Fix identity diag: same 8D TMA indexing fix

This commit is contained in:
2026-05-22 21:21:52 +00:00
parent bb92af5b0c
commit 27116110ab

View File

@@ -138,7 +138,7 @@ class FmhaV3Diag:
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))
tVsV,tVgV = cpasync.tma_partition(tma_v,0,b_lay,cute.group_modes(sV,0,3),cute.group_modes(tCgV,0,3))
tAgQ = tAgQ[(None,0,None,0)]; tBgK = tBgK[(None,None,0,0)]; tVgV = tVgV[(None,0,None,0)]
tAgQ = tAgQ[(None,0,None,0)]; tBgK = tBgK[(None,None,None,None,None,None,None,None)]; tVgV = tVgV[(None,None,None,None,None,None,None,None)]
tCrQ = qk_mma.make_fragment_A(sQ); tCrK = qk_mma.make_fragment_B(sK)
tCrV = pv_mma.make_fragment_B(sV)
@@ -171,8 +171,8 @@ class FmhaV3Diag:
kv_coord = Int32(0 + 0)
for kt in cutlass.range(self.n_kv_tiles, unroll=1):
kvh = kvp.acquire_and_advance(pk)
cute.copy(tma_k, tBgK[(None, kv_coord)], tBsK[(None, kvh.index)], tma_bar_ptr=kvh.barrier)
cute.copy(tma_v, tVgV[(None, kv_coord)], tVsV[(None, kvh.index)], tma_bar_ptr=kvh.barrier)
cute.copy(tma_k, tBgK[None, None, None, None, kv_coord, None, None, None], tBsK[(None, kvh.index)], tma_bar_ptr=kvh.barrier)
cute.copy(tma_v, tVgV[None, None, None, None, kv_coord, None, None, None], tVsV[(None, kvh.index)], tma_bar_ptr=kvh.barrier)
kv_coord += 1
pk = cutlass.Boolean(1)
kvp.tail()