Skip to content

[Unity][CUTLASS][Cherry-pick] Skip profiling all conv2d output alignments when possible #15583

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

Merged
merged 1 commit into from
Aug 17, 2023
Merged

[Unity][CUTLASS][Cherry-pick] Skip profiling all conv2d output alignments when possible #15583

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: 6 additions & 1 deletion python/tvm/contrib/cutlass/gen_conv2d.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -285,6 +285,11 @@ def alignments(dtype):
raise ValueError("Unsupported data type: %s" % dtype)
return alignments

alignments_c = [align for align in alignments(out_dtype) if OC % align == 0]

if not profile_all_alignments:
alignments_c = [alignments_c[0]]

ops = GENERATOR_FUNC_TABLE[self.sm](
out_dtype,
data_dtype,
Expand All @@ -294,7 +299,7 @@ def alignments(dtype):
conv_kind,
stride_support,
split_k_slices,
[align for align in alignments(out_dtype) if OC % align == 0],
alignments_c,
),
lambda align: all([dim % align == 0 for dim in [IC]]),
use_3xtf32,
Expand Down