
+3. Skip PV entirely — read P from TMEM using the C-fragment composition LOAD path
+4. Output P to GMEM and compare against S.to(BF16)
+
+This tests whether the softmax packing writes P correctly to the same TMEM
+that the PV would read from.
+
+But we can't easily read P from TMEM using the standard epilogue path
+because the epilogue expects FP32 accumulator data.
+
+Alternative: Use the PV MMA with V=I (identity). If P is correct,
+then P @ I = P. But V needs to be MN-major and (128, 128), not (128, 64).
+The output would be (128, 128) which doesn't match our (128, 64) c tensor.
+
+Let me use V that selects the first 64 columns: V[k, n] = delta(k, n) for k in [0,63].
+This gives P @ V = P[:, :64], and the output is (128, 64).
+But V is (128, 128) in the MMA K,N dims. V[k, n] for k in [0,127], n in [0,63].
+Hmm, this is getting complicated. Let me just do the identity approach with a (128, 128) output.
+"""
+import torch, cutlass, cutlass.cute as cute, cutlass.utils as utils, cutlass.pipeline as pipeline
+from cutlass.cute.nvgpu import cpasync, tcgen05
+from cutlass import Float32, BFloat16, Int32, Boolean, const_expr
+from cutlass.utils import LayoutEnum
+from cutlass.utils.tmem_allocator import find_tmem_tensor_col_offset
+import cuda.bindings.driver as cuda
+import cutlass.torch as ct
+
+
+class TmemCol3Kernel:
+    """QK + softmax packing + PV with V=I to isolate PV MMA correctness.
+    Output should be P = S.to(BF16), i.e. (Q@K^T).bfloat16()
+    With V=I, O = P @ I = P.
+    But V is (K=128, N=128) in the MMA. We need a 128x128 identity in MN-major.
+    Output tensor is (128, 128).
+    """
+    def __init__(self, mma_tiler_mn):
+        self.acc_dtype = Float32; self.qk_acc_dtype = Float32
+        self.q_dtype = BFloat16; self.o_dtype = BFloat16; self.c_dtype = BFloat16
+        self.mma_tiler_mn = mma_tiler_mn; self.mma_tiler = (*mma_tiler_mn, 1)
+        self.use_2cta_instrs = False  # needed by epilogue_tma_store
+        self.epilog_sync_bar_id = 1  # needed by epilogue_tma_store
+        self.cluster_shape_mn = (1, 1)
+        self.cta_group = tcgen05.CtaGroup.ONE
+        self.epilogue_warp_id = (0, 1, 2, 3)
+        self.mma_warp_id = 4; self.tma_warp_id = 5
+        self.threads_per_cta = 192
+        self.num_c_stage = 2
+
+    def _setup(self, qk_mma, pv_mma):
+        qk_inst_k = int(cute.size(qk_mma.shape_mnk, mode=[2]))
+        self.qk_mma_tiler = (*self.mma_tiler_mn, qk_inst_k * 4)
+        # PV with V=I: output is (128, 128), same as QK
+        self.pv_mma_tiler = (self.qk_mma_tiler[0], self.qk_mma_tiler[1], self.qk_mma_tiler[1])
+        # pv_mma_tiler = (128, 128, 128) since V is 128x128
+        self.mma_tiler = self.qk_mma_tiler
+
+        self.cluster_layout_vmnk = cute.tiled_divide(cute.make_layout((1,1,1)), (qk_mma.thr_id.shape,))
+        self.cta_tile_shape_mnk = (
+            self.qk_mma_tiler[0] // cute.size(qk_mma.thr_id.shape),
+            self.qk_mma_tiler[1], self.qk_mma_tiler[2])
+        self.c_layout = LayoutEnum.ROW_MAJOR
+        self.epi_tile = utils.sm100.compute_epilogue_tile_shape(
+            self.cta_tile_shape_mnk, False, self.c_layout, self.o_dtype)
+        self.num_ab_stage = 1; self.num_acc_stage = 1
+
+        self.a_smem_s = utils.sm100.make_smem_layout_a(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.b_smem_s = utils.sm100.make_smem_layout_b(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.v_smem_s = utils.sm100.make_smem_layout_b(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.p_tmem_s = utils.sm100.make_smem_layout_a(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.c_smem_s = utils.sm100.make_smem_layout_epi(self.o_dtype, self.c_layout, self.epi_tile, 2)
+
+        qk_thr = qk_mma.get_slice(0)
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        s_cols = find_tmem_tensor_col_offset(tStS)
+        pv_thr = pv_mma.get_slice(0)
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        o_cols = find_tmem_tensor_col_offset(tOtO)
+
+        self.tilePlikeFP32 = self.qk_mma_tiler[1] // Float32.width * self.o_dtype.width
+        self.tmem_s0_offset = 0
+        self.tmem_p0_offset = 32
+        self.tmem_o0_offset = s_cols
+        # DIAGNOSTIC: compare layout sizes and shapes
+        tP = cute.make_tensor(tStS.iterator, self.p_tmem_s.outer)
+        tOrP2 = pv_mma.get_slice(0).make_fragment_A(tP)
+        tOrP2_s = tOrP2[(None, None, None, 0)]
+        print(f"LAYOUT: tStS_size={int(cute.size(tStS))} tP_size={int(cute.size(tP))} tOrP2_size={int(cute.size(tOrP2))} tOrP2_s_size={int(cute.size(tOrP2_s))}")
+        
+dims = [int(cute.size(tOrP2_s, mode=[i])) for i in range(tOrP2_s.layout.ndim)]
+print(f"LAYOUT: tOrP2_s_dims={dims}")
+        print(f"LAYOUT: pv_mma_tiler={self.pv_mma_tiler} tilePlikeFP32={self.tilePlikeFP32}")
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+        self.num_tmem_alloc_cols = utils.get_num_tmem_alloc_cols([tCtS_fake, tCtO_fake], arch="sm_100")
+
+        # ⛔⛔⛔ CRITICAL: num_tma_load_bytes MUST include ALL TMA-loaded tensors (Q + K + V). Missing V → DEADLOCK. See FOOTGUN #0 in README.
+        a_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        b_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        self.num_tma_load_bytes = (
+            cute.size_in_bytes(self.q_dtype, a_smem) + cute.size_in_bytes(self.q_dtype, b_smem) +
+            cute.size_in_bytes(self.q_dtype, v_smem)
+        ) * cute.size(qk_mma.thr_id.shape)
+
+    @cute.jit
+    def __call__(self, q, k, v, c, stream):
+        self.q_dtype = q.element_type; self.o_dtype = c.element_type; self.c_dtype = self.o_dtype
+        self.a_major = LayoutEnum.from_tensor(q).mma_major_mode()
+        self.b_major = LayoutEnum.from_tensor(k).mma_major_mode()
+        self.v_major = LayoutEnum.from_tensor(v).mma_major_mode()
+        self.c_layout = LayoutEnum.from_tensor(c)
+
+        qk_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, self.a_major, self.b_major,
+            self.qk_acc_dtype, self.cta_group, self.mma_tiler_mn, tcgen05.OperandSource.SMEM)
+        # PV with 128x128 output (V=I)
+        pv_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, cute.nvgpu.OperandMajorMode.K, self.v_major,
+            self.qk_acc_dtype, self.cta_group, self.mma_tiler_mn, tcgen05.OperandSource.TMEM)
+        self._setup(qk_mma, pv_mma)
+
+        q_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        k_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        tma_q, tma_tq = cute.nvgpu.make_tiled_tma_atom_A(
+            utils.sm100.cluster_shape_to_tma_atom_A(self.cluster_shape_mn, qk_mma.thr_id),
+            q, q_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_k, tma_tk = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, qk_mma.thr_id),
+            k, k_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_v, tma_tv = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, pv_mma.thr_id),
+            v, v_smem, self.pv_mma_tiler, pv_mma, self.cluster_layout_vmnk.shape)
+        epi_smem = cute.select(self.c_smem_s, mode=[0, 1])
+        tma_c, tma_tc = cpasync.make_tiled_tma_atom(cpasync.CopyBulkTensorTileS2GOp(), c, epi_smem, self.epi_tile)
+
+        self._kernel(qk_mma, pv_mma, tma_q, tma_tq, tma_k, tma_tk, tma_v, tma_tv,
+            tma_c, tma_tc, self.cluster_layout_vmnk,
+            self.a_smem_s, self.b_smem_s, self.v_smem_s, self.p_tmem_s, self.c_smem_s, self.epi_tile
+        ).launch(grid=(1,1,1), block=[self.threads_per_cta,1,1], stream=stream)
+
+    @cute.kernel
+    def _kernel(self, qk_mma, pv_mma, tma_q, mQ, tma_k, mK, tma_v, mV,
+                tma_c, mC, cl_vmnk, a_smem_s, b_smem_s, v_smem_s, p_tmem_s, c_smem_s, epi_tile):
+        warp_idx = cute.arch.make_warp_uniform(cute.arch.warp_idx())
+        tidx, _, _ = cute.arch.thread_idx()
+        use_2cta = cute.size(qk_mma.thr_id.shape) == 2
+
+        if warp_idx == self.tma_warp_id:
+            cpasync.prefetch_descriptor(tma_q); cpasync.prefetch_descriptor(tma_k)
+            cpasync.prefetch_descriptor(tma_v); cpasync.prefetch_descriptor(tma_c)
+
+        @cute.struct
+        class SS:
+            ab_bar: cute.struct.MemRange[cutlass.Int64, self.num_ab_stage * 2]
+            mma_si_bar: cute.struct.MemRange[cutlass.Int64, 2]
+            acc_bar: cute.struct.MemRange[cutlass.Int64, self.num_acc_stage * 2]
+            tmem_dealloc: cutlass.Int64
+            holding: cutlass.Int32
+
+        smem = utils.SmemAllocator(); st = smem.allocate(SS)
+
+        ab_p, ab_c = pipeline.PipelineTmaUmma.create(
+            barrier_storage=st.ab_bar.data_ptr(), num_stages=self.num_ab_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 1),
+            tx_count=self.num_tma_load_bytes, cta_layout_vmnk=cl_vmnk, defer_sync=True
+        ).make_participants()
+
+        mma_si_prod, mma_si_cons = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.mma_si_bar.data_ptr(), num_stages=1,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id)),
+        ).make_participants()
+
+        acc_pipe = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.acc_bar.data_ptr(), num_stages=self.num_acc_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(
+                pipeline.Agent.Thread, len(self.epilogue_warp_id) * (2 if use_2cta else 1)),
+            cta_layout_vmnk=cl_vmnk, defer_sync=True)
+
+        tmem_bar = pipeline.NamedBarrier(barrier_id=2,
+            num_threads=32 * len((self.mma_warp_id, *self.epilogue_warp_id)))
+        tmem = utils.TmemAllocator(st.holding.ptr, barrier_for_retrieve=tmem_bar,
+            allocator_warp_id=self.epilogue_warp_id[0], is_two_cta=use_2cta,
+            two_cta_tmem_dealloc_mbar_ptr=st.tmem_dealloc.ptr)
+
+        pipeline.pipeline_init_arrive(cluster_shape_mn=cl_vmnk, is_relaxed=True)
+
+        sQ = smem.allocate_tensor(element_type=self.q_dtype, layout=a_smem_s.outer, byte_alignment=128, swizzle=a_smem_s.inner)
+        sK = smem.allocate_tensor(element_type=self.q_dtype, layout=b_smem_s.outer, byte_alignment=128, swizzle=b_smem_s.inner)
+        sV = smem.allocate_tensor(element_type=self.q_dtype, layout=v_smem_s.outer, byte_alignment=128, swizzle=v_smem_s.inner)
+        sC = smem.allocate_tensor(element_type=self.o_dtype, layout=c_smem_s.outer, byte_alignment=128, swizzle=c_smem_s.inner)
+
+        gQ = cute.local_tile(mQ, cute.slice_(self.qk_mma_tiler, (None,0,None)), (None,None,None))
+        gK = cute.local_tile(mK, cute.slice_(self.qk_mma_tiler, (0,None,None)), (None,None,None))
+        gC = cute.local_tile(mC, cute.slice_(self.qk_mma_tiler, (None,None,0)), (None,None,None))
+        k_cnt = cute.size(gQ, mode=[3])
+
+        qk_thr = qk_mma.get_slice(0)
+        pv_thr = pv_mma.get_slice(0)
+        tCgQ = qk_thr.partition_A(gQ); tCgK = qk_thr.partition_B(gK); tCgC = qk_thr.partition_C(gC)
+        a_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,0,None,0)).shape)
+        tAsQ, tAgQ = cpasync.tma_partition(tma_q, 0, a_lay, cute.group_modes(sQ,0,3), cute.group_modes(tCgQ,0,3))
+        b_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,None,0,0)).shape)
+        tBsK, tBgK = cpasync.tma_partition(tma_k, 0, b_lay, cute.group_modes(sK,0,3), cute.group_modes(tCgK,0,3))
+        tAgQ = tAgQ[(None,0,None,0)]; tBgK = tBgK[(None,0,None,0)]
+
+        gV = cute.local_tile(mV, cute.slice_(self.pv_mma_tiler, (0,None,None)), (None,None,None))
+        tCgV = pv_thr.partition_B(gV)
+        tVsV, tVgV = cpasync.tma_partition(tma_v, 0, b_lay, cute.group_modes(sV,0,3), cute.group_modes(tCgV,0,3))
+        tVgV = tVgV[(None,0,None,0)]
+
+        tCrQ = qk_mma.make_fragment_A(sQ); tCrK = qk_mma.make_fragment_B(sK)
+        tCrV = pv_mma.make_fragment_B(sV)
+
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        tStS0 = cute.make_tensor(tStS.iterator + self.tmem_s0_offset, tStS.layout)
+
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        tOtO0 = cute.make_tensor(tOtO.iterator + self.tmem_o0_offset, tOtO.layout)
+
+        tP = cute.make_tensor(tStS.iterator, p_tmem_s.outer)
+        tOrP_base = pv_thr.make_fragment_A(tP)
+        tOrP = tOrP_base[(None, None, None, 0)]
+        tOrP0 = cute.make_tensor(
+            tOrP.iterator + self.qk_acc_dtype.width // self.q_dtype.width * self.tmem_p0_offset,
+            tOrP.layout)
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+
+        pipeline.pipeline_init_wait(cluster_shape_mn=cl_vmnk)
+
+        # ═══ TMA LOAD WARP ═══
+        if warp_idx == self.tma_warp_id:
+            ab_p.reset(); peek = ab_p.try_acquire()
+            for kt in cutlass.range(k_cnt, unroll=1):
+                h = ab_p.acquire_and_advance(peek)
+                cute.copy(tma_q, tAgQ[(None,h.count)], tAsQ[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_k, tBgK[(None,h.count)], tBsK[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_v, tVgV[(None,h.count)], tVsV[(None,h.index)], tma_bar_ptr=h.barrier)
+                peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_p.try_acquire()
+            ab_p.tail()
+
+        # ═══ MMA WARP ═══
+        if warp_idx == self.mma_warp_id:
+            tmem.wait_for_alloc()
+            ab_c.reset(); peek = ab_c.try_wait()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            acc_prod_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Producer, self.num_acc_stage)
+            acc_pipe.producer_acquire(acc_prod_st)
+            qk_mma.set(tcgen05.Field.ACCUMULATE, False)
+            for kt in range(k_cnt):
+                h = ab_c.wait_and_advance(peek)
+                nblk = cute.size(tCrQ, mode=[2])
+                for kb in cutlass.range(nblk, unroll_full=True):
+                    cute.gemm(qk_mma, tStS0, tCrQ[(None,None,kb,h.index)], tCrK[(None,None,kb,h.index)], tStS0)
+                    qk_mma.set(tcgen05.Field.ACCUMULATE, True)
+                h.release(); peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_c.try_wait()
+
+            cute.arch.fence_view_async_tmem_store()
+            s0_handle.commit()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            # PV MMA: P @ V where V=I → O = P
+            pv_mma.set(tcgen05.Field.ACCUMULATE, False)
+            tCrV_s = tCrV[(None, None, None, 0)]
+            nblk_pv = cute.size(tOrP0, mode=[2])
+            for kb in cutlass.range(nblk_pv, unroll_full=True):
+                cute.gemm(pv_mma, tOtO0, tOrP0[(None,None,kb)], tCrV_s[(None,None,kb)], tOtO0)
+                pv_mma.set(tcgen05.Field.ACCUMULATE, True)
+
+            acc_pipe.producer_commit(acc_prod_st)
+            acc_prod_st.advance()
+            acc_pipe.producer_tail(acc_prod_st)
+
+        # ═══ EPILOGUE WARPS ═══
+        if warp_idx < self.mma_warp_id:
+            tmem.allocate(self.num_tmem_alloc_cols)
+            tmem.wait_for_alloc()
+            tmem_ptr = tmem.retrieve_ptr(self.qk_acc_dtype)
+            sfw_idx = tidx % (32 * len(self.epilogue_warp_id))
+
+            tmem_load_atom = cute.make_copy_atom(
+                tcgen05.copy.Ld32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_load = tcgen05.make_tmem_copy(tmem_load_atom, tStS0)
+            thr_load = tiled_tmem_load.get_slice(sfw_idx)
+            tTMEM_LOADtS = thr_load.partition_S(tStS0)
+            cS = cute.make_identity_tensor((self.qk_mma_tiler[0], self.qk_mma_tiler[1]))
+            tScS = qk_thr.partition_C(cS)
+            tTMEM_LOADcS = thr_load.partition_D(tScS)
+
+            tStS_P_layout = cute.composition(tStS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tStS_P = cute.make_tensor(tStS.iterator + self.tmem_p0_offset, tStS_P_layout)
+            tmem_store_atom = cute.make_copy_atom(
+                tcgen05.copy.St32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_store = tcgen05.make_tmem_copy(tmem_store_atom, tStS_P)
+            thr_store = tiled_tmem_store.get_slice(sfw_idx)
+            tTMEM_STOREtS_x4 = thr_store.partition_D(tStS_P)
+            tScS_P_layout = cute.composition(tScS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tScS_P = cute.make_tensor(tScS.iterator, tScS_P_layout)
+            tTMEM_STOREcS = thr_store.partition_S(tScS_P)
+
+            si_handle = mma_si_cons.wait_and_advance()
+
+            tTMEM_LOADrS = cute.make_rmem_tensor(tTMEM_LOADcS.shape, self.qk_acc_dtype)
+            cute.copy(tiled_tmem_load, tTMEM_LOADtS, tTMEM_LOADrS)
+            tTMEM_STORErS_x4 = cute.make_rmem_tensor(tTMEM_STOREcS.shape, self.qk_acc_dtype)
+            tTMEM_STORErS_x4_e = cute.make_tensor(
+                cute.recast_ptr(tTMEM_STORErS_x4.iterator, dtype=self.q_dtype),
+                tTMEM_LOADrS.layout)
+            frg_cnt = 4
+            frg_tile = cute.size(tTMEM_LOADrS) // frg_cnt
+            tTMEM_LOADrS_frg = cute.logical_divide(tTMEM_LOADrS, cute.make_layout(frg_tile))
+            tTMEM_STORErS_x4_e_frg = cute.logical_divide(
+                tTMEM_STORErS_x4_e, cute.make_layout(frg_tile))
+            for j in range(frg_cnt):
+                s_vec = tTMEM_LOADrS_frg[None, j].load()
+                tTMEM_STORErS_x4_e_frg[None, j].store(s_vec.to(self.q_dtype))
+            cute.copy(tiled_tmem_store, tTMEM_STORErS_x4, tTMEM_STOREtS_x4)
+            cute.arch.fence_view_async_tmem_store()
+            si_handle.release()
+
+            # Output epilogue
+            tCtO_base = cute.make_tensor(tmem_ptr + self.tmem_o0_offset, tCtO_fake.layout)
+            acc_cons_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Consumer, self.num_acc_stage)
+            c_grp = pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id))
+            c_pipe = pipeline.PipelineTmaStore.create(num_stages=self.num_c_stage, producer_group=c_grp)
+            acc_cons_st = utils.gemm.sm100.epilogue_tma_store(
+                self, tidx, warp_idx, tma_c, tCtO_base, sC, tCgC,
+                epi_tile, 0, const_expr(lambda x: x), (0,0,0), acc_cons_st, acc_pipe, c_pipe)
+            c_pipe.producer_tail()
+            tmem.relinquish_alloc_permit()
+            tmem.free(tmem_ptr)
+
+
+def test():
+    torch.manual_seed(42)
+    m, n, head_dim = 128, 128, 64
+    q = torch.randn(m, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    k = torch.randn(n, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    # V = identity (128x128) in MN-major: (128,128) with strides (1,128)
+    v = torch.eye(128, dtype=torch.bfloat16, device='cuda')
+    # MN-major: (128,128) with strides (1,128) — row is fast dim
+    v = v.as_strided((128, 128), (1, 128)).unsqueeze(-1)
+    c = torch.zeros(m, n, 1, dtype=torch.bfloat16, device='cuda')
+
+    qf = q[:,:,0].float(); kf = k[:,:,0].float()
+    # With V=I and identity softmax: O = (Q@K^T).bf16() @ I = (Q@K^T).bf16()
+    ref = (qf @ kf.T).bfloat16().float()
+
+    mQ = ct.from_dlpack(q).mark_layout_dynamic(leading_dim=ct.get_leading_dim(q))
+    mK = ct.from_dlpack(k).mark_layout_dynamic(leading_dim=ct.get_leading_dim(k))
+    mV = ct.from_dlpack(v).mark_layout_dynamic(leading_dim=ct.get_leading_dim(v))
+    mC = ct.from_dlpack(c).mark_layout_dynamic(leading_dim=ct.get_leading_dim(c))
+    stream = cuda.CUstream(torch.cuda.current_stream().cuda_stream)
+    kernel = TmemCol3Kernel(mma_tiler_mn=(128, 128))
+    print('Compiling...', flush=True)
+    compiled = cute.compile(kernel, mQ, mK, mV, mC, stream)
+    print('Running...', flush=True)
+    compiled(mQ, mK, mV, mC, stream)
+    torch.cuda.synchronize()
+    out = c[:,:,0].float()
+    cos = torch.nn.functional.cosine_similarity(out.flatten().unsqueeze(0), ref.flatten().unsqueeze(0)).item()
+    print('TmemCol3(128,128): cosine {:.6f} {}'.format(cos, 'PASS' if cos >= 0.99 else 'FAIL'))
+
+
+if __name__ == '__main__':
+    test()

tests/test_tmem_col4.py

@@ -0,0 +1,388 @@
+"""
+Minimal PV-only test: Load P from GMEM to TMEM via QK-style MMA, then PV from TMEM.
