leavelet/cutlass
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

v4.4 tag release update. (#3032)

Browse Source
This commit is contained in:
Junkai-Wu
2026-02-14 12:27:58 +08:00
committed by GitHub
parent 01687cfba1
commit d4bbf728ca
140 changed files with 41624 additions and 3691 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
examples/python/CuTeDSL/experimental/blackwell/dense_gemm_ptr_array.py
View File

@@ -198,10 +198,7 @@ class DenseGemmPtrArrayKernel:
)
# UMMA ACC TMEM Layout
acc_shape = tiled_mma.partition_shape_C(mnk_tiler[:2])
tmem_layout = tiled_mma.make_fragment_C(
cute.append(acc_shape, acc_stage)
).layout
tmem_layout = cute_ext.make_tmem_layout_acc(tiled_mma, mnk_tiler, acc_stage)
# Allocate UMMA Buffers
bufferA = cute_ext.allocate(
@@ -251,11 +248,9 @@ class DenseGemmPtrArrayKernel:
tiled_copy_t2r = cute.nvgpu.tcgen05.make_tmem_copy(copy_atom_t2r, acc_epi_div)
# Calculate the per thread destination size per iteration for output of TMEM and input of SMEM
thr_copy_t2r = tiled_copy_t2r.get_slice(tid_x)
gC_mnl_epi = cute.flat_divide(gD_tile, epi_tile)
tTR_gC = thr_copy_t2r.partition_D(gC_mnl_epi)
acc_d_rmem_layout = cute.make_fragment_like(
tTR_gC[(None, None, None, 0, 0)].layout
acc_d_rmem_layout = cute_ext.make_t2r_rmem_layout(
tiled_copy_t2r, gC_mnl_epi, tid_x
)
# Allocate RMEM buffers