nvfp4-megamoe-kernel/README.md

DeepSeek-V4 NVFP4 Kernel Suite

CuTeDSL kernels for DeepSeek-V4 (Blackwell B200, SM100). All kernels use cutlass.cute (CuTeDSL) with Blackwell tensor cores.

Status (May 21, 2026 — 05:15 UTC)

✅ Stage A: Bare Q@K^T via tcgen05.mma → TMEM → GMEM — COMPLETE

File: tests/test_stage_a_v2.py Result: Q(128,128) @ K^T(128,128) → S(128,128), cosine 0.999999

🔨 Stage B: Two MMAs + Identity Softmax — IN PROGRESS

Pipeline deadlock: FIXED. Kernel runs without deadlock. Bug 1 (V MN-major): ✅ Fix applied. Bug 2 (softmax packing): ✅ Confirmed correct (V=I test: cosine 1.0). Bug 3 (ACCUMULATE): ✅ Fix applied. Bug 4 (non-square PV): 🔨 ROOT CAUSE IDENTIFIED — TMEM layout mismatch.

Bug 4 (CURRENT): PV MMA Broken for (128,64) Output — ROOT CAUSE IDENTIFIED

Root Cause: The (128,64) PV MMA's A-fragment reads P from TMEM with a different layout than the softmax packing writes it.

The softmax packing writes P using the QK C-fragment layout (MMA atom = (128,128,16), N_MMA=128). The PV MMA reads P using its A-fragment layout (MMA atom = (128,64,16), N_MMA=64). These two layouts produce different physical TMEM addresses for the same logical (m,k) coordinate.

Evidence:

• Truncated identity V (64,128) MN-major: O[m,d] ≈ P[m, 2d] — the MMA reads every other column of P
• All-ones V: cosine 0.999999 ✅ (uniform data hides the layout mismatch)
• Single-element V: cosine 1.0 ✅ (sparse data also hides it)
• (128,128) PV with same softmax packing: cosine 0.999999 ✅ (N_MMA=128 matches QK, no mismatch)

C++ TMEM Fragment Layout (from mma_traits_sm100.hpp):

// For M_MMA = 128, N_MMA varies with the MMA atom's N dimension
Layout tmem_atom = Layout<Shape <_128, Int<N_MMA>>,
                          Stride<  _1,     _128>>{};

• QK C-fragment: N_MMA=128 → 128 TMEM columns, stride 128
• PV A-fragment (128,64): N_MMA=64 → 64 TMEM columns, stride 128

When the softmax packing writes P at tmem_p0_offset using the QK C-fragment layout (N_MMA=128), P's (m,k) elements land at TMEM address m + 128*k. But the PV A-fragment (N_MMA=64) reads the same TMEM region as if P were stored with N_MMA=64, so it interprets the data with stride 64 instead of 128, causing the every-other-column effect (O[m,d] ≈ P[m, 2d]).

Fix (not yet applied): The softmax packing must write P using the PV MMA's A-fragment layout, not the QK C-fragment layout. FMHA does this correctly because its softmax writes P using a composition that matches the PV A-fragment — the tStS_P layout is derived from tStS.layout (QK C-fragment) but the TMEM store uses a C-fragment composition that's based on the PV MMA's tiling. The key is that FMHA's tilePlikeFP32 computation adapts the packing width to match the PV output N.

Additional fix: epi_tile must be computed from PV cta_tile, not QK cta_tile. Using QK's cta_tile for the epilogue produces epi_tile=(128,128) which is wrong for a (128,64) output. Computing from PV's cta_tile gives epi_tile=(128:1, 32:1). This fix alone improved cosine from 0.01 to 0.876, but the TMEM layout mismatch remains.

