Skip to content

[ISSUE-262, 85] Unfix the Compute Capabilities across CMakeLists, NVCC Compiler Options#921

Open
stardriftfx wants to merge 5 commits intoSharedDevelopmentfrom
262-unfix-the-compute-capabilities-across-cmakelists
Open

[ISSUE-262, 85] Unfix the Compute Capabilities across CMakeLists, NVCC Compiler Options#921
stardriftfx wants to merge 5 commits intoSharedDevelopmentfrom
262-unfix-the-compute-capabilities-across-cmakelists

Conversation

@stardriftfx
Copy link
Contributor

Closes #262, #85

Description

  • updated nvcc compile flag to compile on local machine, if target not user-specified
  • added wiki page on GPU Arch levels and their differences

Checklist (Mandatory for new features)

  • Added Documentation

Testing (Mandatory for all changes)

  • GPU Test: test-medium-connected.xml Passed
  • GPU Test: test-large-long.xml Passed

…t user-specified, documentation

- wiki page on GPU Arch levels and their differences
@stardriftfx stardriftfx requested a review from stiber March 16, 2026 20:19
@stardriftfx stardriftfx self-assigned this Mar 16, 2026
Copilot AI review requested due to automatic review settings March 16, 2026 20:19
Copy link
Contributor

Copilot AI left a comment

Choose a reason for hiding this comment

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

Pull request overview

Updates Graphitti’s GPU build configuration to make CUDA architecture targeting more flexible, and adds developer documentation describing GPU architecture / compute capability levels.

Changes:

  • Change default TARGET_ARCH behavior to prefer a local-machine (“native”) target when not specified.
  • Update NVCC gencode flag handling (real + virtual/PTX) in the CUDA-enabled CMake path.
  • Add docs/Developer/GPUArchLevels.md documenting supported GPU architectures and recommended TARGET_ARCH values.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 8 comments.

File Description
CMakeLists.txt Adjusts CUDA architecture selection defaults and NVCC compiler flag configuration.
docs/Developer/GPUArchLevels.md Adds developer-facing reference on GPU architecture levels and target selection.

Comment on lines 38 to 41
if(NOT DEFINED TARGET_ARCH)
set(TARGET_ARCH 37)
# If target not specified by user, use local machine's GPU architecture
set(TARGET_ARCH "native")
endif()
Copy link
Contributor Author

Choose a reason for hiding this comment

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

Tests passed successfully though.

Copy link
Contributor

Choose a reason for hiding this comment

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

This generates worrisome warnings that don't exist previously:

CMake Warning (dev) in CMakeLists.txt:
  Policy CMP0104 is not set: CMAKE_CUDA_ARCHITECTURES now detected for NVCC,
  empty CUDA_ARCHITECTURES not allowed.  Run "cmake --help-policy CMP0104"
  for policy details.  Use the cmake_policy command to set the policy and
  suppress this warning.

  CUDA_ARCHITECTURES is empty for target "Core".
This warning is for project developers.  Use -Wno-dev to suppress it.```

stardriftfx and others added 2 commits March 17, 2026 15:05
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
@@ -0,0 +1,41 @@
# GPU Architecture Levels
Copy link
Contributor

Choose a reason for hiding this comment

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

This doesn't appear to be linked in to the documentation.

Comment on lines 38 to 41
if(NOT DEFINED TARGET_ARCH)
set(TARGET_ARCH 37)
# If target not specified by user, use local machine's GPU architecture
set(TARGET_ARCH "native")
endif()
Copy link
Contributor

Choose a reason for hiding this comment

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

This generates worrisome warnings that don't exist previously:

CMake Warning (dev) in CMakeLists.txt:
  Policy CMP0104 is not set: CMAKE_CUDA_ARCHITECTURES now detected for NVCC,
  empty CUDA_ARCHITECTURES not allowed.  Run "cmake --help-policy CMP0104"
  for policy details.  Use the cmake_policy command to set the policy and
  suppress this warning.

  CUDA_ARCHITECTURES is empty for target "Core".
This warning is for project developers.  Use -Wno-dev to suppress it.```

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Unfix the Compute Capabilities across CMakeLists Look into nvcc compiler options

3 participants