+Step 1: QK MMA writes FP32 S to TMEM (we know this works)
+Step 2: Softmax packing writes BF16 P to TMEM (test this)
+Step 3: PV MMA reads BF16 P from TMEM and V from SMEM, produces O
+
+But to isolate the bug, let me test just the PV MMA in isolation.
+I'll write known BF16 values to TMEM using the softmax packing path,
+then immediately read them back using the PV A-fragment path,
+and compare.
+
+Actually, the simplest isolation test:
+1. Do QK MMA to get S in TMEM (cosine 0.999999 verified)
+2. Do softmax packing: S → P in TMEM (at offset 32)
+3. Skip PV entirely — read P from TMEM using the C-fragment composition LOAD path
+4. Output P to GMEM and compare against S.to(BF16)
+
+This tests whether the softmax packing writes P correctly to the same TMEM
+that the PV would read from.
+
+But we can't easily read P from TMEM using the standard epilogue path
+because the epilogue expects FP32 accumulator data.
+
+Alternative: Use the PV MMA with V=I (identity). If P is correct,
+then P @ I = P. But V needs to be MN-major and (128, 128), not (128, 64).
+The output would be (128, 128) which doesn't match our (128, 64) c tensor.
+
+Let me use V that selects the first 64 columns: V[k, n] = delta(k, n) for k in [0,63].
+This gives P @ V = P[:, :64], and the output is (128, 64).
+But V is (128, 128) in the MMA K,N dims. V[k, n] for k in [0,127], n in [0,63].
+Hmm, this is getting complicated. Let me just do the identity approach with a (128, 128) output.
+"""
+import torch, cutlass, cutlass.cute as cute, cutlass.utils as utils, cutlass.pipeline as pipeline
+from cutlass.cute.nvgpu import cpasync, tcgen05
+from cutlass import Float32, BFloat16, Int32, Boolean, const_expr
+from cutlass.utils import LayoutEnum
+from cutlass.utils.tmem_allocator import find_tmem_tensor_col_offset
+import cuda.bindings.driver as cuda
+import cutlass.torch as ct
+
+
+class TmemCol4:
+    """QK + softmax packing + PV with V=I to isolate PV MMA correctness.
+    Output should be P = S.to(BF16), i.e. (Q@K^T).bfloat16()
+    With V=I, O = P @ I = P.
+    But V is (K=128, N=128) in the MMA. We need a 128x128 identity in MN-major.
+    Output tensor is (128, 128).
+    """
+    def __init__(self, mma_tiler_mn):
+        self.acc_dtype = Float32; self.qk_acc_dtype = Float32
+        self.q_dtype = BFloat16; self.o_dtype = BFloat16; self.c_dtype = BFloat16
+        self.mma_tiler_mn = mma_tiler_mn; self.mma_tiler = (*mma_tiler_mn, 1)
+        self.use_2cta_instrs = False  # needed by epilogue_tma_store
+        self.epilog_sync_bar_id = 1  # needed by epilogue_tma_store
+        self.cluster_shape_mn = (1, 1)
+        self.cta_group = tcgen05.CtaGroup.ONE
+        self.epilogue_warp_id = (0, 1, 2, 3)
+        self.mma_warp_id = 4; self.tma_warp_id = 5
+        self.threads_per_cta = 192
+        self.num_c_stage = 2
+
+    def _setup(self, qk_mma, pv_mma):
+        qk_inst_k = int(cute.size(qk_mma.shape_mnk, mode=[2]))
+        self.qk_mma_tiler = (*self.mma_tiler_mn, qk_inst_k * 4)
+        # PV with V=I: output is (128, 128), same as QK
+        self.pv_mma_tiler = (self.qk_mma_tiler[0], self.qk_mma_tiler[1], self.qk_mma_tiler[1])
+        # pv_mma_tiler = (128, 128, 128) since V is 128x128
+        self.mma_tiler = self.qk_mma_tiler
+
+        self.cluster_layout_vmnk = cute.tiled_divide(cute.make_layout((1,1,1)), (qk_mma.thr_id.shape,))
+        self.cta_tile_shape_mnk = (
+            self.qk_mma_tiler[0] // cute.size(qk_mma.thr_id.shape),
+            self.qk_mma_tiler[1], self.qk_mma_tiler[2])
+        self.c_layout = LayoutEnum.ROW_MAJOR
+        self.epi_tile = utils.sm100.compute_epilogue_tile_shape(
+            self.cta_tile_shape_mnk, False, self.c_layout, self.o_dtype)
+        self.num_ab_stage = 1; self.num_acc_stage = 1
+
+        self.a_smem_s = utils.sm100.make_smem_layout_a(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.b_smem_s = utils.sm100.make_smem_layout_b(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.v_smem_s = utils.sm100.make_smem_layout_b(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.p_tmem_s = utils.sm100.make_smem_layout_a(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.c_smem_s = utils.sm100.make_smem_layout_epi(self.o_dtype, self.c_layout, self.epi_tile, 2)
+
+        qk_thr = qk_mma.get_slice(0)
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        s_cols = find_tmem_tensor_col_offset(tStS)
+        pv_thr = pv_mma.get_slice(0)
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        o_cols = find_tmem_tensor_col_offset(tOtO)
+
+        self.tilePlikeFP32 = self.qk_mma_tiler[1] // Float32.width * self.o_dtype.width
+        self.tmem_s0_offset = 0
+        self.tmem_p0_offset = 32
+        self.tmem_o0_offset = s_cols
+        tP = cute.make_tensor(tStS.iterator, self.p_tmem_s.outer)
+        tOrP2 = pv_mma.get_slice(0).make_fragment_A(tP)
+        tOrP2_s = tOrP2[(None, None, None, 0)]
+        ndim = tOrP2_s.layout.ndim
+        print(int(cute.size(tStS)), int(cute.size(tP)), int(cute.size(tOrP2)), int(cute.size(tOrP2_s)), int(ndim))
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+        self.num_tmem_alloc_cols = utils.get_num_tmem_alloc_cols([tCtS_fake, tCtO_fake], arch="sm_100")
+
+        # ⛔⛔⛔ CRITICAL: num_tma_load_bytes MUST include ALL TMA-loaded tensors (Q + K + V). Missing V → DEADLOCK. See FOOTGUN #0 in README.
+        a_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        b_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        self.num_tma_load_bytes = (
+            cute.size_in_bytes(self.q_dtype, a_smem) + cute.size_in_bytes(self.q_dtype, b_smem) +
+            cute.size_in_bytes(self.q_dtype, v_smem)
+        ) * cute.size(qk_mma.thr_id.shape)
+
+    @cute.jit
+    def __call__(self, q, k, v, c, stream):
+        self.q_dtype = q.element_type; self.o_dtype = c.element_type; self.c_dtype = self.o_dtype
+        self.a_major = LayoutEnum.from_tensor(q).mma_major_mode()
+        self.b_major = LayoutEnum.from_tensor(k).mma_major_mode()
+        self.v_major = LayoutEnum.from_tensor(v).mma_major_mode()
+        self.c_layout = LayoutEnum.from_tensor(c)
+
+        qk_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, self.a_major, self.b_major,
+            self.qk_acc_dtype, self.cta_group, self.mma_tiler_mn, tcgen05.OperandSource.SMEM)
+        # PV with 128x128 output (V=I)
+        pv_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, cute.nvgpu.OperandMajorMode.K, self.v_major,
+            self.qk_acc_dtype, self.cta_group, self.mma_tiler_mn, tcgen05.OperandSource.TMEM)
+        self._setup(qk_mma, pv_mma)
+
+        q_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        k_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        tma_q, tma_tq = cute.nvgpu.make_tiled_tma_atom_A(
+            utils.sm100.cluster_shape_to_tma_atom_A(self.cluster_shape_mn, qk_mma.thr_id),
+            q, q_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_k, tma_tk = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, qk_mma.thr_id),
+            k, k_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_v, tma_tv = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, pv_mma.thr_id),
+            v, v_smem, self.pv_mma_tiler, pv_mma, self.cluster_layout_vmnk.shape)
+        epi_smem = cute.select(self.c_smem_s, mode=[0, 1])
+        tma_c, tma_tc = cpasync.make_tiled_tma_atom(cpasync.CopyBulkTensorTileS2GOp(), c, epi_smem, self.epi_tile)
+
+        self._kernel(qk_mma, pv_mma, tma_q, tma_tq, tma_k, tma_tk, tma_v, tma_tv,
+            tma_c, tma_tc, self.cluster_layout_vmnk,
+            self.a_smem_s, self.b_smem_s, self.v_smem_s, self.p_tmem_s, self.c_smem_s, self.epi_tile
+        ).launch(grid=(1,1,1), block=[self.threads_per_cta,1,1], stream=stream)
+
+    @cute.kernel
+    def _kernel(self, qk_mma, pv_mma, tma_q, mQ, tma_k, mK, tma_v, mV,
+                tma_c, mC, cl_vmnk, a_smem_s, b_smem_s, v_smem_s, p_tmem_s, c_smem_s, epi_tile):
+        warp_idx = cute.arch.make_warp_uniform(cute.arch.warp_idx())
+        tidx, _, _ = cute.arch.thread_idx()
+        use_2cta = cute.size(qk_mma.thr_id.shape) == 2
+
+        if warp_idx == self.tma_warp_id:
+            cpasync.prefetch_descriptor(tma_q); cpasync.prefetch_descriptor(tma_k)
+            cpasync.prefetch_descriptor(tma_v); cpasync.prefetch_descriptor(tma_c)
+
+        @cute.struct
+        class SS:
+            ab_bar: cute.struct.MemRange[cutlass.Int64, self.num_ab_stage * 2]
+            mma_si_bar: cute.struct.MemRange[cutlass.Int64, 2]
+            acc_bar: cute.struct.MemRange[cutlass.Int64, self.num_acc_stage * 2]
+            tmem_dealloc: cutlass.Int64
+            holding: cutlass.Int32
+
+        smem = utils.SmemAllocator(); st = smem.allocate(SS)
+
+        ab_p, ab_c = pipeline.PipelineTmaUmma.create(
+            barrier_storage=st.ab_bar.data_ptr(), num_stages=self.num_ab_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 1),
+            tx_count=self.num_tma_load_bytes, cta_layout_vmnk=cl_vmnk, defer_sync=True
+        ).make_participants()
+
+        mma_si_prod, mma_si_cons = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.mma_si_bar.data_ptr(), num_stages=1,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id)),
+        ).make_participants()
+
+        acc_pipe = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.acc_bar.data_ptr(), num_stages=self.num_acc_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(
+                pipeline.Agent.Thread, len(self.epilogue_warp_id) * (2 if use_2cta else 1)),
+            cta_layout_vmnk=cl_vmnk, defer_sync=True)
+
+        tmem_bar = pipeline.NamedBarrier(barrier_id=2,
+            num_threads=32 * len((self.mma_warp_id, *self.epilogue_warp_id)))
+        tmem = utils.TmemAllocator(st.holding.ptr, barrier_for_retrieve=tmem_bar,
+            allocator_warp_id=self.epilogue_warp_id[0], is_two_cta=use_2cta,
+            two_cta_tmem_dealloc_mbar_ptr=st.tmem_dealloc.ptr)
+
+        pipeline.pipeline_init_arrive(cluster_shape_mn=cl_vmnk, is_relaxed=True)
+
+        sQ = smem.allocate_tensor(element_type=self.q_dtype, layout=a_smem_s.outer, byte_alignment=128, swizzle=a_smem_s.inner)
+        sK = smem.allocate_tensor(element_type=self.q_dtype, layout=b_smem_s.outer, byte_alignment=128, swizzle=b_smem_s.inner)
+        sV = smem.allocate_tensor(element_type=self.q_dtype, layout=v_smem_s.outer, byte_alignment=128, swizzle=v_smem_s.inner)
+        sC = smem.allocate_tensor(element_type=self.o_dtype, layout=c_smem_s.outer, byte_alignment=128, swizzle=c_smem_s.inner)
+
+        gQ = cute.local_tile(mQ, cute.slice_(self.qk_mma_tiler, (None,0,None)), (None,None,None))
+        gK = cute.local_tile(mK, cute.slice_(self.qk_mma_tiler, (0,None,None)), (None,None,None))
+        gC = cute.local_tile(mC, cute.slice_(self.qk_mma_tiler, (None,None,0)), (None,None,None))
+        k_cnt = cute.size(gQ, mode=[3])
+
+        qk_thr = qk_mma.get_slice(0)
+        pv_thr = pv_mma.get_slice(0)
+        tCgQ = qk_thr.partition_A(gQ); tCgK = qk_thr.partition_B(gK); tCgC = qk_thr.partition_C(gC)
+        a_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,0,None,0)).shape)
+        tAsQ, tAgQ = cpasync.tma_partition(tma_q, 0, a_lay, cute.group_modes(sQ,0,3), cute.group_modes(tCgQ,0,3))
+        b_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,None,0,0)).shape)
+        tBsK, tBgK = cpasync.tma_partition(tma_k, 0, b_lay, cute.group_modes(sK,0,3), cute.group_modes(tCgK,0,3))
+        tAgQ = tAgQ[(None,0,None,0)]; tBgK = tBgK[(None,0,None,0)]
+
+        gV = cute.local_tile(mV, cute.slice_(self.pv_mma_tiler, (0,None,None)), (None,None,None))
+        tCgV = pv_thr.partition_B(gV)
+        tVsV, tVgV = cpasync.tma_partition(tma_v, 0, b_lay, cute.group_modes(sV,0,3), cute.group_modes(tCgV,0,3))
+        tVgV = tVgV[(None,0,None,0)]
+
+        tCrQ = qk_mma.make_fragment_A(sQ); tCrK = qk_mma.make_fragment_B(sK)
+        tCrV = pv_mma.make_fragment_B(sV)
+
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        tStS0 = cute.make_tensor(tStS.iterator + self.tmem_s0_offset, tStS.layout)
+
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        tOtO0 = cute.make_tensor(tOtO.iterator + self.tmem_o0_offset, tOtO.layout)
+
+        tP = cute.make_tensor(tStS.iterator, p_tmem_s.outer)
+        tOrP_base = pv_thr.make_fragment_A(tP)
+        tOrP = tOrP_base[(None, None, None, 0)]
+        tOrP0 = cute.make_tensor(
+            tOrP.iterator + self.qk_acc_dtype.width // self.q_dtype.width * self.tmem_p0_offset,
+            tOrP.layout)
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+
+        pipeline.pipeline_init_wait(cluster_shape_mn=cl_vmnk)
+
+        # ═══ TMA LOAD WARP ═══
+        if warp_idx == self.tma_warp_id:
+            ab_p.reset(); peek = ab_p.try_acquire()
+            for kt in cutlass.range(k_cnt, unroll=1):