V SMEM Layouts (confirmed correct):

• PV(128,64) V SMEM: outer=((64,16),1,8,1):((1,64),0,1024,0), inner=S<3,4,3>
• PV(128,128) V SMEM: outer=(((64,2),16),1,8,1):(((1,8192),64),0,1024,0), inner=S<3,4,3>

---

Bug 1: V B-Operand Must Be MN-Major — ✅ FIX APPLIED

V must be shaped (head_dim, seq) = (64, 128) with strides (1, 64) — MN-major. PV MMA uses v_major (OperandMajorMode.MN) instead of b_major (K).

V must use as_strided — default PyTorch (64,128) gives strides (128,1) which is K-major.

Bug 2: C-Fragment Composition Store — ✅ CONFIRMED CORRECT

FP32→BF16 packing via C-fragment composition store (FMHA pattern) is correct. Proven by V=I test (cosine 1.0) and random V 128x128 test (cosine 0.999999).

⛔ FOOTGUN: St32x32bOp MUST use Float32, NOT BFloat16. ⚠️ The recast view for P packing uses the LOAD layout (128 BF16 elements), not the store composition shape.

Bug 3: First PV Must Use ACCUMULATE=False — ✅ FIX APPLIED

If ACCUMULATE=True on the first PV, O = P@V + old_O adds uninitialized TMEM. Always ACCUMULATE=False for first PV, then True for subsequent tiles.

---

🔨 Stage C: Online Softmax — AFTER B

Per the pseudocode: epilogue warps compute per-row tile_max, rescale, exp, store P back to TMEM.

🔨 Stage D: FP8 Paged KV Gather — AFTER C

Replace BF16 TMA load with FP8 paged KV gather + per-position dequant.

---

Pipeline Deadlock — ✅ FIXED (May 21)

v20-v25 all deadlocked on GPU. Three root causes found and fixed:

Fix 1: PipelineUmmaAsync for mma_si Must NOT Pass cta_layout_vmnk

FMHA's mma_s0/mma_s1 PipelineUmmaAsync calls do NOT pass cta_layout_vmnk. Removing it fixes the deadlock.

Fix 2: TMA Warp Must NOT Call tmem.wait_for_alloc()

The tmem allocation barrier has num_threads = 32 * (mma_warp + epilogue_warps). The TMA warp is NOT part of this barrier. Calling wait_for_alloc() from the TMA warp corrupts the barrier.

Fix 3: PipelineTmaStore (not TmaStorePipeline)

pipeline.TmaStorePipeline does not exist. The correct name is pipeline.PipelineTmaStore.

---

⛔ DEAD TEST: test_stage_b_v21.py — DELETED, DO NOT RECREATE

v21 attempted both Bug 1 and Bug 2 fixes in a hand-rolled pipeline kernel. It deadlocks on GPU. Root cause: pipeline synchronization mismatch. Do not recreate. Write from scratch using fmha.py as the reference.

---

⛔ FOOTGUNS — CUTLASS CuTeDSL Landmines

1. St32x32bOp with 16-bit dtype → ILLEGAL MEMORY ACCESS

St32x32bOp(Repetition(N), BFloat16) crashes at runtime. You MUST use St32x32bOp(Repetition(N), Float32) and pack 2×16-bit values into 1×Float32 backing words via cute.recast_ptr. The 16-bit type only appears in the recast view, never in the store atom itself.

2. V B-Operand Major Mode ≠ K Major Mode

FMHA requires v_major_mode == OperandMajorMode.MN. Passing K's K-major mode for V is WRONG. V must be shaped (head_dim, seq) with strides (1, head_dim) to produce MN-major. Standard PyTorch row-major (seq, head_dim) gives K-major.

3. CuTe Nested Layout Modes Flatten Sequentially

A layout like ((128,16),1,(4,2)):((65536,1),0,(16,64)) looks "non-sequential" but flattens to addr = m*65536 + k when k = k0 + 16k1 + 64k2 (CuTe row-major order). Do NOT assume nested modes imply non-sequential physical addressing. The C-fragment composition and A-fragment alias the same TMEM columns — BUT ONLY WHEN N_MMA MATCHES (i.e., (128,128) PV). For (128,64) PV, N_MMA=64 and the alias breaks.

