chenchenghao/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 11 Packages Projects Releases Wiki Activity

[Ascend] qwen optimization (#12078)

Browse Source 
Co-authored-by: Even Zhou <even.y.zhou@outlook.com>
This commit is contained in:
Liwansi
2025-11-25 19:44:24 +08:00
committed by GitHub
co-authored by Even Zhou
parent 0b3f002daf
commit 432ecf841e
16 changed files with 561 additions and 108 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docker/npu.Dockerfile
+9 -10
View File
@@ -10,11 +10,10 @@ ARG PIP_INDEX_URL="https://pypi.org/simple/"
ARG APTMIRROR=""
ARG PYTORCH_VERSION="2.8.0"
ARG TORCHVISION_VERSION="0.23.0"
ARG PTA_VERSION="v7.2.0-pytorch${PYTORCH_VERSION}"
ARG PTA_NAME="torch_npu-${PYTORCH_VERSION}-cp311-cp311-manylinux_2_28_aarch64.whl"
ARG PTA_URL="https://gitcode.com/Ascend/pytorch/releases/download/${PTA_VERSION}/${PTA_NAME}"
ARG TRITON_ASCEND_URL="https://sglang-ascend.obs.cn-east-3.myhuaweicloud.com/sglang/triton_ascend-3.2.0%2Bgitb0ea0850-cp311-cp311-linux_aarch64.whl"
ARG BISHENG_URL="https://sglang-ascend.obs.cn-east-3.myhuaweicloud.com/sglang/Ascend-BiSheng-toolkit_aarch64.run"
ARG PTA_URL="https://sglang-ascend.obs.cn-east-3.myhuaweicloud.com/sglang/torch_npu/torch_npu-2.8.0.post2.dev20251113-cp311-cp311-manylinux_2_28_aarch64.whl"
ARG TRITON_ASCEND_URL="https://sglang-ascend.obs.cn-east-3.myhuaweicloud.com/sglang/triton_ascend/triton_ascend-3.2.0.dev2025112116-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl"
ARG BISHENG_NAME="Ascend-BiSheng-toolkit_aarch64_20251121.run"
ARG BISHENG_URL="https://sglang-ascend.obs.cn-east-3.myhuaweicloud.com/sglang/triton_ascend/${BISHENG_NAME}"
ARG SGLANG_TAG=main
ARG ASCEND_CANN_PATH=/usr/local/Ascend/ascend-toolkit
ARG SGLANG_KERNEL_NPU_TAG=main
@@ -64,13 +63,13 @@ RUN ${PIP_INSTALL} sglang-router
### Install PyTorch and PTA
RUN (${PIP_INSTALL} torch==${PYTORCH_VERSION} torchvision==${TORCHVISION_VERSION} --index-url https://download.pytorch.org/whl/cpu) && \
(wget -O "${PTA_NAME}" "${PTA_URL}" && ${PIP_INSTALL} "./${PTA_NAME}" && rm "./${PTA_NAME}")
RUN (${PIP_INSTALL} torch==${PYTORCH_VERSION} torchvision==${TORCHVISION_VERSION} --index-url https://download.pytorch.org/whl/cpu) \
&& (${PIP_INSTALL} ${PTA_URL})
# TODO: install from pypi released triton-ascend
RUN ${PIP_INSTALL} attrs==24.2.0 numpy==1.26.4 scipy==1.13.1 decorator==5.1.1 psutil==6.0.0 pytest==8.3.2 pytest-xdist==3.6.1 pyyaml pybind11 && \
${PIP_INSTALL} ${TRITON_ASCEND_URL}
RUN (${PIP_INSTALL} pybind11) \
&& (${PIP_INSTALL} ${TRITON_ASCEND_URL})
# Install SGLang
RUN git clone https://github.com/sgl-project/sglang --branch $SGLANG_TAG && \
@@ -96,6 +95,6 @@ RUN wget https://sglang-ascend.obs.cn-east-3.myhuaweicloud.com/ops/CANN-custom_o
${PIP_INSTALL} ./custom_ops-1.0.$DEVICE_TYPE-cp311-cp311-linux_aarch64.whl
# Install Bisheng
RUN wget ${BISHENG_URL} && chmod a+x Ascend-BiSheng-toolkit_aarch64.run && ./Ascend-BiSheng-toolkit_aarch64.run --install && rm Ascend-BiSheng-toolkit_aarch64.run
RUN wget -O "${BISHENG_NAME}" "${BISHENG_URL}" && chmod a+x "${BISHENG_NAME}" && "./${BISHENG_NAME}" --install && rm "${BISHENG_NAME}"
CMD ["/bin/bash"]