+                h = ab_p.acquire_and_advance(peek)
+                cute.copy(tma_q, tAgQ[(None,h.count)], tAsQ[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_k, tBgK[(None,h.count)], tBsK[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_v, tVgV[(None,h.count)], tVsV[(None,h.index)], tma_bar_ptr=h.barrier)
+                peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_p.try_acquire()
+            ab_p.tail()
+
+        # ═══ MMA WARP ═══
+        if warp_idx == self.mma_warp_id:
+            tmem.wait_for_alloc()
+            ab_c.reset(); peek = ab_c.try_wait()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            acc_prod_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Producer, self.num_acc_stage)
+            acc_pipe.producer_acquire(acc_prod_st)
+            qk_mma.set(tcgen05.Field.ACCUMULATE, False)
+            for kt in range(k_cnt):
+                h = ab_c.wait_and_advance(peek)
+                nblk = cute.size(tCrQ, mode=[2])
+                for kb in cutlass.range(nblk, unroll_full=True):
+                    cute.gemm(qk_mma, tStS0, tCrQ[(None,None,kb,h.index)], tCrK[(None,None,kb,h.index)], tStS0)
+                    qk_mma.set(tcgen05.Field.ACCUMULATE, True)
+                h.release(); peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_c.try_wait()
+
+            cute.arch.fence_view_async_tmem_store()
+            s0_handle.commit()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            # PV MMA: P @ V where V=I → O = P
+            pv_mma.set(tcgen05.Field.ACCUMULATE, False)
+            tCrV_s = tCrV[(None, None, None, 0)]
+            nblk_pv = cute.size(tOrP0, mode=[2])
+            for kb in cutlass.range(nblk_pv, unroll_full=True):
+                cute.gemm(pv_mma, tOtO0, tOrP0[(None,None,kb)], tCrV_s[(None,None,kb)], tOtO0)
+                pv_mma.set(tcgen05.Field.ACCUMULATE, True)
+
+            acc_pipe.producer_commit(acc_prod_st)
+            acc_prod_st.advance()
+            acc_pipe.producer_tail(acc_prod_st)
+
+        # ═══ EPILOGUE WARPS ═══
+        if warp_idx < self.mma_warp_id:
+            tmem.allocate(self.num_tmem_alloc_cols)
+            tmem.wait_for_alloc()
+            tmem_ptr = tmem.retrieve_ptr(self.qk_acc_dtype)
+            sfw_idx = tidx % (32 * len(self.epilogue_warp_id))
+
+            tmem_load_atom = cute.make_copy_atom(
+                tcgen05.copy.Ld32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_load = tcgen05.make_tmem_copy(tmem_load_atom, tStS0)
+            thr_load = tiled_tmem_load.get_slice(sfw_idx)
+            tTMEM_LOADtS = thr_load.partition_S(tStS0)
+            cS = cute.make_identity_tensor((self.qk_mma_tiler[0], self.qk_mma_tiler[1]))
+            tScS = qk_thr.partition_C(cS)
+            tTMEM_LOADcS = thr_load.partition_D(tScS)
+
+            tStS_P_layout = cute.composition(tStS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tStS_P = cute.make_tensor(tStS.iterator + self.tmem_p0_offset, tStS_P_layout)
+            tmem_store_atom = cute.make_copy_atom(
+                tcgen05.copy.St32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_store = tcgen05.make_tmem_copy(tmem_store_atom, tStS_P)
+            thr_store = tiled_tmem_store.get_slice(sfw_idx)
+            tTMEM_STOREtS_x4 = thr_store.partition_D(tStS_P)
+            tScS_P_layout = cute.composition(tScS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tScS_P = cute.make_tensor(tScS.iterator, tScS_P_layout)
+            tTMEM_STOREcS = thr_store.partition_S(tScS_P)
+
+            si_handle = mma_si_cons.wait_and_advance()
+
+            tTMEM_LOADrS = cute.make_rmem_tensor(tTMEM_LOADcS.shape, self.qk_acc_dtype)
+            cute.copy(tiled_tmem_load, tTMEM_LOADtS, tTMEM_LOADrS)
+            tTMEM_STORErS_x4 = cute.make_rmem_tensor(tTMEM_STOREcS.shape, self.qk_acc_dtype)
+            tTMEM_STORErS_x4_e = cute.make_tensor(
+                cute.recast_ptr(tTMEM_STORErS_x4.iterator, dtype=self.q_dtype),
+                tTMEM_LOADrS.layout)
+            frg_cnt = 4
+            frg_tile = cute.size(tTMEM_LOADrS) // frg_cnt
+            tTMEM_LOADrS_frg = cute.logical_divide(tTMEM_LOADrS, cute.make_layout(frg_tile))
+            tTMEM_STORErS_x4_e_frg = cute.logical_divide(
+                tTMEM_STORErS_x4_e, cute.make_layout(frg_tile))
+            for j in range(frg_cnt):
+                s_vec = tTMEM_LOADrS_frg[None, j].load()
+                tTMEM_STORErS_x4_e_frg[None, j].store(s_vec.to(self.q_dtype))
+            cute.copy(tiled_tmem_store, tTMEM_STORErS_x4, tTMEM_STOREtS_x4)
+            cute.arch.fence_view_async_tmem_store()
+            si_handle.release()
+
+            # Output epilogue
+            tCtO_base = cute.make_tensor(tmem_ptr + self.tmem_o0_offset, tCtO_fake.layout)
+            acc_cons_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Consumer, self.num_acc_stage)
+            c_grp = pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id))
+            c_pipe = pipeline.PipelineTmaStore.create(num_stages=self.num_c_stage, producer_group=c_grp)
+            acc_cons_st = utils.gemm.sm100.epilogue_tma_store(
+                self, tidx, warp_idx, tma_c, tCtO_base, sC, tCgC,
+                epi_tile, 0, const_expr(lambda x: x), (0,0,0), acc_cons_st, acc_pipe, c_pipe)
+            c_pipe.producer_tail()
+            tmem.relinquish_alloc_permit()
+            tmem.free(tmem_ptr)
+
+
+def test():
+    torch.manual_seed(42)
+    m, n, head_dim = 128, 128, 64
+    q = torch.randn(m, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    k = torch.randn(n, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    # V = identity (128x128) in MN-major: (128,128) with strides (1,128)
+    v = torch.eye(128, dtype=torch.bfloat16, device='cuda')
+    # MN-major: (128,128) with strides (1,128) — row is fast dim
+    v = v.as_strided((128, 128), (1, 128)).unsqueeze(-1)
+    c = torch.zeros(m, n, 1, dtype=torch.bfloat16, device='cuda')
+
+    qf = q[:,:,0].float(); kf = k[:,:,0].float()
+    # With V=I and identity softmax: O = (Q@K^T).bf16() @ I = (Q@K^T).bf16()
+    ref = (qf @ kf.T).bfloat16().float()
+
+    mQ = ct.from_dlpack(q).mark_layout_dynamic(leading_dim=ct.get_leading_dim(q))
+    mK = ct.from_dlpack(k).mark_layout_dynamic(leading_dim=ct.get_leading_dim(k))
+    mV = ct.from_dlpack(v).mark_layout_dynamic(leading_dim=ct.get_leading_dim(v))
+    mC = ct.from_dlpack(c).mark_layout_dynamic(leading_dim=ct.get_leading_dim(c))
+    stream = cuda.CUstream(torch.cuda.current_stream().cuda_stream)
+    kernel = TmemCol4(mma_tiler_mn=(128, 128))
+    print('Compiling...', flush=True)
+    compiled = cute.compile(kernel, mQ, mK, mV, mC, stream)
+    print('Running...', flush=True)
+    compiled(mQ, mK, mV, mC, stream)
+    torch.cuda.synchronize()
+    out = c[:,:,0].float()
+    cos = torch.nn.functional.cosine_similarity(out.flatten().unsqueeze(0), ref.flatten().unsqueeze(0)).item()
+    print('Col4: cosine {:.6f} {}'.format(cos, 'PASS' if cos >= 0.99 else 'FAIL'))
+
+
+if __name__ == '__main__':
+    test()

tests/test_tmem_col5.py

@@ -0,0 +1,392 @@
+"""
+Minimal PV-only test: Load P from GMEM to TMEM via QK-style MMA, then PV from TMEM.
+Step 1: QK MMA writes FP32 S to TMEM (we know this works)
+Step 2: Softmax packing writes BF16 P to TMEM (test this)
+Step 3: PV MMA reads BF16 P from TMEM and V from SMEM, produces O
+
+But to isolate the bug, let me test just the PV MMA in isolation.
+I'll write known BF16 values to TMEM using the softmax packing path,
+then immediately read them back using the PV A-fragment path,
+and compare.
+
+Actually, the simplest isolation test:
+1. Do QK MMA to get S in TMEM (cosine 0.999999 verified)
+2. Do softmax packing: S → P in TMEM (at offset 32)
+3. Skip PV entirely — read P from TMEM using the C-fragment composition LOAD path
+4. Output P to GMEM and compare against S.to(BF16)
+
+This tests whether the softmax packing writes P correctly to the same TMEM
+that the PV would read from.
+
+But we can't easily read P from TMEM using the standard epilogue path
+because the epilogue expects FP32 accumulator data.
+
+Alternative: Use the PV MMA with V=I (identity). If P is correct,
+then P @ I = P. But V needs to be MN-major and (128, 128), not (128, 64).
+The output would be (128, 128) which doesn't match our (128, 64) c tensor.
+
+Let me use V that selects the first 64 columns: V[k, n] = delta(k, n) for k in [0,63].
+This gives P @ V = P[:, :64], and the output is (128, 64).
+But V is (128, 128) in the MMA K,N dims. V[k, n] for k in [0,127], n in [0,63].
+Hmm, this is getting complicated. Let me just do the identity approach with a (128, 128) output.
+"""
+import torch, cutlass, cutlass.cute as cute, cutlass.utils as utils, cutlass.pipeline as pipeline
+from cutlass.cute.nvgpu import cpasync, tcgen05
+from cutlass import Float32, BFloat16, Int32, Boolean, const_expr
+from cutlass.utils import LayoutEnum
+from cutlass.utils.tmem_allocator import find_tmem_tensor_col_offset
+import cuda.bindings.driver as cuda
+import cutlass.torch as ct
+
+
+class TmemCol5:
+    """QK + softmax packing + PV with V=I to isolate PV MMA correctness.
+    Output should be P = S.to(BF16), i.e. (Q@K^T).bfloat16()
+    With V=I, O = P @ I = P.
+    But V is (K=128, N=128) in the MMA. We need a 128x128 identity in MN-major.
+    Output tensor is (128, 128).
+    """
+    def __init__(self, mma_tiler_mn):
+        self.acc_dtype = Float32; self.qk_acc_dtype = Float32
+        self.q_dtype = BFloat16; self.o_dtype = BFloat16; self.c_dtype = BFloat16
+        self.mma_tiler_mn = mma_tiler_mn; self.mma_tiler = (*mma_tiler_mn, 1)
+        self.use_2cta_instrs = False  # needed by epilogue_tma_store
+        self.epilog_sync_bar_id = 1  # needed by epilogue_tma_store
+        self.cluster_shape_mn = (1, 1)
+        self.cta_group = tcgen05.CtaGroup.ONE
+        self.epilogue_warp_id = (0, 1, 2, 3)
+        self.mma_warp_id = 4; self.tma_warp_id = 5
+        self.threads_per_cta = 192
+        self.num_c_stage = 2
+
+    def _setup(self, qk_mma, pv_mma):
+        qk_inst_k = int(cute.size(qk_mma.shape_mnk, mode=[2]))
+        self.qk_mma_tiler = (*self.mma_tiler_mn, qk_inst_k * 4)
+        # PV with V=I: output is (128, 128), same as QK
+        self.pv_mma_tiler = (self.qk_mma_tiler[0], self.qk_mma_tiler[1], self.qk_mma_tiler[1])
+        # pv_mma_tiler = (128, 128, 128) since V is 128x128
+        self.mma_tiler = self.qk_mma_tiler
+
+        self.cluster_layout_vmnk = cute.tiled_divide(cute.make_layout((1,1,1)), (qk_mma.thr_id.shape,))
+        self.cta_tile_shape_mnk = (
+            self.qk_mma_tiler[0] // cute.size(qk_mma.thr_id.shape),
+            self.qk_mma_tiler[1], self.qk_mma_tiler[2])
+        self.c_layout = LayoutEnum.ROW_MAJOR
+        self.epi_tile = utils.sm100.compute_epilogue_tile_shape(
+            self.cta_tile_shape_mnk, False, self.c_layout, self.o_dtype)
+        self.num_ab_stage = 1; self.num_acc_stage = 1
+
+        self.a_smem_s = utils.sm100.make_smem_layout_a(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.b_smem_s = utils.sm100.make_smem_layout_b(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.v_smem_s = utils.sm100.make_smem_layout_b(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.p_tmem_s = utils.sm100.make_smem_layout_a(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.c_smem_s = utils.sm100.make_smem_layout_epi(self.o_dtype, self.c_layout, self.epi_tile, 2)
+
+        qk_thr = qk_mma.get_slice(0)
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        s_cols = find_tmem_tensor_col_offset(tStS)
+        pv_thr = pv_mma.get_slice(0)
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        o_cols = find_tmem_tensor_col_offset(tOtO)
+
+        self.tilePlikeFP32 = self.qk_mma_tiler[1] // Float32.width * self.o_dtype.width
+        self.tmem_s0_offset = 0
+        self.tmem_p0_offset = 32
+        self.tmem_o0_offset = s_cols
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        print(f"V: smem_size={int(cute.size(v_smem))} O: tOtO={int(cute.size(tOtO))} o_cols={o_cols} cta_tile={self.cta_tile_shape_mnk} epi_tile={self.epi_tile}")
+        tP = cute.make_tensor(tStS.iterator, self.p_tmem_s.outer)
+        tOrP2 = pv_mma.get_slice(0).make_fragment_A(tP)
+        tOrP2_s = tOrP2[(None, None, None, 0)]
+        sz0 = int(cute.size(tOrP2_s, mode=[0]))
+        sz1 = int(cute.size(tOrP2_s, mode=[1]))
+        sz2 = int(cute.size(tOrP2_s, mode=[2])) if cute.rank(tOrP2_s.layout) >= 3 else 0
+        print(sz0, sz1, sz2, int(cute.size(tStS)), int(cute.size(tP)), int(cute.size(tOrP2_s)))
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+        self.num_tmem_alloc_cols = utils.get_num_tmem_alloc_cols([tCtS_fake, tCtO_fake], arch="sm_100")
+
+        # ⛔⛔⛔ CRITICAL: num_tma_load_bytes MUST include ALL TMA-loaded tensors (Q + K + V). Missing V → DEADLOCK. See FOOTGUN #0 in README.