4. PipelineUmmaAsync Consumer Group = Thread Count, NOT Warp Count

# WRONG: consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 4)
# CORRECT: consumer_group=pipeline.CooperativeGroup(pipeline.Agent.Thread, 32 * len(warp_ids))

5. PipelineUmmaAsync for mma_si Must NOT Pass cta_layout_vmnk

Passing cta_layout_vmnk to the mma_si PipelineUmmaAsync causes deadlock. FMHA does not pass it. Remove it.

6. TMA Warp Must NOT Call tmem.wait_for_alloc()

The tmem allocation barrier only includes MMA + epilogue warps. The TMA warp is excluded. Calling wait_for_alloc() from the TMA warp corrupts the barrier.

7. PV MMA ACCUMULATE Must Be False on First Tile

If ACCUMULATE=True on the first PV MMA, O = P@V + old_O adds uninitialized TMEM to the result. Always set ACCUMULATE=False for the first PV, then True for subsequent tiles. FMHA: pv_tiled_mma.set(tcgen05.Field.ACCUMULATE, kphase_idx != 0).

8. TMEM Pointer Arithmetic: Offset Units Depend on Pointer Type

When computing PV A-fragment offset from the softmax P offset:

# Softmax store: FP32 pointer + tmem_p0_offset (in FP32 elements)
tStS_P = cute.make_tensor(tStS.iterator + tmem_p0_offset, tStS_P_layout)

# PV A-fragment: BF16 pointer + scaled offset (in BF16 elements)
p_offset = acc_dtype.width // q_dtype.width * tmem_p0_offset  # 2 * 32 = 64
tOrP0 = cute.make_tensor(tOrP.iterator + p_offset, tOrP.layout)

Both must address the same physical TMEM column. The 2× scaling accounts for FP32→BF16 element size difference.

9. C-Fragment → A-Fragment TMEM Alias Only Works When N_MMA Matches

The softmax packing writes P using the QK C-fragment layout. The PV A-fragment reads P. These alias correctly ONLY when both MMA atoms have the same N_MMA (i.e., both (128,128,16) → N_MMA=128). When the PV MMA uses (128,64,16) → N_MMA=64, the A-fragment has a different TMEM stride and reads garbage. The softmax packing must be adapted to write P in the PV A-fragment's layout.

10. epi_tile Must Match PV Output Shape, Not QK

compute_epilogue_tile_shape must use PV's cta_tile_shape_mnk, not QK's. Also, self.cta_tile_shape_mnk must be set to PV's cta tile before calling epilogue_tma_store (it reads gemm_kernel.cta_tile_shape_mnk internally). FMHA sets self.epi_tile = self.pv_mma_tiler[:2] directly.

---

Architecture: Per-Tile Flow

For each KV tile:
  1. Load warp writes sKV[stage] (paged FP8 gather via indexed cp.async)
  2. MMA warp issues MMA1: sQ @ sKV[stage]^T → tmem_scores (accumulate=False)
     Signals scores_full_mbar (via PipelineUmmaAsync commit)
  3. Epilogue warps wait on mma_si consumer (scores ready), then:
     a. tcgen05.ld scores from TMEM → register fragments
     b. Compute tile_max, new_max, rescale = exp(old_max - new_max)
     c. Apply rescale to tmem_output IN PLACE (tmem_output *= rescale)
     d. tcgen05.st exp(scores - new_max) back to TMEM → P operand (via C-fragment composition)
     e. Release mma_si (softmax_done — MMA warp can re-acquire and issue PV MMA)
  4. MMA warp waits on mma_si acquire (softmax done), MMA2: P @ sV → tmem_output (accumulate=True)
  5. Stage released, load warp can refill it

After all tiles: epilogue warps tcgen05.ld tmem_output, divide by row_sum, cast to BF16, store to GMEM

