diff --git a/.github/workflows/accuracy_test.yaml b/.github/workflows/accuracy_test.yaml index 999fb6ad5d..0297f625d0 100644 --- a/.github/workflows/accuracy_test.yaml +++ b/.github/workflows/accuracy_test.yaml @@ -173,6 +173,8 @@ jobs: - name: Install vllm-project/vllm-ascend working-directory: ./vllm-ascend + env: + PIP_EXTRA_INDEX_URL: https://mirrors.huaweicloud.com/ascend/repos/pypi run: | pip install -r requirements-dev.txt pip install -e . diff --git a/.github/workflows/image_openeuler.yml b/.github/workflows/image_openeuler.yml index 0d585e690f..690d8145cd 100644 --- a/.github/workflows/image_openeuler.yml +++ b/.github/workflows/image_openeuler.yml @@ -19,6 +19,12 @@ on: - '.github/workflows/image_openeuler.yml' - 'Dockerfile.openEuler' - 'vllm_ascend/**' + - 'setup.py' + - 'pyproject.toml' + - 'requirements.txt' + - 'cmake/**' + - 'CMakeLists.txt' + - 'csrc/**' push: # Publish image when tagging, the Dockerfile in tag will be build as tag image branches: diff --git a/.github/workflows/image_ubuntu.yml b/.github/workflows/image_ubuntu.yml index 63dadc64a7..a2cfbcefb0 100644 --- a/.github/workflows/image_ubuntu.yml +++ b/.github/workflows/image_ubuntu.yml @@ -19,6 +19,12 @@ on: - '.github/workflows/image_ubuntu.yml' - 'Dockerfile' - 'vllm_ascend/**' + - 'setup.py' + - 'pyproject.toml' + - 'requirements.txt' + - 'cmake/**' + - 'CMakeLists.txt' + - 'csrc/**' push: # Publish image when tagging, the Dockerfile in tag will be build as tag image branches: diff --git a/.github/workflows/nightly_benchmarks.yaml b/.github/workflows/nightly_benchmarks.yaml index b060076c9c..6e03c1e6be 100644 --- a/.github/workflows/nightly_benchmarks.yaml +++ b/.github/workflows/nightly_benchmarks.yaml @@ -115,6 +115,8 @@ jobs: VLLM_TARGET_DEVICE=empty pip install -e . - name: Install vllm-project/vllm-ascend + env: + PIP_EXTRA_INDEX_URL: https://mirrors.huaweicloud.com/ascend/repos/pypi run: | pip install -e . pip install -r benchmarks/requirements-bench.txt diff --git a/.github/workflows/vllm_ascend_test.yaml b/.github/workflows/vllm_ascend_test.yaml index 62c91de8cc..540680dd2f 100644 --- a/.github/workflows/vllm_ascend_test.yaml +++ b/.github/workflows/vllm_ascend_test.yaml @@ -151,6 +151,7 @@ jobs: - name: Install vllm-project/vllm-ascend run: | + export PIP_EXTRA_INDEX_URL=https://mirrors.huaweicloud.com/ascend/repos/pypi export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/Ascend/ascend-toolkit/latest/x86_64-linux/devlib python3 -m pip install -r requirements-dev.txt --extra-index https://download.pytorch.org/whl/cpu/ python3 -m pip install -v . --extra-index https://download.pytorch.org/whl/cpu/ @@ -214,6 +215,8 @@ jobs: VLLM_TARGET_DEVICE=empty pip install -e . - name: Install vllm-project/vllm-ascend + env: + PIP_EXTRA_INDEX_URL: https://mirrors.huaweicloud.com/ascend/repos/pypi run: | pip install -r requirements-dev.txt pip install -v -e . @@ -311,6 +314,8 @@ jobs: VLLM_TARGET_DEVICE=empty pip install -e . - name: Install vllm-project/vllm-ascend + env: + PIP_EXTRA_INDEX_URL: https://mirrors.huaweicloud.com/ascend/repos/pypi run: | pip install -r requirements-dev.txt pip install -v -e . diff --git a/.github/workflows/vllm_ascend_test_long_term.yaml b/.github/workflows/vllm_ascend_test_long_term.yaml index 46123590ab..e249849e19 100644 --- a/.github/workflows/vllm_ascend_test_long_term.yaml +++ b/.github/workflows/vllm_ascend_test_long_term.yaml @@ -88,6 +88,8 @@ jobs: VLLM_TARGET_DEVICE=empty pip install -e . - name: Install vllm-project/vllm-ascend + env: + PIP_EXTRA_INDEX_URL: https://mirrors.huaweicloud.com/ascend/repos/pypi run: | pip install -r requirements-dev.txt pip install -v -e . diff --git a/.github/workflows/vllm_ascend_test_pd.yaml b/.github/workflows/vllm_ascend_test_pd.yaml index c2c76c9f23..932b3e59b3 100644 --- a/.github/workflows/vllm_ascend_test_pd.yaml +++ b/.github/workflows/vllm_ascend_test_pd.yaml @@ -97,6 +97,8 @@ jobs: VLLM_TARGET_DEVICE=empty pip install -e . - name: Install vllm-project/vllm-ascend + env: + PIP_EXTRA_INDEX_URL: https://mirrors.huaweicloud.com/ascend/repos/pypi run: | pip install -r requirements-dev.txt pip install -v -e . diff --git a/Dockerfile b/Dockerfile index 952e77fe92..e6ec009d28 100644 --- a/Dockerfile +++ b/Dockerfile @@ -46,7 +46,8 @@ RUN VLLM_TARGET_DEVICE="empty" python3 -m pip install -v -e /vllm-workspace/vllm # Install vllm-ascend # Append `libascend_hal.so` path (devlib) to LD_LIBRARY_PATH -RUN source /usr/local/Ascend/ascend-toolkit/set_env.sh && \ +RUN export PIP_EXTRA_INDEX_URL=https://mirrors.huaweicloud.com/ascend/repos/pypi && \ + source /usr/local/Ascend/ascend-toolkit/set_env.sh && \ source /usr/local/Ascend/nnal/atb/set_env.sh && \ export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/Ascend/ascend-toolkit/latest/`uname -i`-linux/devlib && \ python3 -m pip install -v -e /vllm-workspace/vllm-ascend/ --extra-index https://download.pytorch.org/whl/cpu/ && \ diff --git a/Dockerfile.openEuler b/Dockerfile.openEuler index 2ff3d0b39a..4e414e04b7 100644 --- a/Dockerfile.openEuler +++ b/Dockerfile.openEuler @@ -43,7 +43,8 @@ RUN VLLM_TARGET_DEVICE="empty" python3 -m pip install -e /vllm-workspace/vllm/ - python3 -m pip cache purge # Install vllm-ascend -RUN source /usr/local/Ascend/ascend-toolkit/set_env.sh && \ +RUN export PIP_EXTRA_INDEX_URL=https://mirrors.huaweicloud.com/ascend/repos/pypi && \ + source /usr/local/Ascend/ascend-toolkit/set_env.sh && \ source /usr/local/Ascend/nnal/atb/set_env.sh && \ export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/Ascend/ascend-toolkit/latest/`uname -i`-linux/devlib && \ python3 -m pip install -v -e /vllm-workspace/vllm-ascend/ --extra-index https://download.pytorch.org/whl/cpu/ && \ diff --git a/README.md b/README.md index f3f9296c1f..7d0966c8d4 100644 --- a/README.md +++ b/README.md @@ -38,7 +38,7 @@ By using vLLM Ascend plugin, popular open-source models, including Transformer-l - Software: * Python >= 3.9, < 3.12 * CANN >= 8.1.RC1 - * PyTorch >= 2.5.1, torch-npu >= 2.5.1 + * PyTorch >= 2.5.1, torch-npu >= 2.5.1.post1.dev20250528 * vLLM (the same version as