+        a_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        b_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        self.num_tma_load_bytes = (
+            cute.size_in_bytes(self.q_dtype, a_smem) + cute.size_in_bytes(self.q_dtype, b_smem) +
+            cute.size_in_bytes(self.q_dtype, v_smem)
+        ) * cute.size(qk_mma.thr_id.shape)
+
+    @cute.jit
+    def __call__(self, q, k, v, c, stream):
+        self.q_dtype = q.element_type; self.o_dtype = c.element_type; self.c_dtype = self.o_dtype
+        self.a_major = LayoutEnum.from_tensor(q).mma_major_mode()
+        self.b_major = LayoutEnum.from_tensor(k).mma_major_mode()
+        self.v_major = LayoutEnum.from_tensor(v).mma_major_mode()
+        self.c_layout = LayoutEnum.from_tensor(c)
+
+        qk_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, self.a_major, self.b_major,
+            self.qk_acc_dtype, self.cta_group, self.mma_tiler_mn, tcgen05.OperandSource.SMEM)
+        # PV with 128x128 output (V=I)
+        pv_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, cute.nvgpu.OperandMajorMode.K, self.v_major,
+            self.qk_acc_dtype, self.cta_group, self.mma_tiler_mn, tcgen05.OperandSource.TMEM)
+        self._setup(qk_mma, pv_mma)
+
+        q_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        k_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        tma_q, tma_tq = cute.nvgpu.make_tiled_tma_atom_A(
+            utils.sm100.cluster_shape_to_tma_atom_A(self.cluster_shape_mn, qk_mma.thr_id),
+            q, q_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_k, tma_tk = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, qk_mma.thr_id),
+            k, k_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_v, tma_tv = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, pv_mma.thr_id),
+            v, v_smem, self.pv_mma_tiler, pv_mma, self.cluster_layout_vmnk.shape)
+        epi_smem = cute.select(self.c_smem_s, mode=[0, 1])
+        tma_c, tma_tc = cpasync.make_tiled_tma_atom(cpasync.CopyBulkTensorTileS2GOp(), c, epi_smem, self.epi_tile)
+
+        self._kernel(qk_mma, pv_mma, tma_q, tma_tq, tma_k, tma_tk, tma_v, tma_tv,
+            tma_c, tma_tc, self.cluster_layout_vmnk,
+            self.a_smem_s, self.b_smem_s, self.v_smem_s, self.p_tmem_s, self.c_smem_s, self.epi_tile
+        ).launch(grid=(1,1,1), block=[self.threads_per_cta,1,1], stream=stream)
+
+    @cute.kernel
+    def _kernel(self, qk_mma, pv_mma, tma_q, mQ, tma_k, mK, tma_v, mV,
+                tma_c, mC, cl_vmnk, a_smem_s, b_smem_s, v_smem_s, p_tmem_s, c_smem_s, epi_tile):
+        warp_idx = cute.arch.make_warp_uniform(cute.arch.warp_idx())
+        tidx, _, _ = cute.arch.thread_idx()
+        use_2cta = cute.size(qk_mma.thr_id.shape) == 2
+
+        if warp_idx == self.tma_warp_id:
+            cpasync.prefetch_descriptor(tma_q); cpasync.prefetch_descriptor(tma_k)
+            cpasync.prefetch_descriptor(tma_v); cpasync.prefetch_descriptor(tma_c)
+
+        @cute.struct
+        class SS:
+            ab_bar: cute.struct.MemRange[cutlass.Int64, self.num_ab_stage * 2]
+            mma_si_bar: cute.struct.MemRange[cutlass.Int64, 2]
+            acc_bar: cute.struct.MemRange[cutlass.Int64, self.num_acc_stage * 2]
+            tmem_dealloc: cutlass.Int64
+            holding: cutlass.Int32
+
+        smem = utils.SmemAllocator(); st = smem.allocate(SS)
+
+        ab_p, ab_c = pipeline.PipelineTmaUmma.create(
+            barrier_storage=st.ab_bar.data_ptr(), num_stages=self.num_ab_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 1),
+            tx_count=self.num_tma_load_bytes, cta_layout_vmnk=cl_vmnk, defer_sync=True
+        ).make_participants()
+
+        mma_si_prod, mma_si_cons = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.mma_si_bar.data_ptr(), num_stages=1,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id)),
+        ).make_participants()
+
+        acc_pipe = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.acc_bar.data_ptr(), num_stages=self.num_acc_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(
+                pipeline.Agent.Thread, len(self.epilogue_warp_id) * (2 if use_2cta else 1)),
+            cta_layout_vmnk=cl_vmnk, defer_sync=True)
+
+        tmem_bar = pipeline.NamedBarrier(barrier_id=2,
+            num_threads=32 * len((self.mma_warp_id, *self.epilogue_warp_id)))
+        tmem = utils.TmemAllocator(st.holding.ptr, barrier_for_retrieve=tmem_bar,
+            allocator_warp_id=self.epilogue_warp_id[0], is_two_cta=use_2cta,
+            two_cta_tmem_dealloc_mbar_ptr=st.tmem_dealloc.ptr)
+
+        pipeline.pipeline_init_arrive(cluster_shape_mn=cl_vmnk, is_relaxed=True)
+
+        sQ = smem.allocate_tensor(element_type=self.q_dtype, layout=a_smem_s.outer, byte_alignment=128, swizzle=a_smem_s.inner)
+        sK = smem.allocate_tensor(element_type=self.q_dtype, layout=b_smem_s.outer, byte_alignment=128, swizzle=b_smem_s.inner)
+        sV = smem.allocate_tensor(element_type=self.q_dtype, layout=v_smem_s.outer, byte_alignment=128, swizzle=v_smem_s.inner)
+        sC = smem.allocate_tensor(element_type=self.o_dtype, layout=c_smem_s.outer, byte_alignment=128, swizzle=c_smem_s.inner)
+
+        gQ = cute.local_tile(mQ, cute.slice_(self.qk_mma_tiler, (None,0,None)), (None,None,None))
+        gK = cute.local_tile(mK, cute.slice_(self.qk_mma_tiler, (0,None,None)), (None,None,None))
+        gC = cute.local_tile(mC, cute.slice_(self.qk_mma_tiler, (None,None,0)), (None,None,None))
+        k_cnt = cute.size(gQ, mode=[3])
+
+        qk_thr = qk_mma.get_slice(0)
+        pv_thr = pv_mma.get_slice(0)
+        tCgQ = qk_thr.partition_A(gQ); tCgK = qk_thr.partition_B(gK); tCgC = qk_thr.partition_C(gC)
+        a_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,0,None,0)).shape)
+        tAsQ, tAgQ = cpasync.tma_partition(tma_q, 0, a_lay, cute.group_modes(sQ,0,3), cute.group_modes(tCgQ,0,3))
+        b_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,None,0,0)).shape)
+        tBsK, tBgK = cpasync.tma_partition(tma_k, 0, b_lay, cute.group_modes(sK,0,3), cute.group_modes(tCgK,0,3))
+        tAgQ = tAgQ[(None,0,None,0)]; tBgK = tBgK[(None,0,None,0)]
+
+        gV = cute.local_tile(mV, cute.slice_(self.pv_mma_tiler, (0,None,None)), (None,None,None))
+        tCgV = pv_thr.partition_B(gV)
+        tVsV, tVgV = cpasync.tma_partition(tma_v, 0, b_lay, cute.group_modes(sV,0,3), cute.group_modes(tCgV,0,3))
+        tVgV = tVgV[(None,0,None,0)]
+
+        tCrQ = qk_mma.make_fragment_A(sQ); tCrK = qk_mma.make_fragment_B(sK)
+        tCrV = pv_mma.make_fragment_B(sV)
+
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        tStS0 = cute.make_tensor(tStS.iterator + self.tmem_s0_offset, tStS.layout)
+
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        tOtO0 = cute.make_tensor(tOtO.iterator + self.tmem_o0_offset, tOtO.layout)
+
+        tP = cute.make_tensor(tStS.iterator, p_tmem_s.outer)
+        tOrP_base = pv_thr.make_fragment_A(tP)
+        tOrP = tOrP_base[(None, None, None, 0)]
+        tOrP0 = cute.make_tensor(
+            tOrP.iterator + self.qk_acc_dtype.width // self.q_dtype.width * self.tmem_p0_offset,
+            tOrP.layout)
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+
+        pipeline.pipeline_init_wait(cluster_shape_mn=cl_vmnk)
+
+        # ═══ TMA LOAD WARP ═══
+        if warp_idx == self.tma_warp_id:
+            ab_p.reset(); peek = ab_p.try_acquire()
+            for kt in cutlass.range(k_cnt, unroll=1):
+                h = ab_p.acquire_and_advance(peek)
+                cute.copy(tma_q, tAgQ[(None,h.count)], tAsQ[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_k, tBgK[(None,h.count)], tBsK[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_v, tVgV[(None,h.count)], tVsV[(None,h.index)], tma_bar_ptr=h.barrier)
+                peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_p.try_acquire()
+            ab_p.tail()
+
+        # ═══ MMA WARP ═══
+        if warp_idx == self.mma_warp_id:
+            tmem.wait_for_alloc()
+            ab_c.reset(); peek = ab_c.try_wait()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            acc_prod_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Producer, self.num_acc_stage)
+            acc_pipe.producer_acquire(acc_prod_st)
+            qk_mma.set(tcgen05.Field.ACCUMULATE, False)
+            for kt in range(k_cnt):
+                h = ab_c.wait_and_advance(peek)
+                nblk = cute.size(tCrQ, mode=[2])
+                for kb in cutlass.range(nblk, unroll_full=True):
+                    cute.gemm(qk_mma, tStS0, tCrQ[(None,None,kb,h.index)], tCrK[(None,None,kb,h.index)], tStS0)
+                    qk_mma.set(tcgen05.Field.ACCUMULATE, True)
+                h.release(); peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_c.try_wait()
+
+            cute.arch.fence_view_async_tmem_store()
+            s0_handle.commit()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            # PV MMA: P @ V where V=I → O = P
+            pv_mma.set(tcgen05.Field.ACCUMULATE, False)
+            tCrV_s = tCrV[(None, None, None, 0)]
+            nblk_pv = cute.size(tOrP0, mode=[2])
+            for kb in cutlass.range(nblk_pv, unroll_full=True):
+                cute.gemm(pv_mma, tOtO0, tOrP0[(None,None,kb)], tCrV_s[(None,None,kb)], tOtO0)
+                pv_mma.set(tcgen05.Field.ACCUMULATE, True)
+
+            acc_pipe.producer_commit(acc_prod_st)
+            acc_prod_st.advance()
+            acc_pipe.producer_tail(acc_prod_st)
+
+        # ═══ EPILOGUE WARPS ═══
+        if warp_idx < self.mma_warp_id:
+            tmem.allocate(self.num_tmem_alloc_cols)
+            tmem.wait_for_alloc()
+            tmem_ptr = tmem.retrieve_ptr(self.qk_acc_dtype)
+            sfw_idx = tidx % (32 * len(self.epilogue_warp_id))
+
+            tmem_load_atom = cute.make_copy_atom(
+                tcgen05.copy.Ld32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_load = tcgen05.make_tmem_copy(tmem_load_atom, tStS0)
+            thr_load = tiled_tmem_load.get_slice(sfw_idx)
+            tTMEM_LOADtS = thr_load.partition_S(tStS0)
+            cS = cute.make_identity_tensor((self.qk_mma_tiler[0], self.qk_mma_tiler[1]))
+            tScS = qk_thr.partition_C(cS)
+            tTMEM_LOADcS = thr_load.partition_D(tScS)
+
+            tStS_P_layout = cute.composition(tStS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tStS_P = cute.make_tensor(tStS.iterator + self.tmem_p0_offset, tStS_P_layout)
+            tmem_store_atom = cute.make_copy_atom(
+                tcgen05.copy.St32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_store = tcgen05.make_tmem_copy(tmem_store_atom, tStS_P)
+            thr_store = tiled_tmem_store.get_slice(sfw_idx)
+            tTMEM_STOREtS_x4 = thr_store.partition_D(tStS_P)
+            tScS_P_layout = cute.composition(tScS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tScS_P = cute.make_tensor(tScS.iterator, tScS_P_layout)
+            tTMEM_STOREcS = thr_store.partition_S(tScS_P)
+
+            si_handle = mma_si_cons.wait_and_advance()
+
+            tTMEM_LOADrS = cute.make_rmem_tensor(tTMEM_LOADcS.shape, self.qk_acc_dtype)
+            cute.copy(tiled_tmem_load, tTMEM_LOADtS, tTMEM_LOADrS)
+            tTMEM_STORErS_x4 = cute.make_rmem_tensor(tTMEM_STOREcS.shape, self.qk_acc_dtype)
+            tTMEM_STORErS_x4_e = cute.make_tensor(
+                cute.recast_ptr(tTMEM_STORErS_x4.iterator, dtype=self.q_dtype),
+                tTMEM_LOADrS.layout)
+            frg_cnt = 4
+            frg_tile = cute.size(tTMEM_LOADrS) // frg_cnt
+            tTMEM_LOADrS_frg = cute.logical_divide(tTMEM_LOADrS, cute.make_layout(frg_tile))
+            tTMEM_STORErS_x4_e_frg = cute.logical_divide(
+                tTMEM_STORErS_x4_e, cute.make_layout(frg_tile))
+            for j in range(frg_cnt):
+                s_vec = tTMEM_LOADrS_frg[None, j].load()
+                tTMEM_STORErS_x4_e_frg[None, j].store(s_vec.to(self.q_dtype))
+            cute.copy(tiled_tmem_store, tTMEM_STORErS_x4, tTMEM_STOREtS_x4)
+            cute.arch.fence_view_async_tmem_store()
+            si_handle.release()
+
+            # Output epilogue
+            tCtO_base = cute.make_tensor(tmem_ptr + self.tmem_o0_offset, tCtO_fake.layout)
+            acc_cons_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Consumer, self.num_acc_stage)
+            c_grp = pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id))
+            c_pipe = pipeline.PipelineTmaStore.create(num_stages=self.num_c_stage, producer_group=c_grp)
+            acc_cons_st = utils.gemm.sm100.epilogue_tma_store(
+                self, tidx, warp_idx, tma_c, tCtO_base, sC, tCgC,
+                epi_tile, 0, const_expr(lambda x: x), (0,0,0), acc_cons_st, acc_pipe, c_pipe)
+            c_pipe.producer_tail()
+            tmem.relinquish_alloc_permit()
+            tmem.free(tmem_ptr)
+
+
+def test():
+    torch.manual_seed(42)
+    m, n, head_dim = 128, 128, 64
+    q = torch.randn(m, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    k = torch.randn(n, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    # V = identity (128x128) in MN-major: (128,128) with strides (1,128)
+    v = torch.eye(128, dtype=torch.bfloat16, device='cuda')
+    # MN-major: (128,128) with strides (1,128) — row is fast dim
+    v = v.as_strided((128, 128), (1, 128)).unsqueeze(-1)
+    c = torch.zeros(m, n, 1, dtype=torch.bfloat16, device='cuda')
+
+    qf = q[:,:,0].float(); kf = k[:,:,0].float()
+    # With V=I and identity softmax: O = (Q@K^T).bf16() @ I = (Q@K^T).bf16()
+    ref = (qf @ kf.T).bfloat16().float()
+
+    mQ = ct.from_dlpack(q).mark_layout_dynamic(leading_dim=ct.get_leading_dim(q))
+    mK = ct.from_dlpack(k).mark_layout_dynamic(leading_dim=ct.get_leading_dim(k))
+    mV = ct.from_dlpack(v).mark_layout_dynamic(leading_dim=ct.get_leading_dim(v))
+    mC = ct.from_dlpack(c).mark_layout_dynamic(leading_dim=ct.get_leading_dim(c))
+    stream = cuda.CUstream(torch.cuda.current_stream().cuda_stream)
+    kernel = TmemCol5(mma_tiler_mn=(128, 128))
+    print('Compiling...', flush=True)
+    compiled = cute.compile(kernel, mQ, mK, mV, mC, stream)
+    print('Running...', flush=True)
+    compiled(mQ, mK, mV, mC, stream)
+    torch.cuda.synchronize()
+    out = c[:,:,0].float()
+    cos = torch.nn.functional.cosine_similarity(out.flatten().unsqueeze(0), ref.flatten().unsqueeze(0)).item()
+    print('Col5: cosine {:.6f} {}'.format(cos, 'PASS' if cos >= 0.99 else 'FAIL'))
+
+
+if __name__ == '__main__':
+    test()

tests/test_tmem_col5_16.py

@@ -0,0 +1,397 @@
+"""
+Minimal PV-only test: Load P from GMEM to TMEM via QK-style MMA, then PV from TMEM.
+Step 1: QK MMA writes FP32 S to TMEM (we know this works)
+Step 2: Softmax packing writes BF16 P to TMEM (test this)
+Step 3: PV MMA reads BF16 P from TMEM and V from SMEM, produces O
+
+But to isolate the bug, let me test just the PV MMA in isolation.
+I'll write known BF16 values to TMEM using the softmax packing path,
+then immediately read them back using the PV A-fragment path,
+and compare.
+
+Actually, the simplest isolation test:
+1. Do QK MMA to get S in TMEM (cosine 0.999999 verified)
+2. Do softmax packing: S → P in TMEM (at offset 32)
+3. Skip PV entirely — read P from TMEM using the C-fragment composition LOAD path
+4. Output P to GMEM and compare against S.to(BF16)
+
+This tests whether the softmax packing writes P correctly to the same TMEM
+that the PV would read from.
+
+But we can't easily read P from TMEM using the standard epilogue path
+because the epilogue expects FP32 accumulator data.
+
+Alternative: Use the PV MMA with V=I (identity). If P is correct,
+then P @ I = P. But V needs to be MN-major and (128, 128), not (128, 64).
+The output would be (128, 128) which doesn't match our (128, 64) c tensor.
+
+Let me use V that selects the first 64 columns: V[k, n] = delta(k, n) for k in [0,63].
