Skip to content

fix: isCuMemMapAllocated crashes on non-NVLS systems even with MSCCLPP_FORCE_DISABLE_NVLS=true#790

Draft
Copilot wants to merge 3 commits intomainfrom
copilot/fix-runtimeerror-on-perlmutter-gpus
Draft

fix: isCuMemMapAllocated crashes on non-NVLS systems even with MSCCLPP_FORCE_DISABLE_NVLS=true#790
Copilot wants to merge 3 commits intomainfrom
copilot/fix-runtimeerror-on-perlmutter-gpus

Conversation

Copy link
Copy Markdown
Contributor

Copilot AI commented Apr 20, 2026

  • Fix isCuMemMapAllocated() to just return true/false without throwing when NVLS is not supported
  • Fix isNvlsSupported() caching bug where result/isChecked were never updated
  • Restore [[maybe_unused]] on result and isChecked statics — needed in HIP/ROCm env where CUDA_NVLS_API_AVAILABLE is not defined and the variables would otherwise be unused
  • Run linter (./tools/lint.sh)

…sSupported caching

Agent-Logs-Url: https://github.com/microsoft/mscclpp/sessions/f8417c19-e697-4a32-a298-9a958b9a16a1

Co-authored-by: Binyang2014 <9415966+Binyang2014@users.noreply.github.com>
Copilot AI changed the title [WIP] Fix RuntimeError for MSCCLPP on Perlmutter A100 GPUs fix: isCuMemMapAllocated crashes on non-NVLS systems even with MSCCLPP_FORCE_DISABLE_NVLS=true Apr 20, 2026
Copilot AI requested a review from Binyang2014 April 20, 2026 00:39
Comment thread src/core/gpu_utils.cc Outdated
Comment thread src/core/gpu_utils.cc
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.

"RuntimeError: cuMemMap is used in env without NVLS support (mscclpp failure: InvalidUsage)" Error

2 participants