Skip to content

[Feature] support blackwell gemm in ht #7053

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
heavengate merged 6 commits into from
Apr 7, 2026
Merged

[Feature] support blackwell gemm in ht #7053

Show file tree
Hide file tree
Changes from 5 commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
30b152c
[Feature] support blackwell gemm in ht
lizhenyun01 Mar 27, 2026
3a8450c
[Feature] support ops for convert
lizhenyun01 Mar 27, 2026
2dfa2e8
fix cuda error 716
lizhenyun01 Mar 31, 2026
c0949b8
fix cuda error
lizhenyun01 Apr 1, 2026
f67be54
opt memory
lizhenyun01 Apr 2, 2026
181436a
remove unused code
lizhenyun01 Apr 2, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions fastdeploy/envs.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -76,6 +76,8 @@ def _validate_split_kv_size(value: int) -> int:
"FD_PD_CHANGEABLE": lambda: os.getenv("FD_PD_CHANGEABLE", "0"),
# Whether to use DeepGemm for FP8 blockwise MoE.
"FD_USE_DEEP_GEMM": lambda: bool(int(os.getenv("FD_USE_DEEP_GEMM", "0"))),
# Whether to use DeepGemm for FP8 blockwise MoE.
Copy link
Copy Markdown

@PaddlePaddle-bot PaddlePaddle-bot Apr 2, 2026

Choose a reason for hiding this comment

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

🟡 建议 注释复制粘贴错误

当前注释 # Whether to use DeepGemm for FP8 blockwise MoE. 与上一行 FD_USE_DEEP_GEMM 的注释完全相同，应该描述 Blackwell GEMM 的用途。

建议修改为：

# Whether to use Blackwell GEMM for FP8 blockwise MoE (SM100+).
"FD_USE_BLACKWELL_GEMM": lambda: bool(int(os.getenv("FD_USE_BLACKWELL_GEMM", "0"))),

"FD_USE_BLACKWELL_GEMM": lambda: bool(int(os.getenv("FD_USE_BLACKWELL_GEMM", "0"))),
# Whether to use PFCCLab/DeepEP.
"FD_USE_PFCC_DEEP_EP": lambda: bool(int(os.getenv("FD_USE_PFCC_DEEP_EP", "0"))),
# Whether to use aggregate send.
Expand Down
Loading
Loading