fix: add SM120 (RTX PRO 6000 Blackwell) support (#124)#146
Open
yurekami wants to merge 3 commits into
Open
Conversation
Changed `flash_mla_with_kvcache_sm90` to `flash_mla_with_kvcache` in get_mla_metadata docstring to match the actual function name. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
CUDA 13 moved cuda/std/utility and other standard library headers to CCCL (CUDA C++ Core Library). This adds the CUDA include path explicitly to resolve build errors on CUDA 13+ ARM64 systems. Fixes deepseek-ai#121 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add get_gpu_arch() function to detect GPU compute capability - Add FLASH_MLA_DISABLE_SM120 environment variable - Generate SM120 arch flags when NVCC 12.9+ is available - Auto-detect SM120 GPUs and log detection message 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
|
Does this PR indicate that the kernel actually works on sm120, after the compilation flags added properly by this PR? |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Add support for SM120 architecture (RTX PRO 6000 Blackwell workstation GPUs).
Changes
get_gpu_arch()function to detect GPU compute capability via nvidia-smiFLASH_MLA_DISABLE_SM120environment variable to control SM120 compilationTest plan
🤖 Generated with Claude Code