Skip to content

Enhancement- Add SM80/90 architecture to CUTLASS builds for CUDA 12.8+ for A100 #785

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
polarG wants to merge 3 commits into
base: main
Choose a base branch
from
+4 −4
Open

Enhancement- Add SM80/90 architecture to CUTLASS builds for CUDA 12.8+ for A100 #785

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
e0bb811
Add SM80 architecture to CUTLASS builds for CUDA 12.8+ and 12.9+
Mar 11, 2026
d56fb5c
Add arch support for MSCCL.
Mar 17, 2026
b8cfc30
Apply suggestion from @Copilot
polarG Mar 17, 2026
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
8 changes: 4 additions & 4 deletions third_party/Makefile
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -39,14 +39,14 @@ sb_micro_path:
mkdir -p $(SB_MICRO_PATH)/lib

# Build cutlass.
# for cuda 12.9 and later Build from commit v3.9 (3.9 release commit) for blackwell support
# For CUDA 12.9 and later, build from CUTLASS v4.1.0 (required for Blackwell support).
cuda_cutlass:
ifeq ($(shell echo $(CUDA_VER)">=12.9" | bc -l), 1)
$(eval ARCHS := "100;103")
$(eval ARCHS := "80;90;100;103")
if [ -d cutlass ]; then rm -rf cutlass; fi
git clone --branch v4.1.0 --depth 1 https://github.com/NVIDIA/cutlass.git && cd cutlass
else ifeq ($(shell echo $(CUDA_VER)">=12.8" | bc -l), 1)
$(eval ARCHS := "90;100")
$(eval ARCHS := "80;90;100")
if [ -d cutlass ]; then rm -rf cutlass; fi
git clone --branch v3.9.2 --depth 1 https://github.com/NVIDIA/cutlass.git && cd cutlass
Comment on lines 46 to 51
Copy link
Copy Markdown
Contributor

@guoshzhao guoshzhao Apr 13, 2026 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@polarG Please address the comment from Copilot.

else ifeq ($(shell echo $(CUDA_VER)">=11.8" | bc -l), 1)
Expand Down Expand Up @@ -270,7 +270,7 @@ apex_rocm:
# Build MSCCL for CUDA
cuda_msccl: sb_micro_path
ifeq ($(shell echo $(CUDA_VER)">=12.9" | bc -l), 1)
$(eval ARCHS := 100 103)
$(eval ARCHS := 80 90 100 103)
if [ -d msccl ]; then rm -rf msccl; fi; \
git clone --single-branch --branch main https://github.com/Azure/msccl.git \
&& git -C msccl checkout 87048bd && git -C msccl submodule update --recursive --init
Expand Down
Loading