+This gives P @ V = P[:, :64], and the output is (128, 64).
+But V is (128, 128) in the MMA K,N dims. V[k, n] for k in [0,127], n in [0,63].
+Hmm, this is getting complicated. Let me just do the identity approach with a (128, 128) output.
+"""
+import torch, cutlass, cutlass.cute as cute, cutlass.utils as utils, cutlass.pipeline as pipeline
+from cutlass.cute.nvgpu import cpasync, tcgen05
+from cutlass import Float32, BFloat16, Int32, Boolean, const_expr
+from cutlass.utils import LayoutEnum
+from cutlass.utils.tmem_allocator import find_tmem_tensor_col_offset
+import cuda.bindings.driver as cuda
+import cutlass.torch as ct
+
+
+class TmemCol5_16_32:
+    """QK + softmax packing + PV with V=I to isolate PV MMA correctness.
+    Output should be P = S.to(BF16), i.e. (Q@K^T).bfloat16()
+    With V=I, O = P @ I = P.
+    But V is (K=128, N=128) in the MMA. We need a 128x128 identity in MN-major.
+    Output tensor is (128, 128).
+    """
+    def __init__(self, mma_tiler_mn):
+        self.acc_dtype = Float32; self.qk_acc_dtype = Float32
+        self.q_dtype = BFloat16; self.o_dtype = BFloat16; self.c_dtype = BFloat16
+        self.mma_tiler_mn = mma_tiler_mn; self.mma_tiler = (*mma_tiler_mn, 1)
+        self.use_2cta_instrs = False  # needed by epilogue_tma_store
+        self.epilog_sync_bar_id = 1  # needed by epilogue_tma_store
+        self.cluster_shape_mn = (1, 1)
+        self.cta_group = tcgen05.CtaGroup.ONE
+        self.epilogue_warp_id = (0, 1, 2, 3)
+        self.mma_warp_id = 4; self.tma_warp_id = 5
+        self.threads_per_cta = 192
+        self.num_c_stage = 2
+
+    def _setup(self, qk_mma, pv_mma):
+        qk_inst_k = int(cute.size(qk_mma.shape_mnk, mode=[2]))
+        self.qk_mma_tiler = (*self.mma_tiler_mn, qk_inst_k * 4)
+        # PV with V=I: output is (128, 128), same as QK
+        self.pv_mma_tiler = (self.qk_mma_tiler[0], qk_inst_k, self.qk_mma_tiler[1])
+        # pv_mma_tiler = (128, 128, 128) since V is 128x128
+        self.mma_tiler = self.qk_mma_tiler
+
+        self.cluster_layout_vmnk = cute.tiled_divide(cute.make_layout((1,1,1)), (qk_mma.thr_id.shape,))
+        self.cta_tile_shape_mnk = (
+            self.qk_mma_tiler[0] // cute.size(qk_mma.thr_id.shape),
+            self.qk_mma_tiler[1], self.qk_mma_tiler[2])
+        self.c_layout = LayoutEnum.ROW_MAJOR
+        self.epi_tile = utils.sm100.compute_epilogue_tile_shape(
+            (self.pv_mma_tiler[0], self.pv_mma_tiler[1], self.pv_mma_tiler[2]), False, self.c_layout, self.o_dtype)
+        self.num_ab_stage = 1; self.num_acc_stage = 1
+
+        self.a_smem_s = utils.sm100.make_smem_layout_a(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.b_smem_s = utils.sm100.make_smem_layout_b(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.v_smem_s = utils.sm100.make_smem_layout_b(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.p_tmem_s = utils.sm100.make_smem_layout_a(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.c_smem_s = utils.sm100.make_smem_layout_epi(self.o_dtype, self.c_layout, self.epi_tile, 2)
+
+        qk_thr = qk_mma.get_slice(0)
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        s_cols = find_tmem_tensor_col_offset(tStS)
+        pv_thr = pv_mma.get_slice(0)
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        o_cols = find_tmem_tensor_col_offset(tOtO)
+
+        self.tilePlikeFP32 = self.qk_mma_tiler[1] // Float32.width * self.o_dtype.width
+        self.tmem_s0_offset = 0
+        self.tmem_p0_offset = 32
+        self.tmem_o0_offset = s_cols
+        # V SMEM layout
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        print(f"V: smem_size={int(cute.size(v_smem))}")
+        # O C-fragment
+        print(f"O: tOtO_size={int(cute.size(tOtO))} o_cols={o_cols}")
+        # PV tiler
+        print(f"PV: mma_tiler={self.pv_mma_tiler} epi_tile={self.epi_tile} cta_tile={self.cta_tile_shape_mnk}")
+        tP = cute.make_tensor(tStS.iterator, self.p_tmem_s.outer)
+        tOrP2 = pv_mma.get_slice(0).make_fragment_A(tP)
+        tOrP2_s = tOrP2[(None, None, None, 0)]
+        sz0 = int(cute.size(tOrP2_s, mode=[0]))
+        sz1 = int(cute.size(tOrP2_s, mode=[1]))
+        sz2 = int(cute.size(tOrP2_s, mode=[2])) if cute.rank(tOrP2_s.layout) >= 3 else 0
+        print(sz0, sz1, sz2, int(cute.size(tStS)), int(cute.size(tP)), int(cute.size(tOrP2_s)))
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+        self.num_tmem_alloc_cols = utils.get_num_tmem_alloc_cols([tCtS_fake, tCtO_fake], arch="sm_100")
+
+        # ⛔⛔⛔ CRITICAL: num_tma_load_bytes MUST include ALL TMA-loaded tensors (Q + K + V). Missing V → DEADLOCK. See FOOTGUN #0 in README.
+        a_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        b_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        self.num_tma_load_bytes = (
+            cute.size_in_bytes(self.q_dtype, a_smem) + cute.size_in_bytes(self.q_dtype, b_smem) +
+            cute.size_in_bytes(self.q_dtype, v_smem)
+        ) * cute.size(qk_mma.thr_id.shape)
+
+    @cute.jit
+    def __call__(self, q, k, v, c, stream):
+        self.q_dtype = q.element_type; self.o_dtype = c.element_type; self.c_dtype = self.o_dtype
+        self.a_major = LayoutEnum.from_tensor(q).mma_major_mode()
+        self.b_major = LayoutEnum.from_tensor(k).mma_major_mode()
+        self.v_major = LayoutEnum.from_tensor(v).mma_major_mode()
+        self.c_layout = LayoutEnum.from_tensor(c)
+
+        qk_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, self.a_major, self.b_major,
+            self.qk_acc_dtype, self.cta_group, self.mma_tiler_mn, tcgen05.OperandSource.SMEM)
+        # PV with 128x128 output (V=I)
+        pv_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, cute.nvgpu.OperandMajorMode.K, self.v_major,
+            self.qk_acc_dtype, self.cta_group, (128, 16), tcgen05.OperandSource.TMEM)
+        self._setup(qk_mma, pv_mma)
+
+        q_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        k_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        tma_q, tma_tq = cute.nvgpu.make_tiled_tma_atom_A(
+            utils.sm100.cluster_shape_to_tma_atom_A(self.cluster_shape_mn, qk_mma.thr_id),
+            q, q_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_k, tma_tk = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, qk_mma.thr_id),
+            k, k_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_v, tma_tv = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, pv_mma.thr_id),
+            v, v_smem, self.pv_mma_tiler, pv_mma, self.cluster_layout_vmnk.shape)
+        epi_smem = cute.select(self.c_smem_s, mode=[0, 1])
+        tma_c, tma_tc = cpasync.make_tiled_tma_atom(cpasync.CopyBulkTensorTileS2GOp(), c, epi_smem, self.epi_tile)
+
+        self._kernel(qk_mma, pv_mma, tma_q, tma_tq, tma_k, tma_tk, tma_v, tma_tv,
+            tma_c, tma_tc, self.cluster_layout_vmnk,
+            self.a_smem_s, self.b_smem_s, self.v_smem_s, self.p_tmem_s, self.c_smem_s, self.epi_tile
+        ).launch(grid=(1,1,1), block=[self.threads_per_cta,1,1], stream=stream)
+
+    @cute.kernel
+    def _kernel(self, qk_mma, pv_mma, tma_q, mQ, tma_k, mK, tma_v, mV,
+                tma_c, mC, cl_vmnk, a_smem_s, b_smem_s, v_smem_s, p_tmem_s, c_smem_s, epi_tile):
+        warp_idx = cute.arch.make_warp_uniform(cute.arch.warp_idx())
+        tidx, _, _ = cute.arch.thread_idx()
+        use_2cta = cute.size(qk_mma.thr_id.shape) == 2
+
+        if warp_idx == self.tma_warp_id:
+            cpasync.prefetch_descriptor(tma_q); cpasync.prefetch_descriptor(tma_k)
+            cpasync.prefetch_descriptor(tma_v); cpasync.prefetch_descriptor(tma_c)
+
+        @cute.struct
+        class SS:
+            ab_bar: cute.struct.MemRange[cutlass.Int64, self.num_ab_stage * 2]
+            mma_si_bar: cute.struct.MemRange[cutlass.Int64, 2]
+            acc_bar: cute.struct.MemRange[cutlass.Int64, self.num_acc_stage * 2]
+            tmem_dealloc: cutlass.Int64
+            holding: cutlass.Int32
+
+        smem = utils.SmemAllocator(); st = smem.allocate(SS)
+
+        ab_p, ab_c = pipeline.PipelineTmaUmma.create(
+            barrier_storage=st.ab_bar.data_ptr(), num_stages=self.num_ab_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 1),
+            tx_count=self.num_tma_load_bytes, cta_layout_vmnk=cl_vmnk, defer_sync=True
+        ).make_participants()
+
+        mma_si_prod, mma_si_cons = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.mma_si_bar.data_ptr(), num_stages=1,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id)),
+        ).make_participants()
+
+        acc_pipe = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.acc_bar.data_ptr(), num_stages=self.num_acc_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(
+                pipeline.Agent.Thread, len(self.epilogue_warp_id) * (2 if use_2cta else 1)),
+            cta_layout_vmnk=cl_vmnk, defer_sync=True)
+
+        tmem_bar = pipeline.NamedBarrier(barrier_id=2,
+            num_threads=32 * len((self.mma_warp_id, *self.epilogue_warp_id)))
+        tmem = utils.TmemAllocator(st.holding.ptr, barrier_for_retrieve=tmem_bar,
+            allocator_warp_id=self.epilogue_warp_id[0], is_two_cta=use_2cta,
+            two_cta_tmem_dealloc_mbar_ptr=st.tmem_dealloc.ptr)
+
+        pipeline.pipeline_init_arrive(cluster_shape_mn=cl_vmnk, is_relaxed=True)
+
+        sQ = smem.allocate_tensor(element_type=self.q_dtype, layout=a_smem_s.outer, byte_alignment=128, swizzle=a_smem_s.inner)
+        sK = smem.allocate_tensor(element_type=self.q_dtype, layout=b_smem_s.outer, byte_alignment=128, swizzle=b_smem_s.inner)
+        sV = smem.allocate_tensor(element_type=self.q_dtype, layout=v_smem_s.outer, byte_alignment=128, swizzle=v_smem_s.inner)
+        sC = smem.allocate_tensor(element_type=self.o_dtype, layout=c_smem_s.outer, byte_alignment=128, swizzle=c_smem_s.inner)
+
+        gQ = cute.local_tile(mQ, cute.slice_(self.qk_mma_tiler, (None,0,None)), (None,None,None))
+        gK = cute.local_tile(mK, cute.slice_(self.qk_mma_tiler, (0,None,None)), (None,None,None))
+        gC = cute.local_tile(mC, cute.slice_(self.qk_mma_tiler, (None,None,0)), (None,None,None))
+        k_cnt = cute.size(gQ, mode=[3])
+
+        qk_thr = qk_mma.get_slice(0)
+        pv_thr = pv_mma.get_slice(0)
+        tCgQ = qk_thr.partition_A(gQ); tCgK = qk_thr.partition_B(gK); tCgC = qk_thr.partition_C(gC)
+        a_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,0,None,0)).shape)
+        tAsQ, tAgQ = cpasync.tma_partition(tma_q, 0, a_lay, cute.group_modes(sQ,0,3), cute.group_modes(tCgQ,0,3))
+        b_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,None,0,0)).shape)
+        tBsK, tBgK = cpasync.tma_partition(tma_k, 0, b_lay, cute.group_modes(sK,0,3), cute.group_modes(tCgK,0,3))
+        tAgQ = tAgQ[(None,0,None,0)]; tBgK = tBgK[(None,0,None,0)]
+
+        gV = cute.local_tile(mV, cute.slice_(self.pv_mma_tiler, (0,None,None)), (None,None,None))
+        tCgV = pv_thr.partition_B(gV)
+        tVsV, tVgV = cpasync.tma_partition(tma_v, 0, b_lay, cute.group_modes(sV,0,3), cute.group_modes(tCgV,0,3))
+        tVgV = tVgV[(None,0,None,0)]
+
+        tCrQ = qk_mma.make_fragment_A(sQ); tCrK = qk_mma.make_fragment_B(sK)
+        tCrV = pv_mma.make_fragment_B(sV)
+
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        tStS0 = cute.make_tensor(tStS.iterator + self.tmem_s0_offset, tStS.layout)
+
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        tOtO0 = cute.make_tensor(tOtO.iterator + self.tmem_o0_offset, tOtO.layout)
+
+        tP = cute.make_tensor(tStS.iterator, p_tmem_s.outer)
+        tOrP_base = pv_thr.make_fragment_A(tP)
+        tOrP = tOrP_base[(None, None, None, 0)]
+        tOrP0 = cute.make_tensor(
+            tOrP.iterator + self.qk_acc_dtype.width // self.q_dtype.width * self.tmem_p0_offset,
+            tOrP.layout)
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+
+        pipeline.pipeline_init_wait(cluster_shape_mn=cl_vmnk)
+
+        # ═══ TMA LOAD WARP ═══
+        if warp_idx == self.tma_warp_id:
+            ab_p.reset(); peek = ab_p.try_acquire()
+            for kt in cutlass.range(k_cnt, unroll=1):
+                h = ab_p.acquire_and_advance(peek)
+                cute.copy(tma_q, tAgQ[(None,h.count)], tAsQ[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_k, tBgK[(None,h.count)], tBsK[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_v, tVgV[(None,h.count)], tVsV[(None,h.index)], tma_bar_ptr=h.barrier)
+                peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_p.try_acquire()
+            ab_p.tail()
+
+        # ═══ MMA WARP ═══
+        if warp_idx == self.mma_warp_id:
+            tmem.wait_for_alloc()
+            ab_c.reset(); peek = ab_c.try_wait()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            acc_prod_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Producer, self.num_acc_stage)
+            acc_pipe.producer_acquire(acc_prod_st)
+            qk_mma.set(tcgen05.Field.ACCUMULATE, False)
+            for kt in range(k_cnt):
+                h = ab_c.wait_and_advance(peek)
+                nblk = cute.size(tCrQ, mode=[2])
+                for kb in cutlass.range(nblk, unroll_full=True):
+                    cute.gemm(qk_mma, tStS0, tCrQ[(None,None,kb,h.index)], tCrK[(None,None,kb,h.index)], tStS0)
+                    qk_mma.set(tcgen05.Field.ACCUMULATE, True)
+                h.release(); peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_c.try_wait()
+
+            cute.arch.fence_view_async_tmem_store()
+            s0_handle.commit()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            # PV MMA: P @ V where V=I → O = P
+            pv_mma.set(tcgen05.Field.ACCUMULATE, False)
+            tCrV_s = tCrV[(None, None, None, 0)]
+            nblk_pv = cute.size(tOrP0, mode=[2])
+            for kb in cutlass.range(nblk_pv, unroll_full=True):
+                cute.gemm(pv_mma, tOtO0, tOrP0[(None,None,kb)], tCrV_s[(None,None,kb)], tOtO0)
+                pv_mma.set(tcgen05.Field.ACCUMULATE, True)
+
+            acc_pipe.producer_commit(acc_prod_st)
+            acc_prod_st.advance()
+            acc_pipe.producer_tail(acc_prod_st)
+
+        # ═══ EPILOGUE WARPS ═══
+        if warp_idx < self.mma_warp_id:
+            tmem.allocate(self.num_tmem_alloc_cols)
+            tmem.wait_for_alloc()
+            tmem_ptr = tmem.retrieve_ptr(self.qk_acc_dtype)
+            sfw_idx = tidx % (32 * len(self.epilogue_warp_id))
+
+            tmem_load_atom = cute.make_copy_atom(
+                tcgen05.copy.Ld32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_load = tcgen05.make_tmem_copy(tmem_load_atom, tStS0)
+            thr_load = tiled_tmem_load.get_slice(sfw_idx)
+            tTMEM_LOADtS = thr_load.partition_S(tStS0)
+            cS = cute.make_identity_tensor((self.qk_mma_tiler[0], self.qk_mma_tiler[1]))
+            tScS = qk_thr.partition_C(cS)
+            tTMEM_LOADcS = thr_load.partition_D(tScS)
+
+            tStS_P_layout = cute.composition(tStS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tStS_P = cute.make_tensor(tStS.iterator + self.tmem_p0_offset, tStS_P_layout)
+            tmem_store_atom = cute.make_copy_atom(
+                tcgen05.copy.St32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_store = tcgen05.make_tmem_copy(tmem_store_atom, tStS_P)
+            thr_store = tiled_tmem_store.get_slice(sfw_idx)
+            tTMEM_STOREtS_x4 = thr_store.partition_D(tStS_P)
+            tScS_P_layout = cute.composition(tScS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tScS_P = cute.make_tensor(tScS.iterator, tScS_P_layout)
+            tTMEM_STOREcS = thr_store.partition_S(tScS_P)
+
+            si_handle = mma_si_cons.wait_and_advance()
+
+            tTMEM_LOADrS = cute.make_rmem_tensor(tTMEM_LOADcS.shape, self.qk_acc_dtype)
+            cute.copy(tiled_tmem_load, tTMEM_LOADtS, tTMEM_LOADrS)
+            tTMEM_STORErS_x4 = cute.make_rmem_tensor(tTMEM_STOREcS.shape, self.qk_acc_dtype)
+            tTMEM_STORErS_x4_e = cute.make_tensor(
+                cute.recast_ptr(tTMEM_STORErS_x4.iterator, dtype=self.q_dtype),
+                tTMEM_LOADrS.layout)
+            frg_cnt = 4
+            frg_tile = cute.size(tTMEM_LOADrS) // frg_cnt
+            tTMEM_LOADrS_frg = cute.logical_divide(tTMEM_LOADrS, cute.make_layout(frg_tile))
+            tTMEM_STORErS_x4_e_frg = cute.logical_divide(
+                tTMEM_STORErS_x4_e, cute.make_layout(frg_tile))
+            for j in range(frg_cnt):
+                s_vec = tTMEM_LOADrS_frg[None, j].load()
+                tTMEM_STORErS_x4_e_frg[None, j].store(s_vec.to(self.q_dtype))
+            cute.copy(tiled_tmem_store, tTMEM_STORErS_x4, tTMEM_STOREtS_x4)
+            cute.arch.fence_view_async_tmem_store()
+            si_handle.release()
+
+            # Output epilogue
+            tCtO_base = cute.make_tensor(tmem_ptr + self.tmem_o0_offset, tCtO_fake.layout)
+            acc_cons_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Consumer, self.num_acc_stage)
+            c_grp = pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id))
+            c_pipe = pipeline.PipelineTmaStore.create(num_stages=self.num_c_stage, producer_group=c_grp)
+            acc_cons_st = utils.gemm.sm100.epilogue_tma_store(
+                self, tidx, warp_idx, tma_c, tCtO_base, sC, tCgC,
+                epi_tile, 0, const_expr(lambda x: x), (0,0,0), acc_cons_st, acc_pipe, c_pipe)
+            c_pipe.producer_tail()
+            tmem.relinquish_alloc_permit()
+            tmem.free(tmem_ptr)
+
+
+def test():
+    torch.manual_seed(42)
+    m, n, head_dim = 128, 128, 64
+    q = torch.randn(m, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    k = torch.randn(n, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    # V = identity (128x128) in MN-major: (128,128) with strides (1,128)
+    v = torch.eye(128, dtype=torch.bfloat16, device='cuda')
+    # MN-major: (128,128) with strides (1,128) — row is fast dim
+    v = v.as_strided((128, 128), (1, 128)).unsqueeze(-1)
+    c = torch.zeros(m, n, 1, dtype=torch.bfloat16, device='cuda')
+
+    qf = q[:,:,0].float(); kf = k[:,:,0].float()
+    # With V=I and identity softmax: O = (Q@K^T).bf16() @ I = (Q@K^T).bf16()
+    ref = (qf @ kf.T).bfloat16().float()
+
+    mQ = ct.from_dlpack(q).mark_layout_dynamic(leading_dim=ct.get_leading_dim(q))
+    mK = ct.from_dlpack(k).mark_layout_dynamic(leading_dim=ct.get_leading_dim(k))
+    mV = ct.from_dlpack(v).mark_layout_dynamic(leading_dim=ct.get_leading_dim(v))
+    mC = ct.from_dlpack(c).mark_layout_dynamic(leading_dim=ct.get_leading_dim(c))
+    stream = cuda.CUstream(torch.cuda.current_stream().cuda_stream)
+    kernel = TmemCol5_16_32(mma_tiler_mn=(128, 128))
+    print('Compiling...', flush=True)
+    compiled = cute.compile(kernel, mQ, mK, mV, mC, stream)
+    print('Running...', flush=True)
+    compiled(mQ, mK, mV, mC, stream)
+    torch.cuda.synchronize()
+    out = c[:,:,0].float()
+    cos = torch.nn.functional.cosine_similarity(out.flatten().unsqueeze(0), ref.flatten().unsqueeze(0)).item()
+    print('Col5(128,16): cosine {:.6f} {}'.format(cos, 'PASS' if cos >= 0.99 else 'FAIL'))
+
+
+if __name__ == '__main__':
+    test()

tests/test_tmem_col5_32.py

@@ -0,0 +1,390 @@
+"""
+Minimal PV-only test: Load P from GMEM to TMEM via QK-style MMA, then PV from TMEM.
