[Deps] Upgrade CUDA PyTorch stack to 2.13 (#28836)
Co-authored-by: Brayden Zhong <b8zhong@uwaterloo.ca>
This commit is contained in:
Mohammad Miadh Angkad
co-authored by
Brayden Zhong
Brayden Zhong
parent
4ad990ba7d
commit
434e646282
+21
-27
@@ -1,4 +1,4 @@
|
||||
ARG CUDA_VERSION=13.0.1
|
||||
ARG CUDA_VERSION=13.0.3
|
||||
FROM nvidia/cuda:${CUDA_VERSION}-cudnn-devel-ubuntu24.04 AS base
|
||||
|
||||
ARG TARGETARCH
|
||||
@@ -10,9 +10,9 @@ ARG HOPPER_SBO=0
|
||||
ARG HOPPER_SBO_DEEPEP_COMMIT=9f2fc4b3182a51044ae7ecb6610f7c9c3258c4d6
|
||||
ARG DEEPEP_COMMIT=9af0e0d0e74f3577af1979c9b9e1ac2cad0104ee
|
||||
ARG BUILD_AND_DOWNLOAD_PARALLEL=8
|
||||
ARG SGL_KERNEL_VERSION=0.4.5
|
||||
ARG SGL_KERNEL_VERSION=0.4.6.post1
|
||||
ARG SGL_VERSION
|
||||
ARG SGL_DEEP_GEMM_VERSION=0.1.5.post1
|
||||
ARG SGL_DEEP_GEMM_VERSION=0.1.5.post2
|
||||
ARG USE_LATEST_SGLANG=0
|
||||
ARG GDRCOPY_VERSION=2.5.1
|
||||
ARG PIP_DEFAULT_INDEX
|
||||
@@ -194,19 +194,18 @@ RUN curl --proto '=https' --tlsv1.2 --retry 3 --retry-delay 2 -sSf https://sh.ru
|
||||
RUN --mount=type=cache,target=/root/.cache/pip \
|
||||
python3 -m pip install --upgrade pip setuptools wheel html5lib six \
|
||||
&& case "$CUDA_VERSION" in \
|
||||
12.6.1) CUINDEX=126 ;; \
|
||||
12.8.1) CUINDEX=128 ;; \
|
||||
12.9.1) CUINDEX=129 ;; \
|
||||
13.0.1) CUINDEX=130 ;; \
|
||||
12.6.3) CUINDEX=126 ;; \
|
||||
12.9.2) CUINDEX=129 ;; \
|
||||
13.0.3) CUINDEX=130 ;; \
|
||||
*) echo "Unsupported CUDA version: $CUDA_VERSION" && exit 1 ;; \
|
||||
esac \
|
||||
&& if [ "$CUDA_VERSION" = "12.6.1" ]; then \
|
||||
&& if [ "$CUDA_VERSION" = "12.6.3" ]; then \
|
||||
python3 -m pip install https://${GITHUB_ARTIFACTORY}/sgl-project/whl/releases/download/v${SGL_KERNEL_VERSION}/sglang_kernel-${SGL_KERNEL_VERSION}+cu124-cp310-abi3-manylinux2014_$(uname -m).whl --force-reinstall --no-deps \
|
||||
; \
|
||||
elif [ "$CUDA_VERSION" = "12.8.1" ] || [ "$CUDA_VERSION" = "12.9.1" ]; then \
|
||||
python3 -m pip install https://github.com/sgl-project/whl/releases/download/v${SGL_KERNEL_VERSION}/sglang_kernel-${SGL_KERNEL_VERSION}+cu129-cp310-abi3-manylinux2014_$(uname -m).whl --force-reinstall --no-deps \
|
||||
elif [ "$CUDA_VERSION" = "12.9.2" ]; then \
|
||||
python3 -m pip install https://${GITHUB_ARTIFACTORY}/sgl-project/whl/releases/download/v${SGL_KERNEL_VERSION}/sglang_kernel-${SGL_KERNEL_VERSION}+cu129-cp310-abi3-manylinux2014_$(uname -m).whl --force-reinstall --no-deps \
