Skip to content

fix(CuTeDSL): correct FP4 tensor K dimension in grouped blockscaled GEMM #3102

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Hale423 wants to merge 1 commit into
base: main
Choose a base branch
from
+5 −2
Open

fix(CuTeDSL): correct FP4 tensor K dimension in grouped blockscaled GEMM #3102

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 5 additions & 2 deletions examples/python/CuTeDSL/blackwell/grouped_blockscaled_gemm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2490,6 +2490,9 @@ def create_tensors_abc_for_all_groups(
strides_abc = []
ptrs_abc = []

# FP4 packing: 2 elements per byte, so the K storage dimension is halved
k_fct = 2 if ab_dtype == cutlass.Float4E2M1FN else 1

# Iterate through all groups and create tensors for each group
for group_idx, (m, n, k, l) in enumerate(problem_sizes_mnkl):
# Create tensors A, B, C
Expand All @@ -2499,15 +2502,15 @@ def create_tensors_abc_for_all_groups(
cute_tensor_a,
ref_torch_fp32_tensor_a,
stride_mk_a,
) = create_tensor_and_stride(l, m, k, a_major == "m", ab_dtype)
) = create_tensor_and_stride(l, m, k // k_fct, a_major == "m", ab_dtype)

(
ptr_b,
torch_tensor_b,
cute_tensor_b,
ref_torch_fp32_tensor_b,
stride_nk_b,
) = create_tensor_and_stride(l, n, k, b_major == "n", ab_dtype)
) = create_tensor_and_stride(l, n, k // k_fct, b_major == "n", ab_dtype)

(
ptr_c,
Expand Down