vllm-ascend) ## Getting Started diff --git a/README.zh.md b/README.zh.md index 6adb0e1832..2d2062a8b4 100644 --- a/README.zh.md +++ b/README.zh.md @@ -39,7 +39,7 @@ vLLM 昇腾插件 (`vllm-ascend`) 是一个由社区维护的让vLLM在Ascend NP - 软件： * Python >= 3.9, < 3.12 * CANN >= 8.1.RC1 - * PyTorch >= 2.5.1, torch-npu >= 2.5.1 + * PyTorch >= 2.5.1, torch-npu >= 2.5.1.post1.dev20250528 * vLLM (与vllm-ascend版本一致) ## 开始使用 diff --git a/docs/source/installation.md b/docs/source/installation.md index c9f6844915..79e4a88dba 100644 --- a/docs/source/installation.md +++ b/docs/source/installation.md @@ -9,11 +9,11 @@ This document describes how to install vllm-ascend manually. - A hardware with Ascend NPU. It's usually the Atlas 800 A2 series. - Software: - | Software | Supported version | Note | - |-----------|-------------------|----------------------------------------| - | CANN | >= 8.1.RC1 | Required for vllm-ascend and torch-npu | - | torch-npu | >= 2.5.1 | Required for vllm-ascend | - | torch | >= 2.5.1 | Required for torch-npu and vllm | + | Software | Supported version | Note | + |---------------|----------------------------------|-------------------------------------------| + | CANN | >= 8.1.RC1 | Required for vllm-ascend and torch-npu | + | torch-npu | >= 2.5.1.post1.dev20250528 | Required for vllm-ascend | + | torch | >= 2.5.1 | Required for torch-npu and vllm | You have 2 way to install: - **Using pip**: first prepare env manually or via CANN image, then install `vllm-ascend` using pip. @@ -156,6 +156,7 @@ cd .. # Install vLLM Ascend git clone --depth 1 --branch |vllm_ascend_version| https://github.com/vllm-project/vllm-ascend.git cd vllm-ascend +export PIP_EXTRA_INDEX_URL=https://mirrors.huaweicloud.com/ascend/repos/pypi pip install -v -e . cd .. ``` diff --git a/pyproject.toml b/pyproject.toml index b44197038b..514b755c32 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -12,7 +12,7 @@ requires = [ "scipy", "setuptools>=64", "setuptools-scm>=8", - "torch-npu==2.5.1", + "torch-npu==2.5.1.post1.dev20250528", "torch>=2.5.1", "torchvision<0.21.0", "wheel", diff --git a/requirements.txt b/requirements.txt index 2f844df74a..eadb96f1e9 100644 --- a/requirements.txt +++ b/requirements.txt @@ -10,7 +10,6 @@ pyyaml scipy setuptools>=64 setuptools-scm>=8 -torch-npu==2.5.1 torch>=2.5.1 torchvision<0.21.0 wheel @@ -21,3 +20,8 @@ quart # Required for N-gram speculative decoding numba + +# Install torch_npu +--pre +--extra-index-url https://mirrors.huaweicloud.com/ascend/repos/pypi +torch-npu==2.5.1.post1.dev20250528 diff --git a/setup.py b/setup.py index 631e55c1b6..327c2f8791 100644 --- a/setup.py +++ b/setup.py @@ -152,7 +152,7 @@ def configure(self, ext: CMakeExtension) -> None: # if pybind11 is installed via pip pybind11_cmake_path = (subprocess.check_output( [python_executable, "-m", "pybind11", - "--cmake"]).decode().strip()) + "--cmakedir"]).decode().strip()) except subprocess.CalledProcessError as e: # else specify pybind11 path installed from source code on CI container raise RuntimeError(f"CMake configuration failed: {e}")