豆豆友情提示:这是一个非官方 GitHub 代理镜像,主要用于网络测试或访问加速。请勿在此进行登录、注册或处理任何敏感信息。进行这些操作请务必访问官方网站 github.com。 Raw 内容也通过此代理提供。
Skip to content

fix: k_grouped_fp8_gemm_nt_contiguous crashes with n = 768 on H100#238

Merged
zheanxu merged 1 commit intodeepseek-ai:mainfrom
xinqiu:fix/k_grouped_fp8
Feb 25, 2026
Merged

fix: k_grouped_fp8_gemm_nt_contiguous crashes with n = 768 on H100#238
zheanxu merged 1 commit intodeepseek-ai:mainfrom
xinqiu:fix/k_grouped_fp8

Conversation

@xinqiu
Copy link
Copy Markdown
Contributor

@xinqiu xinqiu commented Dec 5, 2025

image

@cailun01
Copy link
Copy Markdown

cailun01 commented Feb 12, 2026

Hi @xinqiu , I’m also encountering the issue that this PR addresses. Do you have an estimate on when this might be merged into the main branch? Thanks for the hard work! cc @LyricZhao @zheanxu

@zheanxu zheanxu merged commit 35c4bc8 into deepseek-ai:main Feb 25, 2026
@zheanxu
Copy link
Copy Markdown
Collaborator

zheanxu commented Feb 25, 2026

@cailun01, apologies for the delayed response. This PR has been merged, and the issue will be addressed with a better solution in the next major release.

@Rowtion
Copy link
Copy Markdown

Rowtion commented Feb 25, 2026

@zheanxu Thanks for the update.When is the next major release?

LyricZhao pushed a commit that referenced this pull request Apr 16, 2026
…t stage (#238)

Co-authored-by: Chenhao Xu <xch@deepseek.com>
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.

4 participants