Skip to content

[Build] add SM120 for wheel build#2873

Merged
ApostaC merged 1 commit intoLMCache:devfrom
deng451e:add_sm120_build
Mar 27, 2026
Merged

[Build] add SM120 for wheel build#2873
ApostaC merged 1 commit intoLMCache:devfrom
deng451e:add_sm120_build

Conversation

@deng451e
Copy link
Copy Markdown
Collaborator

What this PR does / why we need it:

Special notes for your reviewers:

If applicable:

  • this PR contains user facing changes - docs added
  • this PR contains unit tests

Signed-off-by: deng451e <838677410@qq.com>
@gemini-code-assist
Copy link
Copy Markdown
Contributor

Summary of Changes

Hello, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request expands the supported CUDA architectures for wheel builds by incorporating CUDA architecture 12.0, corresponding to NVIDIA's Blackwell generation (SM120). This ensures that builds can leverage the latest GPU hardware, updating both the Docker build configurations and the accompanying documentation to reflect this change.

Highlights

  • CUDA Architecture Support: Added support for CUDA architecture 12.0 (SM120) to the build process, enabling compatibility with NVIDIA's Blackwell generation GPUs.
  • Dockerfile Updates: Modified docker/Dockerfile and docker/Dockerfile.standalone to include 12.0 in the torch_cuda_arch_list build argument.
  • Documentation Update: Updated docker/README.md to reflect the expanded torch_cuda_arch_list with the newly added CUDA architecture 12.0.
Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature Command Description
Code Review /gemini review Performs a code review for the current pull request in its current state.
Pull Request Summary /gemini summary Provides a summary of the current pull request in its current state.
Comment @gemini-code-assist Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help /gemini help Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here.

Footnotes

  1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution.

Copy link
Copy Markdown
Contributor

@gemini-code-assist gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

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

Code Review

This pull request updates the torch_cuda_arch_list in the Dockerfiles and README.md to include 12.0 as a CUDA compute capability. The reviewer notes that 12.0 is not a publicly known NVIDIA compute capability, suggesting this is a typo that could lead to build failures. Consequently, the updated documentation in README.md is also inaccurate. The reviewer recommends correcting the architecture list to 10.0+PTX and updating the documentation to reflect the accurate supported CUDA architectures, while also ensuring correct markdown formatting.

Comment thread docker/Dockerfile

# CUDA arch list used by torch
ARG torch_cuda_arch_list='7.0 7.5 8.0 8.6 8.9 9.0 10.0+PTX'
ARG torch_cuda_arch_list='7.0 7.5 8.0 8.6 8.9 9.0 10.0 12.0+PTX'
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

high

The CUDA compute capability 12.0 is being added, presumably for SM120 as per the PR title. However, 12.0 is not a publicly known NVIDIA compute capability. The latest announced architecture is Blackwell, which has a compute capability of 10.0.

This appears to be a typo and will likely cause the build to fail. If the intention was to ensure support for Blackwell (10.0), the previous value 10.0+PTX was already sufficient, as it builds SASS for 10.0 and PTX for forward compatibility.

ARG torch_cuda_arch_list='7.0 7.5 8.0 8.6 8.9 9.0 10.0+PTX'


# CUDA arch list used by torch
ARG torch_cuda_arch_list='7.0 7.5 8.0 8.6 8.9 9.0 10.0+PTX'
ARG torch_cuda_arch_list='7.0 7.5 8.0 8.6 8.9 9.0 10.0 12.0+PTX'
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

high

The CUDA compute capability 12.0 is being added, presumably for SM120 as per the PR title. However, 12.0 is not a publicly known NVIDIA compute capability. The latest announced architecture is Blackwell, which has a compute capability of 10.0.

This appears to be a typo and will likely cause the build to fail. If the intention was to ensure support for Blackwell (10.0), the previous value 10.0+PTX was already sufficient, as it builds SASS for 10.0 and PTX for forward compatibility.

ARG torch_cuda_arch_list='7.0 7.5 8.0 8.6 8.9 9.0 10.0+PTX'

Comment thread docker/README.md
| `max_jobs` | `2` | Max parallel jobs for build |
| `nvcc_threads` | `8` | Number of nvcc threads |
| `torch_cuda_arch_list` | `7.0 7.5 8.0 8.6 8.9 9.0 10.0+PTX` | CUDA architectures |
| `torch_cuda_arch_list` | `7.0 7.5 8.0 8.6 8.9 9.0 10.0 12.0+PTX` | CUDA architectures |
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

high

This documentation is being updated to include the 12.0 compute capability. As this value is likely a typo (see comments on the Dockerfiles), this documentation update would be inaccurate. Please update this line to reflect the correct list of supported CUDA architectures once confirmed. When applying the suggestion, please ensure the argument name and its value are enclosed in backticks (`) for correct markdown formatting.

Suggested change
| `torch_cuda_arch_list` | `7.0 7.5 8.0 8.6 8.9 9.0 10.0 12.0+PTX` | CUDA architectures |
| torch_cuda_arch_list | 7.0 7.5 8.0 8.6 8.9 9.0 10.0+PTX | CUDA architectures |

Copy link
Copy Markdown
Contributor

@ApostaC ApostaC left a comment

Choose a reason for hiding this comment

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

LGTM!

@ApostaC ApostaC enabled auto-merge (squash) March 27, 2026 22:35
@ApostaC ApostaC added the full Run comprehensive tests on this PR label Mar 27, 2026
@ApostaC ApostaC merged commit 07ec1f9 into LMCache:dev Mar 27, 2026
34 checks passed
jooho-XCENA pushed a commit to xcena-dev/LMCache that referenced this pull request Apr 2, 2026
add SM120 for wheel build

Signed-off-by: deng451e <838677410@qq.com>
jooho-XCENA pushed a commit to xcena-dev/LMCache that referenced this pull request Apr 2, 2026
add SM120 for wheel build

Signed-off-by: deng451e <838677410@qq.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

full Run comprehensive tests on this PR

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants