Skip to content

Remove the requirement for CU_DEVICE_ATTRIBUTE_HANDLE_TYPE_FABRIC_SUPPORTED for NVLS support#489

Merged
Binyang2014 merged 7 commits intomainfrom
binyli/nvls
Mar 28, 2025
Merged

Remove the requirement for CU_DEVICE_ATTRIBUTE_HANDLE_TYPE_FABRIC_SUPPORTED for NVLS support#489
Binyang2014 merged 7 commits intomainfrom
binyli/nvls

Conversation

@Binyang2014
Copy link
Contributor

Remove the requirement for CU_DEVICE_ATTRIBUTE_HANDLE_TYPE_FABRIC_SUPPORTED for NVLS support.

Fix #487

@Binyang2014 Binyang2014 requested review from chhwang and Copilot March 28, 2025 02:24
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

This PR removes the outdated requirement for manually setting up IMEX channels (using CU_DEVICE_ATTRIBUTE_HANDLE_TYPE_FABRIC_SUPPORTED) and updates the documentation to state that GPUs with NVLS support now require kernel version 5.6 or above.

  • Removed manual IMEX channel setup instructions
  • Updated GPU support details to include a kernel version requirement
Files not reviewed (5)
  • include/mscclpp/gpu.hpp: Language not supported
  • include/mscclpp/gpu_utils.hpp: Language not supported
  • src/gpu_utils.cc: Language not supported
  • src/include/registered_memory.hpp: Language not supported
  • src/registered_memory.cc: Language not supported

@Binyang2014
Copy link
Contributor Author

/azp run

@azure-pipelines
Copy link

Azure Pipelines successfully started running 3 pipeline(s).

@chhwang chhwang linked an issue Mar 28, 2025 that may be closed by this pull request
@Binyang2014
Copy link
Contributor Author

/azp run

@azure-pipelines
Copy link

Azure Pipelines successfully started running 3 pipeline(s).

@Binyang2014 Binyang2014 merged commit a3d8d68 into main Mar 28, 2025
25 checks passed
@Binyang2014 Binyang2014 deleted the binyli/nvls branch March 28, 2025 23:46
@Binyang2014 Binyang2014 mentioned this pull request Apr 3, 2025
6 tasks
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.

[Feature] Why CU_DEVICE_ATTRIBUTE_HANDLE_TYPE_FABRIC_SUPPORTED is required for nvls? Problem with mcsslpp QuickStart

3 participants