---

Test Results

File	Description	Cosine	Status
test_stage_a_v2.py	Q@K^T only	0.999999	✅ PASS
test_mma_si_only.py	Q@K^T + mma_si pipeline (no PV)	0.999999	✅ PASS
test_softmax_only.py	Q@K^T + softmax packing, output S	0.52	❌ S overwritten by P (expected)
test_mma_si_pv.py	Q@K^T + softmax + P@V (V MN-major, 128x64)	0.01	❌ PV output garbage
test_pv_diag.py	Q@K^T + softmax + P@V (V=I 128x128)	1.0	✅ PASS
test_pv_diag.py	Q@K^T + softmax + P@V (random V 128x128)	0.999999	✅ PASS
test_diag_v_truncid.py	Q@K^T + softmax + P@V (trunc identity 64x128, epi from PV)	0.02	❌ O[m,d]≈P[m,2d] — TMEM alias mismatch
test_diag_v_ones.py	All-ones V (64x128)	0.999999	✅ uniform data hides mismatch
test_diag_v_ones.py	Single-element V (64x128)	1.0	✅ sparse data hides mismatch
test_diag_layout.py	(128,64) PV with epi from PV cta_tile	0.876	❌ partial fix — epi correct, TMEM alias still broken
test_diag_smem_layout.py	Print V SMEM layouts for (128,64) vs (128,128)	N/A	ℹ️ layouts confirmed correct
test_layout_compare.py	Print TMEM layouts for QK S and PV A-fragment	N/A	ℹ️ layout inspection

---

Critical APIs & Lessons

TMEM offset arithmetic

• find_tmem_tensor_col_offset(fragment) — returns physical TMEM column count
• QK accumulator: 128 TMEM columns
• A-fragment offset: acc_dtype.width // q_dtype.width * tmem_p0_offset (F32/BF16=2)

pv_mma_tiler — FMHA Convention

pv_mma_tiler = (qk_mma_tiler[0], qk_mma_tiler[2], qk_mma_tiler[1])
# = (M, head_dim, QK_N) = (128, 64, 128) for head_dim=64

FMHA passes pv_mma_tiler[:2] = (128, head_dim) to make_trivial_tiled_mma, NOT the QK tiler (128, 128).

make_trivial_tiled_mma — Use New Overload

make_trivial_tiled_mma(a_dtype, b_dtype, a_leading_mode, b_leading_mode,
                        acc_dtype, cta_group, mma_tiler_mn, a_source=SMEM)

3D tensors required

Tensors must be 3D (M, K, L) for cute.local_tile — add L=1 dimension.

Other APIs

1. cutlass_torch.from_dlpack(t).mark_layout_dynamic(leading_dim=...) — CuTe tensor from PyTorch
2. PipelineTmaUmma.create(...).make_participants() — returns (producer, consumer) pair
3. utils.gemm.sm100.epilogue_tma_store — handles transform + partition/dcopy. DO NOT hand-roll.
4. smem.allocate_tensor() — for SMEM tensors
5. LayoutEnum.from_tensor(a).mma_major_mode() — major mode from cute tensor

Environment

• Server: root@45.76.247.107 (B200, 180 GiB HBM3e per GPU)
• venv: source /root/dsv4-nvfp4-workspace/venv/bin/activate
• PYTHONPATH: /root/dsv4-nvfp4-workspace/kernel
• Model: /root/nvidia-meeting/DeepSeek-V4-Pro-NVFP4
• vLLM repo: /root/dsv4-nvfp4-workspace/vllm (modified for Blackwell)
• Pseudocode: /root/fragile-kernel-example/README.md
• fmha.py reference: /root/cutlass/examples/python/CuTeDSL/cute/blackwell/kernel/attention/fmha/fmha.py
• fmha_bwd.py reference: /root/cutlass/examples/python/CuTeDSL/cute/blackwell/kernel/attention/fmha/fmha_bwd.py