+Step 1: QK MMA writes FP32 S to TMEM (we know this works)
+Step 2: Softmax packing writes BF16 P to TMEM (test this)
+Step 3: PV MMA reads BF16 P from TMEM and V from SMEM, produces O
+
+But to isolate the bug, let me test just the PV MMA in isolation.
+I'll write known BF16 values to TMEM using the softmax packing path,
+then immediately read them back using the PV A-fragment path,
+and compare.
+
+Actually, the simplest isolation test:
+1. Do QK MMA to get S in TMEM (cosine 0.999999 verified)
+2. Do softmax packing: S → P in TMEM (at offset 32)
+3. Skip PV entirely — read P from TMEM using the C-fragment composition LOAD path
+4. Output P to GMEM and compare against S.to(BF16)
+
+This tests whether the softmax packing writes P correctly to the same TMEM
+that the PV would read from.
+
+But we can't easily read P from TMEM using the standard epilogue path
+because the epilogue expects FP32 accumulator data.
+
+Alternative: Use the PV MMA with V=I (identity). If P is correct,
+then P @ I = P. But V needs to be MN-major and (128, 128), not (128, 64).
+The output would be (128, 128) which doesn't match our (128, 64) c tensor.
+
+Let me use V that selects the first 64 columns: V[k, n] = delta(k, n) for k in [0,63].
+This gives P @ V = P[:, :64], and the output is (128, 64).
+But V is (128, 128) in the MMA K,N dims. V[k, n] for k in [0,127], n in [0,63].
+Hmm, this is getting complicated. Let me just do the identity approach with a (128, 128) output.
+"""
+import torch, cutlass, cutlass.cute as cute, cutlass.utils as utils, cutlass.pipeline as pipeline
+from cutlass.cute.nvgpu import cpasync, tcgen05
+from cutlass import Float32, BFloat16, Int32, Boolean, const_expr
+from cutlass.utils import LayoutEnum
+from cutlass.utils.tmem_allocator import find_tmem_tensor_col_offset
+import cuda.bindings.driver as cuda
+import cutlass.torch as ct
+
+
+class TmemCol5_32:
+    """QK + softmax packing + PV with V=I to isolate PV MMA correctness.
+    Output should be P = S.to(BF16), i.e. (Q@K^T).bfloat16()
+    With V=I, O = P @ I = P.
+    But V is (K=128, N=128) in the MMA. We need a 128x128 identity in MN-major.
+    Output tensor is (128, 128).
+    """
+    def __init__(self, mma_tiler_mn):
+        self.acc_dtype = Float32; self.qk_acc_dtype = Float32
+        self.q_dtype = BFloat16; self.o_dtype = BFloat16; self.c_dtype = BFloat16
+        self.mma_tiler_mn = mma_tiler_mn; self.mma_tiler = (*mma_tiler_mn, 1)
+        self.use_2cta_instrs = False  # needed by epilogue_tma_store
+        self.epilog_sync_bar_id = 1  # needed by epilogue_tma_store
+        self.cluster_shape_mn = (1, 1)
+        self.cta_group = tcgen05.CtaGroup.ONE
+        self.epilogue_warp_id = (0, 1, 2, 3)
+        self.mma_warp_id = 4; self.tma_warp_id = 5
+        self.threads_per_cta = 192
+        self.num_c_stage = 2
+
+    def _setup(self, qk_mma, pv_mma):
+        qk_inst_k = int(cute.size(qk_mma.shape_mnk, mode=[2]))
+        self.qk_mma_tiler = (*self.mma_tiler_mn, qk_inst_k * 4)
+        # PV with V=I: output is (128, 128), same as QK
+        self.pv_mma_tiler = (self.qk_mma_tiler[0], qk_inst_k, self.qk_mma_tiler[1])
+        # pv_mma_tiler = (128, 128, 128) since V is 128x128
+        self.mma_tiler = self.qk_mma_tiler
+
+        self.cluster_layout_vmnk = cute.tiled_divide(cute.make_layout((1,1,1)), (qk_mma.thr_id.shape,))
+        self.cta_tile_shape_mnk = (
+            self.qk_mma_tiler[0] // cute.size(qk_mma.thr_id.shape),
+            self.qk_mma_tiler[1], self.qk_mma_tiler[2])
+        self.c_layout = LayoutEnum.ROW_MAJOR
+        self.epi_tile = utils.sm100.compute_epilogue_tile_shape(
+            (self.pv_mma_tiler[0], self.pv_mma_tiler[1], self.pv_mma_tiler[2]), False, self.c_layout, self.o_dtype)
+        self.num_ab_stage = 1; self.num_acc_stage = 1
+
+        self.a_smem_s = utils.sm100.make_smem_layout_a(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.b_smem_s = utils.sm100.make_smem_layout_b(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.v_smem_s = utils.sm100.make_smem_layout_b(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.p_tmem_s = utils.sm100.make_smem_layout_a(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.c_smem_s = utils.sm100.make_smem_layout_epi(self.o_dtype, self.c_layout, self.epi_tile, 2)
+
+        qk_thr = qk_mma.get_slice(0)
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        s_cols = find_tmem_tensor_col_offset(tStS)
+        pv_thr = pv_mma.get_slice(0)
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        o_cols = find_tmem_tensor_col_offset(tOtO)
+
+        self.tilePlikeFP32 = self.qk_mma_tiler[1] // Float32.width * self.o_dtype.width
+        self.tmem_s0_offset = 0
+        self.tmem_p0_offset = 32
+        self.tmem_o0_offset = s_cols
+        tP = cute.make_tensor(tStS.iterator, self.p_tmem_s.outer)
+        tOrP2 = pv_mma.get_slice(0).make_fragment_A(tP)
+        tOrP2_s = tOrP2[(None, None, None, 0)]
+        sz0 = int(cute.size(tOrP2_s, mode=[0]))
+        sz1 = int(cute.size(tOrP2_s, mode=[1]))
+        sz2 = int(cute.size(tOrP2_s, mode=[2])) if cute.rank(tOrP2_s.layout) >= 3 else 0
+        print(sz0, sz1, sz2, int(cute.size(tStS)), int(cute.size(tP)), int(cute.size(tOrP2_s)))
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+        self.num_tmem_alloc_cols = utils.get_num_tmem_alloc_cols([tCtS_fake, tCtO_fake], arch="sm_100")
+
+        # ⛔⛔⛔ CRITICAL: num_tma_load_bytes MUST include ALL TMA-loaded tensors (Q + K + V). Missing V → DEADLOCK. See FOOTGUN #0 in README.
+        a_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        b_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        self.num_tma_load_bytes = (
+            cute.size_in_bytes(self.q_dtype, a_smem) + cute.size_in_bytes(self.q_dtype, b_smem) +
+            cute.size_in_bytes(self.q_dtype, v_smem)
+        ) * cute.size(qk_mma.thr_id.shape)
+
+    @cute.jit
+    def __call__(self, q, k, v, c, stream):
+        self.q_dtype = q.element_type; self.o_dtype = c.element_type; self.c_dtype = self.o_dtype
+        self.a_major = LayoutEnum.from_tensor(q).mma_major_mode()
+        self.b_major = LayoutEnum.from_tensor(k).mma_major_mode()
+        self.v_major = LayoutEnum.from_tensor(v).mma_major_mode()
+        self.c_layout = LayoutEnum.from_tensor(c)
+
+        qk_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, self.a_major, self.b_major,
+            self.qk_acc_dtype, self.cta_group, self.mma_tiler_mn, tcgen05.OperandSource.SMEM)
+        # PV with 128x128 output (V=I)
+        pv_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, cute.nvgpu.OperandMajorMode.K, self.v_major,
+            self.qk_acc_dtype, self.cta_group, (128, 32), tcgen05.OperandSource.TMEM)
+        self._setup(qk_mma, pv_mma)
+
+        q_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        k_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        tma_q, tma_tq = cute.nvgpu.make_tiled_tma_atom_A(
+            utils.sm100.cluster_shape_to_tma_atom_A(self.cluster_shape_mn, qk_mma.thr_id),
+            q, q_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_k, tma_tk = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, qk_mma.thr_id),
+            k, k_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_v, tma_tv = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, pv_mma.thr_id),
+            v, v_smem, self.pv_mma_tiler, pv_mma, self.cluster_layout_vmnk.shape)
+        epi_smem = cute.select(self.c_smem_s, mode=[0, 1])
+        tma_c, tma_tc = cpasync.make_tiled_tma_atom(cpasync.CopyBulkTensorTileS2GOp(), c, epi_smem, self.epi_tile)
+
+        self._kernel(qk_mma, pv_mma, tma_q, tma_tq, tma_k, tma_tk, tma_v, tma_tv,
+            tma_c, tma_tc, self.cluster_layout_vmnk,
+            self.a_smem_s, self.b_smem_s, self.v_smem_s, self.p_tmem_s, self.c_smem_s, self.epi_tile
+        ).launch(grid=(1,1,1), block=[self.threads_per_cta,1,1], stream=stream)
+
+    @cute.kernel
+    def _kernel(self, qk_mma, pv_mma, tma_q, mQ, tma_k, mK, tma_v, mV,
+                tma_c, mC, cl_vmnk, a_smem_s, b_smem_s, v_smem_s, p_tmem_s, c_smem_s, epi_tile):
+        warp_idx = cute.arch.make_warp_uniform(cute.arch.warp_idx())
+        tidx, _, _ = cute.arch.thread_idx()
+        use_2cta = cute.size(qk_mma.thr_id.shape) == 2
+
+        if warp_idx == self.tma_warp_id:
+            cpasync.prefetch_descriptor(tma_q); cpasync.prefetch_descriptor(tma_k)
+            cpasync.prefetch_descriptor(tma_v); cpasync.prefetch_descriptor(tma_c)
+
+        @cute.struct
+        class SS:
+            ab_bar: cute.struct.MemRange[cutlass.Int64, self.num_ab_stage * 2]
+            mma_si_bar: cute.struct.MemRange[cutlass.Int64, 2]
+            acc_bar: cute.struct.MemRange[cutlass.Int64, self.num_acc_stage * 2]
+            tmem_dealloc: cutlass.Int64
+            holding: cutlass.Int32
+
+        smem = utils.SmemAllocator(); st = smem.allocate(SS)
+
+        ab_p, ab_c = pipeline.PipelineTmaUmma.create(
+            barrier_storage=st.ab_bar.data_ptr(), num_stages=self.num_ab_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 1),
+            tx_count=self.num_tma_load_bytes, cta_layout_vmnk=cl_vmnk, defer_sync=True
+        ).make_participants()
+
+        mma_si_prod, mma_si_cons = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.mma_si_bar.data_ptr(), num_stages=1,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id)),
+        ).make_participants()
+
+        acc_pipe = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.acc_bar.data_ptr(), num_stages=self.num_acc_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(
+                pipeline.Agent.Thread, len(self.epilogue_warp_id) * (2 if use_2cta else 1)),
+            cta_layout_vmnk=cl_vmnk, defer_sync=True)
+
+        tmem_bar = pipeline.NamedBarrier(barrier_id=2,
+            num_threads=32 * len((self.mma_warp_id, *self.epilogue_warp_id)))
+        tmem = utils.TmemAllocator(st.holding.ptr, barrier_for_retrieve=tmem_bar,
+            allocator_warp_id=self.epilogue_warp_id[0], is_two_cta=use_2cta,
+            two_cta_tmem_dealloc_mbar_ptr=st.tmem_dealloc.ptr)
+
+        pipeline.pipeline_init_arrive(cluster_shape_mn=cl_vmnk, is_relaxed=True)
+
+        sQ = smem.allocate_tensor(element_type=self.q_dtype, layout=a_smem_s.outer, byte_alignment=128, swizzle=a_smem_s.inner)
+        sK = smem.allocate_tensor(element_type=self.q_dtype, layout=b_smem_s.outer, byte_alignment=128, swizzle=b_smem_s.inner)
+        sV = smem.allocate_tensor(element_type=self.q_dtype, layout=v_smem_s.outer, byte_alignment=128, swizzle=v_smem_s.inner)
+        sC = smem.allocate_tensor(element_type=self.o_dtype, layout=c_smem_s.outer, byte_alignment=128, swizzle=c_smem_s.inner)
+
+        gQ = cute.local_tile(mQ, cute.slice_(self.qk_mma_tiler, (None,0,None)), (None,None,None))
+        gK = cute.local_tile(mK, cute.slice_(self.qk_mma_tiler, (0,None,None)), (None,None,None))
+        gC = cute.local_tile(mC, cute.slice_(self.qk_mma_tiler, (None,None,0)), (None,None,None))
+        k_cnt = cute.size(gQ, mode=[3])
+
+        qk_thr = qk_mma.get_slice(0)
+        pv_thr = pv_mma.get_slice(0)
+        tCgQ = qk_thr.partition_A(gQ); tCgK = qk_thr.partition_B(gK); tCgC = qk_thr.partition_C(gC)
+        a_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,0,None,0)).shape)
+        tAsQ, tAgQ = cpasync.tma_partition(tma_q, 0, a_lay, cute.group_modes(sQ,0,3), cute.group_modes(tCgQ,0,3))
+        b_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,None,0,0)).shape)
+        tBsK, tBgK = cpasync.tma_partition(tma_k, 0, b_lay, cute.group_modes(sK,0,3), cute.group_modes(tCgK,0,3))
+        tAgQ = tAgQ[(None,0,None,0)]; tBgK = tBgK[(None,0,None,0)]
+
+        gV = cute.local_tile(mV, cute.slice_(self.pv_mma_tiler, (0,None,None)), (None,None,None))
+        tCgV = pv_thr.partition_B(gV)
+        tVsV, tVgV = cpasync.tma_partition(tma_v, 0, b_lay, cute.group_modes(sV,0,3), cute.group_modes(tCgV,0,3))
+        tVgV = tVgV[(None,0,None,0)]
+
+        tCrQ = qk_mma.make_fragment_A(sQ); tCrK = qk_mma.make_fragment_B(sK)
+        tCrV = pv_mma.make_fragment_B(sV)
+
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        tStS0 = cute.make_tensor(tStS.iterator + self.tmem_s0_offset, tStS.layout)
+
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        tOtO0 = cute.make_tensor(tOtO.iterator + self.tmem_o0_offset, tOtO.layout)
+
+        tP = cute.make_tensor(tStS.iterator, p_tmem_s.outer)
+        tOrP_base = pv_thr.make_fragment_A(tP)
+        tOrP = tOrP_base[(None, None, None, 0)]
+        tOrP0 = cute.make_tensor(
+            tOrP.iterator + self.qk_acc_dtype.width // self.q_dtype.width * self.tmem_p0_offset,
+            tOrP.layout)
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+
+        pipeline.pipeline_init_wait(cluster_shape_mn=cl_vmnk)
