hailin
/
sglang_v0.5.2
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

.

This commit is contained in:
hailin 2025-09-20 22:27:00 +08:00
parent c757ea02fc
commit ae99d6a643
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
Dockerfile
View File

@ -91,12 +91,12 @@ WORKDIR /opt/flashinfer
# 覆盖你的目标算力：3090=8.6，4090=8.9，H100=9.0a；可按需增/减
ENV FLASHINFER_CUDA_ARCH_LIST=8.0,8.6,8.9
ENV FLASHINFER_CUDA_ARCH_LIST="8.0 8.6 8.9"
# 先做 AOT 预编译，再直接打 wheel（不隔离，使用同一份自编 torch）
RUN python3 -m pip install --no-cache-dir numpy requests build "cuda-python>=12.0,<13" "nvidia-nvshmem-cu12" ninja pynvml && \
bash -lc 'unset TORCH_CUDA_ARCH_LIST; \
FLASHINFER_CUDA_ARCH_LIST=8.0,8.6,8.9 python3 -m flashinfer.aot' && \
FLASHINFER_CUDA_ARCH_LIST="8.0 8.6 8.9" python3 -m flashinfer.aot' && \
python3 -m build --no-isolation --wheel && \
ls -lh dist/ && \
python3 -m pip install --no-cache-dir --no-deps dist/*.whl