From c29c9f4e0205d20b82cc3504535c1d82f050abeb Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Daniele=20Trifir=C3=B2?= Date: Wed, 25 Sep 2024 16:55:11 +0200 Subject: [PATCH 1/3] Dockerfile.rocm.ubi: move microdnf update in base stage --- Dockerfile.rocm.ubi | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/Dockerfile.rocm.ubi b/Dockerfile.rocm.ubi index 17909b97f50b2..39093f4630e5c 100644 --- a/Dockerfile.rocm.ubi +++ b/Dockerfile.rocm.ubi @@ -13,6 +13,7 @@ ENV VIRTUAL_ENV=/opt/vllm ENV PATH="$VIRTUAL_ENV/bin:$PATH" RUN --mount=type=cache,target=/root/.cache/pip \ + microdnf -y update && \ microdnf install -y --setopt=install_weak_deps=0 --nodocs \ python${PYTHON_VERSION}-devel \ python${PYTHON_VERSION}-pip \ @@ -40,8 +41,7 @@ gpgcheck=1\n\ gpgkey=https://repo.radeon.com/rocm/rocm.gpg.key" > /etc/yum.repos.d/amdgpu.repo -RUN microdnf -y update && \ - microdnf -y install \ +RUN microdnf -y install \ rocm-hip-libraries rocm-hip-runtime \ miopen-hip && \ microdnf clean all From 69ac6c1e66f9f032059cbb732cef4bf088142eb1 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Daniele=20Trifir=C3=B2?= Date: Wed, 25 Sep 2024 16:55:34 +0200 Subject: [PATCH 2/3] Dockerfile.rocm.ubi: bump torch version to 2.5.0.dev20240912+rocm6.1 --- Dockerfile.rocm.ubi | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/Dockerfile.rocm.ubi b/Dockerfile.rocm.ubi index 39093f4630e5c..d6383a30eb40e 100644 --- a/Dockerfile.rocm.ubi +++ b/Dockerfile.rocm.ubi @@ -49,9 +49,8 @@ RUN microdnf -y install \ RUN --mount=type=cache,target=/root/.cache/pip \ --mount=type=cache,target=/root/.cache/uv \ uv pip install -v --index-url "https://download.pytorch.org/whl/nightly/rocm6.1" \ - torch==2.5.0.dev20240726+rocm6.1 \ - torchvision==0.20.0.dev20240726+rocm6.1 - + torch==2.5.0.dev20240912+rocm6.1 \ + torchvision==0.20.0.dev20240912+rocm6.1 FROM rocm_base as rocm_devel From 399c11499e666be94a954ece49323f644860b2e9 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Daniele=20Trifir=C3=B2?= Date: Wed, 25 Sep 2024 16:56:13 +0200 Subject: [PATCH 3/3] Dockerfile.rocm.ubi: get rid of build triton stage this is a torch dependency when installed from the pytorch/rocm6.1 index: https://download.pytorch.org/whl/nightly/rocm6.1 --- Dockerfile.rocm.ubi | 27 --------------------------- 1 file changed, 27 deletions(-) diff --git a/Dockerfile.rocm.ubi b/Dockerfile.rocm.ubi index d6383a30eb40e..12a1da4ca3181 100644 --- a/Dockerfile.rocm.ubi +++ b/Dockerfile.rocm.ubi @@ -126,31 +126,6 @@ RUN --mount=type=cache,target=/root/.cache/uv \ ################################################################################################## -FROM rocm_devel AS build_triton - -# Whether to build triton on rocm -ARG BUILD_TRITON="0" -ARG TRITON_BRANCH="e0fc12c" - -# Build triton wheel if `BUILD_TRITON` is set to 1 -RUN --mount=type=cache,target=/root/.cache/ccache \ - --mount=type=cache,target=/root/.cache/pip \ - --mount=type=cache,target=/root/.cache/uv \ - if [ "$BUILD_TRITON" = "1" ]; then \ - mkdir -p /libs && cd /libs && \ - git clone https://github.com/OpenAI/triton.git && \ - cd triton && \ - git checkout "${TRITON_BRANCH}" && \ - cd python && \ - uv pip install -v cmake ninja && \ - python3 setup.py bdist_wheel --dist-dir=/install; \ - else \ - mkdir -p /install && \ - wget -P /install "https://download.pytorch.org/whl/pytorch_triton_rocm-3.0.0-cp312-cp312-linux_x86_64.whl#sha256=650ce7a6624ecbbe90c30fccb472a57772c08bc6f12312246f7af584f275b7ca"; \ - fi - -################################################################################################## - FROM rocm_devel AS build_vllm ARG PYTORCH_ROCM_ARCH ARG MAX_JOBS @@ -188,7 +163,6 @@ RUN microdnf install -y --setopt=install_weak_deps=0 --nodocs gcc && \ microdnf clean all RUN --mount=type=bind,from=build_amdsmi,src=/install,target=/install/amdsmi/ \ - --mount=type=bind,from=build_triton,src=/install,target=/install/triton \ --mount=type=bind,from=build_flashattention,src=/install,target=/install/flashattention \ --mount=type=bind,from=build_vllm,src=/workspace/dist,target=/install/vllm/ \ --mount=type=cache,target=/root/.cache/pip \ @@ -197,7 +171,6 @@ RUN --mount=type=bind,from=build_amdsmi,src=/install,target=/install/amdsmi/ \ --index-strategy=unsafe-best-match \ --extra-index-url "https://download.pytorch.org/whl/nightly/rocm6.1" \ /install/amdsmi/*.whl\ - /install/triton/*.whl\ /install/flashattention/*.whl\ /install/vllm/*.whl