+
+        # ═══ TMA LOAD WARP ═══
+        if warp_idx == self.tma_warp_id:
+            ab_p.reset(); peek = ab_p.try_acquire()
+            for kt in cutlass.range(k_cnt, unroll=1):
+                h = ab_p.acquire_and_advance(peek)
+                cute.copy(tma_q, tAgQ[(None,h.count)], tAsQ[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_k, tBgK[(None,h.count)], tBsK[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_v, tVgV[(None,h.count)], tVsV[(None,h.index)], tma_bar_ptr=h.barrier)
+                peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_p.try_acquire()
+            ab_p.tail()
+
+        # ═══ MMA WARP ═══
+        if warp_idx == self.mma_warp_id:
+            tmem.wait_for_alloc()
+            ab_c.reset(); peek = ab_c.try_wait()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            acc_prod_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Producer, self.num_acc_stage)
+            acc_pipe.producer_acquire(acc_prod_st)
+            qk_mma.set(tcgen05.Field.ACCUMULATE, False)
+            for kt in range(k_cnt):
+                h = ab_c.wait_and_advance(peek)
+                nblk = cute.size(tCrQ, mode=[2])
+                for kb in cutlass.range(nblk, unroll_full=True):
+                    cute.gemm(qk_mma, tStS0, tCrQ[(None,None,kb,h.index)], tCrK[(None,None,kb,h.index)], tStS0)
+                    qk_mma.set(tcgen05.Field.ACCUMULATE, True)
+                h.release(); peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_c.try_wait()
+
+            cute.arch.fence_view_async_tmem_store()
+            s0_handle.commit()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            # PV MMA: P @ V where V=I → O = P
+            pv_mma.set(tcgen05.Field.ACCUMULATE, False)
+            tCrV_s = tCrV[(None, None, None, 0)]
+            nblk_pv = cute.size(tOrP0, mode=[2])
+            for kb in cutlass.range(nblk_pv, unroll_full=True):
+                cute.gemm(pv_mma, tOtO0, tOrP0[(None,None,kb)], tCrV_s[(None,None,kb)], tOtO0)
+                pv_mma.set(tcgen05.Field.ACCUMULATE, True)
+
+            acc_pipe.producer_commit(acc_prod_st)
+            acc_prod_st.advance()
+            acc_pipe.producer_tail(acc_prod_st)
+
+        # ═══ EPILOGUE WARPS ═══
+        if warp_idx < self.mma_warp_id:
+            tmem.allocate(self.num_tmem_alloc_cols)
+            tmem.wait_for_alloc()
+            tmem_ptr = tmem.retrieve_ptr(self.qk_acc_dtype)
+            sfw_idx = tidx % (32 * len(self.epilogue_warp_id))
+
+            tmem_load_atom = cute.make_copy_atom(
+                tcgen05.copy.Ld32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_load = tcgen05.make_tmem_copy(tmem_load_atom, tStS0)
+            thr_load = tiled_tmem_load.get_slice(sfw_idx)
+            tTMEM_LOADtS = thr_load.partition_S(tStS0)
+            cS = cute.make_identity_tensor((self.qk_mma_tiler[0], self.qk_mma_tiler[1]))
+            tScS = qk_thr.partition_C(cS)
+            tTMEM_LOADcS = thr_load.partition_D(tScS)
+
+            tStS_P_layout = cute.composition(tStS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tStS_P = cute.make_tensor(tStS.iterator + self.tmem_p0_offset, tStS_P_layout)
+            tmem_store_atom = cute.make_copy_atom(
+                tcgen05.copy.St32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_store = tcgen05.make_tmem_copy(tmem_store_atom, tStS_P)
+            thr_store = tiled_tmem_store.get_slice(sfw_idx)
+            tTMEM_STOREtS_x4 = thr_store.partition_D(tStS_P)
+            tScS_P_layout = cute.composition(tScS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tScS_P = cute.make_tensor(tScS.iterator, tScS_P_layout)
+            tTMEM_STOREcS = thr_store.partition_S(tScS_P)
+
+            si_handle = mma_si_cons.wait_and_advance()
+
+            tTMEM_LOADrS = cute.make_rmem_tensor(tTMEM_LOADcS.shape, self.qk_acc_dtype)
+            cute.copy(tiled_tmem_load, tTMEM_LOADtS, tTMEM_LOADrS)
+            tTMEM_STORErS_x4 = cute.make_rmem_tensor(tTMEM_STOREcS.shape, self.qk_acc_dtype)
+            tTMEM_STORErS_x4_e = cute.make_tensor(
+                cute.recast_ptr(tTMEM_STORErS_x4.iterator, dtype=self.q_dtype),
+                tTMEM_LOADrS.layout)
+            frg_cnt = 4
+            frg_tile = cute.size(tTMEM_LOADrS) // frg_cnt
+            tTMEM_LOADrS_frg = cute.logical_divide(tTMEM_LOADrS, cute.make_layout(frg_tile))
+            tTMEM_STORErS_x4_e_frg = cute.logical_divide(
+                tTMEM_STORErS_x4_e, cute.make_layout(frg_tile))
+            for j in range(frg_cnt):
+                s_vec = tTMEM_LOADrS_frg[None, j].load()
+                tTMEM_STORErS_x4_e_frg[None, j].store(s_vec.to(self.q_dtype))
+            cute.copy(tiled_tmem_store, tTMEM_STORErS_x4, tTMEM_STOREtS_x4)
+            cute.arch.fence_view_async_tmem_store()
+            si_handle.release()
+
+            # Output epilogue
+            tCtO_base = cute.make_tensor(tmem_ptr + self.tmem_o0_offset, tCtO_fake.layout)
+            acc_cons_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Consumer, self.num_acc_stage)
+            c_grp = pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id))
+            c_pipe = pipeline.PipelineTmaStore.create(num_stages=self.num_c_stage, producer_group=c_grp)
+            acc_cons_st = utils.gemm.sm100.epilogue_tma_store(
+                self, tidx, warp_idx, tma_c, tCtO_base, sC, tCgC,
+                epi_tile, 0, const_expr(lambda x: x), (0,0,0), acc_cons_st, acc_pipe, c_pipe)
+            c_pipe.producer_tail()
+            tmem.relinquish_alloc_permit()
+            tmem.free(tmem_ptr)
+
+
+def test():
+    torch.manual_seed(42)
+    m, n, head_dim = 128, 128, 64
+    q = torch.randn(m, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    k = torch.randn(n, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    # V = identity (128x128) in MN-major: (128,128) with strides (1,128)
+    v = torch.eye(128, dtype=torch.bfloat16, device='cuda')
+    # MN-major: (128,128) with strides (1,128) — row is fast dim
+    v = v.as_strided((128, 128), (1, 128)).unsqueeze(-1)
+    c = torch.zeros(m, n, 1, dtype=torch.bfloat16, device='cuda')
+
+    qf = q[:,:,0].float(); kf = k[:,:,0].float()
+    # With V=I and identity softmax: O = (Q@K^T).bf16() @ I = (Q@K^T).bf16()
+    ref = (qf @ kf.T).bfloat16().float()
+
+    mQ = ct.from_dlpack(q).mark_layout_dynamic(leading_dim=ct.get_leading_dim(q))
+    mK = ct.from_dlpack(k).mark_layout_dynamic(leading_dim=ct.get_leading_dim(k))
+    mV = ct.from_dlpack(v).mark_layout_dynamic(leading_dim=ct.get_leading_dim(v))
+    mC = ct.from_dlpack(c).mark_layout_dynamic(leading_dim=ct.get_leading_dim(c))
+    stream = cuda.CUstream(torch.cuda.current_stream().cuda_stream)
+    kernel = TmemCol5_32(mma_tiler_mn=(128, 128))
+    print('Compiling...', flush=True)
+    compiled = cute.compile(kernel, mQ, mK, mV, mC, stream)
+    print('Running...', flush=True)
+    compiled(mQ, mK, mV, mC, stream)
+    torch.cuda.synchronize()
+    out = c[:,:,0].float()
+    cos = torch.nn.functional.cosine_similarity(out.flatten().unsqueeze(0), ref.flatten().unsqueeze(0)).item()
+    print('Col5(128,32): cosine {:.6f} {}'.format(cos, 'PASS' if cos >= 0.99 else 'FAIL'))
+
+
+if __name__ == '__main__':
+    test()

tests/test_tmem_col_offset.py

@@ -0,0 +1,388 @@
+"""
+Minimal PV-only test: Load P from GMEM to TMEM via QK-style MMA, then PV from TMEM.
+Step 1: QK MMA writes FP32 S to TMEM (we know this works)
+Step 2: Softmax packing writes BF16 P to TMEM (test this)
+Step 3: PV MMA reads BF16 P from TMEM and V from SMEM, produces O
+
+But to isolate the bug, let me test just the PV MMA in isolation.
+I'll write known BF16 values to TMEM using the softmax packing path,
+then immediately read them back using the PV A-fragment path,
+and compare.
+
+Actually, the simplest isolation test:
+1. Do QK MMA to get S in TMEM (cosine 0.999999 verified)
+2. Do softmax packing: S → P in TMEM (at offset 32)
+3. Skip PV entirely — read P from TMEM using the C-fragment composition LOAD path
+4. Output P to GMEM and compare against S.to(BF16)
+
+This tests whether the softmax packing writes P correctly to the same TMEM
+that the PV would read from.
+
+But we can't easily read P from TMEM using the standard epilogue path
+because the epilogue expects FP32 accumulator data.
+
+Alternative: Use the PV MMA with V=I (identity). If P is correct,
+then P @ I = P. But V needs to be MN-major and (128, 128), not (128, 64).
+The output would be (128, 128) which doesn't match our (128, 64) c tensor.
+
+Let me use V that selects the first 64 columns: V[k, n] = delta(k, n) for k in [0,63].
+This gives P @ V = P[:, :64], and the output is (128, 64).
+But V is (128, 128) in the MMA K,N dims. V[k, n] for k in [0,127], n in [0,63].
+Hmm, this is getting complicated. Let me just do the identity approach with a (128, 128) output.
+"""
+import torch, cutlass, cutlass.cute as cute, cutlass.utils as utils, cutlass.pipeline as pipeline
+from cutlass.cute.nvgpu import cpasync, tcgen05
+from cutlass import Float32, BFloat16, Int32, Boolean, const_expr
+from cutlass.utils import LayoutEnum
+from cutlass.utils.tmem_allocator import find_tmem_tensor_col_offset
+import cuda.bindings.driver as cuda
+import cutlass.torch as ct
+
+
+class TmemColOffsetKernel:
+    """QK + softmax packing + PV with V=I to isolate PV MMA correctness.
+    Output should be P = S.to(BF16), i.e. (Q@K^T).bfloat16()
+    With V=I, O = P @ I = P.
+    But V is (K=128, N=128) in the MMA. We need a 128x128 identity in MN-major.
+    Output tensor is (128, 128).
+    """
+    def __init__(self, mma_tiler_mn):
+        self.acc_dtype = Float32; self.qk_acc_dtype = Float32
+        self.q_dtype = BFloat16; self.o_dtype = BFloat16; self.c_dtype = BFloat16
+        self.mma_tiler_mn = mma_tiler_mn; self.mma_tiler = (*mma_tiler_mn, 1)
+        self.use_2cta_instrs = False  # needed by epilogue_tma_store
+        self.epilog_sync_bar_id = 1  # needed by epilogue_tma_store
+        self.cluster_shape_mn = (1, 1)
+        self.cta_group = tcgen05.CtaGroup.ONE
+        self.epilogue_warp_id = (0, 1, 2, 3)
+        self.mma_warp_id = 4; self.tma_warp_id = 5
+        self.threads_per_cta = 192
+        self.num_c_stage = 2
+
+    def _setup(self, qk_mma, pv_mma):
+        qk_inst_k = int(cute.size(qk_mma.shape_mnk, mode=[2]))
+        self.qk_mma_tiler = (*self.mma_tiler_mn, qk_inst_k * 4)
+        # PV with V=I: output is (128, 128), same as QK
+        self.pv_mma_tiler = (self.qk_mma_tiler[0], self.qk_mma_tiler[1], self.qk_mma_tiler[1])
+        # pv_mma_tiler = (128, 128, 128) since V is 128x128
+        self.mma_tiler = self.qk_mma_tiler
+
+        self.cluster_layout_vmnk = cute.tiled_divide(cute.make_layout((1,1,1)), (qk_mma.thr_id.shape,))
+        self.cta_tile_shape_mnk = (
+            self.qk_mma_tiler[0] // cute.size(qk_mma.thr_id.shape),
+            self.qk_mma_tiler[1], self.qk_mma_tiler[2])
+        self.c_layout = LayoutEnum.ROW_MAJOR
+        self.epi_tile = utils.sm100.compute_epilogue_tile_shape(
+            self.cta_tile_shape_mnk, False, self.c_layout, self.o_dtype)
+        self.num_ab_stage = 1; self.num_acc_stage = 1
+
+        self.a_smem_s = utils.sm100.make_smem_layout_a(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.b_smem_s = utils.sm100.make_smem_layout_b(qk_mma, self.mma_tiler, self.q_dtype, 1)
+        self.v_smem_s = utils.sm100.make_smem_layout_b(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.p_tmem_s = utils.sm100.make_smem_layout_a(pv_mma, self.pv_mma_tiler, self.q_dtype, 1)
+        self.c_smem_s = utils.sm100.make_smem_layout_epi(self.o_dtype, self.c_layout, self.epi_tile, 2)
+
+        qk_thr = qk_mma.get_slice(0)
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        s_cols = find_tmem_tensor_col_offset(tStS)
+        pv_thr = pv_mma.get_slice(0)
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        o_cols = find_tmem_tensor_col_offset(tOtO)
+
+        self.tilePlikeFP32 = self.qk_mma_tiler[1] // Float32.width * self.o_dtype.width
+        self.tmem_s0_offset = 0
+        self.tmem_p0_offset = 32
+        self.tmem_o0_offset = s_cols
+        # Compare TMEM column offsets: QK C-fragment S vs PV A-fragment P
+        tP = cute.make_tensor(tStS.iterator, self.p_tmem_s.outer)
+        tOrP2 = pv_mma.get_slice(0).make_fragment_A(tP)
+        pv_col = find_tmem_tensor_col_offset(tOrP2)
+        print(f"TMEM col: S_base={s_cols}, P_Afrag_at_S_base={pv_col}, O_base={o_cols}")
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+        self.num_tmem_alloc_cols = utils.get_num_tmem_alloc_cols([tCtS_fake, tCtO_fake], arch="sm_100")
+
+        # ⛔⛔⛔ CRITICAL: num_tma_load_bytes MUST include ALL TMA-loaded tensors (Q + K + V). Missing V → DEADLOCK. See FOOTGUN #0 in README.
+        a_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        b_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        self.num_tma_load_bytes = (
+            cute.size_in_bytes(self.q_dtype, a_smem) + cute.size_in_bytes(self.q_dtype, b_smem) +
+            cute.size_in_bytes(self.q_dtype, v_smem)
+        ) * cute.size(qk_mma.thr_id.shape)
+
+    @cute.jit
+    def __call__(self, q, k, v, c, stream):
+        self.q_dtype = q.element_type; self.o_dtype = c.element_type; self.c_dtype = self.o_dtype
+        self.a_major = LayoutEnum.from_tensor(q).mma_major_mode()
+        self.b_major = LayoutEnum.from_tensor(k).mma_major_mode()
+        self.v_major = LayoutEnum.from_tensor(v).mma_major_mode()
+        self.c_layout = LayoutEnum.from_tensor(c)
+
+        qk_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, self.a_major, self.b_major,
+            self.qk_acc_dtype, self.cta_group, self.mma_tiler_mn, tcgen05.OperandSource.SMEM)
+        # PV with 128x128 output (V=I)
+        pv_mma = utils.sm100.make_trivial_tiled_mma(
+            self.q_dtype, self.q_dtype, cute.nvgpu.OperandMajorMode.K, self.v_major,
+            self.qk_acc_dtype, self.cta_group, self.mma_tiler_mn, tcgen05.OperandSource.TMEM)
+        self._setup(qk_mma, pv_mma)
+
+        q_smem = cute.slice_(self.a_smem_s, (None, None, None, 0))
+        k_smem = cute.slice_(self.b_smem_s, (None, None, None, 0))
+        v_smem = cute.slice_(self.v_smem_s, (None, None, None, 0))
+        tma_q, tma_tq = cute.nvgpu.make_tiled_tma_atom_A(
+            utils.sm100.cluster_shape_to_tma_atom_A(self.cluster_shape_mn, qk_mma.thr_id),
+            q, q_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_k, tma_tk = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, qk_mma.thr_id),
+            k, k_smem, self.mma_tiler, qk_mma, self.cluster_layout_vmnk.shape)
+        tma_v, tma_tv = cute.nvgpu.make_tiled_tma_atom_B(
+            utils.sm100.cluster_shape_to_tma_atom_B(self.cluster_shape_mn, pv_mma.thr_id),
+            v, v_smem, self.pv_mma_tiler, pv_mma, self.cluster_layout_vmnk.shape)
+        epi_smem = cute.select(self.c_smem_s, mode=[0, 1])
+        tma_c, tma_tc = cpasync.make_tiled_tma_atom(cpasync.CopyBulkTensorTileS2GOp(), c, epi_smem, self.epi_tile)
+
+        self._kernel(qk_mma, pv_mma, tma_q, tma_tq, tma_k, tma_tk, tma_v, tma_tv,
+            tma_c, tma_tc, self.cluster_layout_vmnk,
+            self.a_smem_s, self.b_smem_s, self.v_smem_s, self.p_tmem_s, self.c_smem_s, self.epi_tile
+        ).launch(grid=(1,1,1), block=[self.threads_per_cta,1,1], stream=stream)
+
+    @cute.kernel
+    def _kernel(self, qk_mma, pv_mma, tma_q, mQ, tma_k, mK, tma_v, mV,
+                tma_c, mC, cl_vmnk, a_smem_s, b_smem_s, v_smem_s, p_tmem_s, c_smem_s, epi_tile):
+        warp_idx = cute.arch.make_warp_uniform(cute.arch.warp_idx())
+        tidx, _, _ = cute.arch.thread_idx()
+        use_2cta = cute.size(qk_mma.thr_id.shape) == 2
+
+        if warp_idx == self.tma_warp_id:
+            cpasync.prefetch_descriptor(tma_q); cpasync.prefetch_descriptor(tma_k)
+            cpasync.prefetch_descriptor(tma_v); cpasync.prefetch_descriptor(tma_c)
+
+        @cute.struct
+        class SS:
+            ab_bar: cute.struct.MemRange[cutlass.Int64, self.num_ab_stage * 2]
+            mma_si_bar: cute.struct.MemRange[cutlass.Int64, 2]
+            acc_bar: cute.struct.MemRange[cutlass.Int64, self.num_acc_stage * 2]
+            tmem_dealloc: cutlass.Int64
+            holding: cutlass.Int32
+
+        smem = utils.SmemAllocator(); st = smem.allocate(SS)
+
+        ab_p, ab_c = pipeline.PipelineTmaUmma.create(
+            barrier_storage=st.ab_bar.data_ptr(), num_stages=self.num_ab_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 1),
+            tx_count=self.num_tma_load_bytes, cta_layout_vmnk=cl_vmnk, defer_sync=True
+        ).make_participants()
+
+        mma_si_prod, mma_si_cons = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.mma_si_bar.data_ptr(), num_stages=1,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id)),
+        ).make_participants()
+
+        acc_pipe = pipeline.PipelineUmmaAsync.create(
+            barrier_storage=st.acc_bar.data_ptr(), num_stages=self.num_acc_stage,
+            producer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread),
+            consumer_group=pipeline.CooperativeGroup(
+                pipeline.Agent.Thread, len(self.epilogue_warp_id) * (2 if use_2cta else 1)),
+            cta_layout_vmnk=cl_vmnk, defer_sync=True)
+
+        tmem_bar = pipeline.NamedBarrier(barrier_id=2,
+            num_threads=32 * len((self.mma_warp_id, *self.epilogue_warp_id)))
+        tmem = utils.TmemAllocator(st.holding.ptr, barrier_for_retrieve=tmem_bar,
+            allocator_warp_id=self.epilogue_warp_id[0], is_two_cta=use_2cta,
+            two_cta_tmem_dealloc_mbar_ptr=st.tmem_dealloc.ptr)
+
+        pipeline.pipeline_init_arrive(cluster_shape_mn=cl_vmnk, is_relaxed=True)
+
+        sQ = smem.allocate_tensor(element_type=self.q_dtype, layout=a_smem_s.outer, byte_alignment=128, swizzle=a_smem_s.inner)
+        sK = smem.allocate_tensor(element_type=self.q_dtype, layout=b_smem_s.outer, byte_alignment=128, swizzle=b_smem_s.inner)
+        sV = smem.allocate_tensor(element_type=self.q_dtype, layout=v_smem_s.outer, byte_alignment=128, swizzle=v_smem_s.inner)
+        sC = smem.allocate_tensor(element_type=self.o_dtype, layout=c_smem_s.outer, byte_alignment=128, swizzle=c_smem_s.inner)
+
+        gQ = cute.local_tile(mQ, cute.slice_(self.qk_mma_tiler, (None,0,None)), (None,None,None))
+        gK = cute.local_tile(mK, cute.slice_(self.qk_mma_tiler, (0,None,None)), (None,None,None))
+        gC = cute.local_tile(mC, cute.slice_(self.qk_mma_tiler, (None,None,0)), (None,None,None))
+        k_cnt = cute.size(gQ, mode=[3])
+
+        qk_thr = qk_mma.get_slice(0)
+        pv_thr = pv_mma.get_slice(0)
+        tCgQ = qk_thr.partition_A(gQ); tCgK = qk_thr.partition_B(gK); tCgC = qk_thr.partition_C(gC)
+        a_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,0,None,0)).shape)
+        tAsQ, tAgQ = cpasync.tma_partition(tma_q, 0, a_lay, cute.group_modes(sQ,0,3), cute.group_modes(tCgQ,0,3))
+        b_lay = cute.make_layout(cute.slice_(cl_vmnk, (0,None,0,0)).shape)
+        tBsK, tBgK = cpasync.tma_partition(tma_k, 0, b_lay, cute.group_modes(sK,0,3), cute.group_modes(tCgK,0,3))
+        tAgQ = tAgQ[(None,0,None,0)]; tBgK = tBgK[(None,0,None,0)]
+
+        gV = cute.local_tile(mV, cute.slice_(self.pv_mma_tiler, (0,None,None)), (None,None,None))
+        tCgV = pv_thr.partition_B(gV)
+        tVsV, tVgV = cpasync.tma_partition(tma_v, 0, b_lay, cute.group_modes(sV,0,3), cute.group_modes(tCgV,0,3))
+        tVgV = tVgV[(None,0,None,0)]
+
+        tCrQ = qk_mma.make_fragment_A(sQ); tCrK = qk_mma.make_fragment_B(sK)
+        tCrV = pv_mma.make_fragment_B(sV)
+
+        qk_acc_shape = qk_thr.partition_shape_C(self.qk_mma_tiler[:2])
+        tStS = qk_thr.make_fragment_C(qk_acc_shape)
+        tStS0 = cute.make_tensor(tStS.iterator + self.tmem_s0_offset, tStS.layout)
+
+        pv_acc_shape = pv_thr.partition_shape_C(self.pv_mma_tiler[:2])
+        tOtO = pv_thr.make_fragment_C(pv_acc_shape)
+        tOtO0 = cute.make_tensor(tOtO.iterator + self.tmem_o0_offset, tOtO.layout)
+
+        tP = cute.make_tensor(tStS.iterator, self.p_tmem_s.outer)
+        tOrP_base = pv_thr.make_fragment_A(tP)
+        tOrP = tOrP_base[(None, None, None, 0)]
+        tOrP0 = cute.make_tensor(
+            tOrP.iterator + self.qk_acc_dtype.width // self.q_dtype.width * self.tmem_p0_offset,
+            tOrP.layout)
+
+        tCtS_fake = qk_mma.make_fragment_C(cute.append(qk_acc_shape, self.num_acc_stage))
+        tCtO_fake = pv_mma.make_fragment_C(cute.append(pv_acc_shape, self.num_acc_stage))
+
+        pipeline.pipeline_init_wait(cluster_shape_mn=cl_vmnk)
+
+        # ═══ TMA LOAD WARP ═══
+        if warp_idx == self.tma_warp_id:
+            ab_p.reset(); peek = ab_p.try_acquire()
+            for kt in cutlass.range(k_cnt, unroll=1):
+                h = ab_p.acquire_and_advance(peek)
+                cute.copy(tma_q, tAgQ[(None,h.count)], tAsQ[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_k, tBgK[(None,h.count)], tBsK[(None,h.index)], tma_bar_ptr=h.barrier)
+                cute.copy(tma_v, tVgV[(None,h.count)], tVsV[(None,h.index)], tma_bar_ptr=h.barrier)
+                peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_p.try_acquire()
+            ab_p.tail()
+
+        # ═══ MMA WARP ═══
+        if warp_idx == self.mma_warp_id:
+            tmem.wait_for_alloc()
+            ab_c.reset(); peek = ab_c.try_wait()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            acc_prod_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Producer, self.num_acc_stage)
+            acc_pipe.producer_acquire(acc_prod_st)
+            qk_mma.set(tcgen05.Field.ACCUMULATE, False)
+            for kt in range(k_cnt):
+                h = ab_c.wait_and_advance(peek)
+                nblk = cute.size(tCrQ, mode=[2])
+                for kb in cutlass.range(nblk, unroll_full=True):
+                    cute.gemm(qk_mma, tStS0, tCrQ[(None,None,kb,h.index)], tCrK[(None,None,kb,h.index)], tStS0)
+                    qk_mma.set(tcgen05.Field.ACCUMULATE, True)
+                h.release(); peek = cutlass.Boolean(1)
+                if h.count+1<k_cnt: peek = ab_c.try_wait()
+
+            cute.arch.fence_view_async_tmem_store()
+            s0_handle.commit()
+            s0_handle = mma_si_prod.acquire_and_advance()
+
+            # PV MMA: P @ V where V=I → O = P
+            pv_mma.set(tcgen05.Field.ACCUMULATE, False)
+            tCrV_s = tCrV[(None, None, None, 0)]
+            nblk_pv = cute.size(tOrP0, mode=[2])
+            for kb in cutlass.range(nblk_pv, unroll_full=True):
+                cute.gemm(pv_mma, tOtO0, tOrP0[(None,None,kb)], tCrV_s[(None,None,kb)], tOtO0)
+                pv_mma.set(tcgen05.Field.ACCUMULATE, True)
+
+            acc_pipe.producer_commit(acc_prod_st)
+            acc_prod_st.advance()
+            acc_pipe.producer_tail(acc_prod_st)
+
+        # ═══ EPILOGUE WARPS ═══
+        if warp_idx < self.mma_warp_id:
+            tmem.allocate(self.num_tmem_alloc_cols)
+            tmem.wait_for_alloc()
+            tmem_ptr = tmem.retrieve_ptr(self.qk_acc_dtype)
+            sfw_idx = tidx % (32 * len(self.epilogue_warp_id))
+
+            tmem_load_atom = cute.make_copy_atom(
+                tcgen05.copy.Ld32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_load = tcgen05.make_tmem_copy(tmem_load_atom, tStS0)
+            thr_load = tiled_tmem_load.get_slice(sfw_idx)
+            tTMEM_LOADtS = thr_load.partition_S(tStS0)
+            cS = cute.make_identity_tensor((self.qk_mma_tiler[0], self.qk_mma_tiler[1]))
+            tScS = qk_thr.partition_C(cS)
+            tTMEM_LOADcS = thr_load.partition_D(tScS)
+
+            tStS_P_layout = cute.composition(tStS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tStS_P = cute.make_tensor(tStS.iterator + self.tmem_p0_offset, tStS_P_layout)
+            tmem_store_atom = cute.make_copy_atom(
+                tcgen05.copy.St32x32bOp(tcgen05.copy.Repetition(32)), self.qk_acc_dtype)
+            tiled_tmem_store = tcgen05.make_tmem_copy(tmem_store_atom, tStS_P)
+            thr_store = tiled_tmem_store.get_slice(sfw_idx)
+            tTMEM_STOREtS_x4 = thr_store.partition_D(tStS_P)
+            tScS_P_layout = cute.composition(tScS.layout, cute.make_layout((128, self.tilePlikeFP32)))
+            tScS_P = cute.make_tensor(tScS.iterator, tScS_P_layout)
+            tTMEM_STOREcS = thr_store.partition_S(tScS_P)
+
+            si_handle = mma_si_cons.wait_and_advance()
+
+            tTMEM_LOADrS = cute.make_rmem_tensor(tTMEM_LOADcS.shape, self.qk_acc_dtype)
+            cute.copy(tiled_tmem_load, tTMEM_LOADtS, tTMEM_LOADrS)
+            tTMEM_STORErS_x4 = cute.make_rmem_tensor(tTMEM_STOREcS.shape, self.qk_acc_dtype)
+            tTMEM_STORErS_x4_e = cute.make_tensor(
+                cute.recast_ptr(tTMEM_STORErS_x4.iterator, dtype=self.q_dtype),
+                tTMEM_LOADrS.layout)
+            frg_cnt = 4
+            frg_tile = cute.size(tTMEM_LOADrS) // frg_cnt
+            tTMEM_LOADrS_frg = cute.logical_divide(tTMEM_LOADrS, cute.make_layout(frg_tile))
+            tTMEM_STORErS_x4_e_frg = cute.logical_divide(
+                tTMEM_STORErS_x4_e, cute.make_layout(frg_tile))
+            for j in range(frg_cnt):
+                s_vec = tTMEM_LOADrS_frg[None, j].load()
+                tTMEM_STORErS_x4_e_frg[None, j].store(s_vec.to(self.q_dtype))
+            cute.copy(tiled_tmem_store, tTMEM_STORErS_x4, tTMEM_STOREtS_x4)
+            cute.arch.fence_view_async_tmem_store()
+            si_handle.release()
+
+            # Output epilogue
+            tCtO_base = cute.make_tensor(tmem_ptr + self.tmem_o0_offset, tCtO_fake.layout)
+            acc_cons_st = pipeline.make_pipeline_state(pipeline.PipelineUserType.Consumer, self.num_acc_stage)
+            c_grp = pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(self.epilogue_warp_id))
+            c_pipe = pipeline.PipelineTmaStore.create(num_stages=self.num_c_stage, producer_group=c_grp)
+            acc_cons_st = utils.gemm.sm100.epilogue_tma_store(
+                self, tidx, warp_idx, tma_c, tCtO_base, sC, tCgC,
+                epi_tile, 0, const_expr(lambda x: x), (0,0,0), acc_cons_st, acc_pipe, c_pipe)
+            c_pipe.producer_tail()
+            tmem.relinquish_alloc_permit()
+            tmem.free(tmem_ptr)
+
+
+def test():
+    torch.manual_seed(42)
+    m, n, head_dim = 128, 128, 64
+    q = torch.randn(m, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    k = torch.randn(n, head_dim, 1, dtype=torch.bfloat16, device='cuda')
+    # V = identity (128x128) in MN-major: (128,128) with strides (1,128)
+    v = torch.eye(128, dtype=torch.bfloat16, device='cuda')
+    # MN-major: (128,128) with strides (1,128) — row is fast dim
+    v = v.as_strided((128, 128), (1, 128)).unsqueeze(-1)
+    c = torch.zeros(m, n, 1, dtype=torch.bfloat16, device='cuda')
+
+    qf = q[:,:,0].float(); kf = k[:,:,0].float()
+    # With V=I and identity softmax: O = (Q@K^T).bf16() @ I = (Q@K^T).bf16()
+    ref = (qf @ kf.T).bfloat16().float()
+
+    mQ = ct.from_dlpack(q).mark_layout_dynamic(leading_dim=ct.get_leading_dim(q))
+    mK = ct.from_dlpack(k).mark_layout_dynamic(leading_dim=ct.get_leading_dim(k))
+    mV = ct.from_dlpack(v).mark_layout_dynamic(leading_dim=ct.get_leading_dim(v))
+    mC = ct.from_dlpack(c).mark_layout_dynamic(leading_dim=ct.get_leading_dim(c))
+    stream = cuda.CUstream(torch.cuda.current_stream().cuda_stream)
+    kernel = TmemColOffsetKernel(mma_tiler_mn=(128, 128))
+    print('Compiling...', flush=True)
+    compiled = cute.compile(kernel, mQ, mK, mV, mC, stream)
+    print('Running...', flush=True)
+    compiled(mQ, mK, mV, mC, stream)
+    torch.cuda.synchronize()
+    out = c[:,:,0].float()
+    cos = torch.nn.functional.cosine_similarity(out.flatten().unsqueeze(0), ref.flatten().unsqueeze(0)).item()
+    print('TmemColOffset(128,128): cosine {:.6f} {}'.format(cos, 'PASS' if cos >= 0.99 else 'FAIL'))
+
+
+if __name__ == '__main__':
+    test()