|
||||
; \
|
||||
elif [ "$CUDA_VERSION" = "13.0.1" ]; then \
|
||||
elif [ "$CUDA_VERSION" = "13.0.3" ]; then \
|
||||
# --no-deps prevents pip from pulling torch from default PyPI
|
||||
python3 -m pip install sglang-kernel==${SGL_KERNEL_VERSION} --force-reinstall --no-deps \
|
||||
; \
|
||||
@@ -229,10 +228,9 @@ COPY proto /tmp/sglang_deps/proto
|
||||
RUN --mount=type=cache,target=/root/.cache/pip \
|
||||
--mount=type=cache,target=/root/.cargo/registry \
|
||||
case "$CUDA_VERSION" in \
|
||||
12.6.1) CUINDEX=126 ;; \
|
||||
12.8.1) CUINDEX=128 ;; \
|
||||
12.9.1) CUINDEX=129 ;; \
|
||||
13.0.1) CUINDEX=130 ;; \
|
||||
12.6.3) CUINDEX=126 ;; \
|
||||
12.9.2) CUINDEX=129 ;; \
|
||||
13.0.3) CUINDEX=130 ;; \
|
||||
*) echo "Unsupported CUDA version: $CUDA_VERSION" && exit 1 ;; \
|
||||
esac \
|
||||
&& cd /tmp/sglang_deps/python \
|
||||
@@ -251,7 +249,7 @@ RUN --mount=type=cache,target=/root/.cache/pip \
|
||||
pip list --format=freeze | awk -F'==' '/-cu13(==|$)/ {print $1}' \
|
||||
| xargs -r python3 -m pip uninstall -y && \
|
||||
python3 -m pip install --index-url https://download.pytorch.org/whl/cu${CUINDEX} \
|
||||
torch==2.11.0 torchvision==0.26.0 torchaudio==2.11.0 --force-reinstall; \
|
||||
torch==2.13.0 torchvision==0.28.0 torchaudio==2.11.0 --force-reinstall; \
|
||||
python3 -m pip install https://github.com/sgl-project/whl/releases/download/v${SGL_DEEP_GEMM_VERSION}/sgl_deep_gemm-${SGL_DEEP_GEMM_VERSION}+cu129-py3-none-manylinux2014_$(uname -m).whl --force-reinstall; \
|
||||
fi \
|
||||
&& cd /sgl-workspace \
|
||||
@@ -315,13 +313,10 @@ RUN set -eux; \
|
||||
RUN --mount=type=cache,target=/root/.cache/pip \
|
||||
cd /build/DeepEP && \
|
||||
case "$CUDA_VERSION" in \
|
||||
12.6.1) \
|
||||
12.6.3) \
|
||||
CHOSEN_TORCH_CUDA_ARCH_LIST='9.0' \
|
||||
;; \
|
||||
12.8.1) \
|
||||
CHOSEN_TORCH_CUDA_ARCH_LIST='9.0;10.0' \
|
||||
;; \
|
||||
12.9.1|13.0.1) \
|
||||
12.9.2|13.0.3) \
|
||||
CHOSEN_TORCH_CUDA_ARCH_LIST='9.0;10.0;10.3' \
|
||||
;; \
|
||||
*) \
|
||||
@@ -370,10 +365,9 @@ ARG FLASHINFER_VERSION
|
||||
# Stage jit-cache/cubin artifacts into /flashinfer_jit_output for clean COPY later
|
||||
RUN --mount=type=cache,target=/root/.cache/pip \
|
||||
case "$CUDA_VERSION" in \
|
||||
12.6.1) CUINDEX=126 ;; \
|
||||
12.8.1) CUINDEX=128 ;; \
|
||||
12.9.1) CUINDEX=129 ;; \
|
||||
13.0.1) CUINDEX=130 ;; \
|
||||
12.6.3) CUINDEX=126 ;; \
|
||||
12.9.2) CUINDEX=129 ;; \
|
||||
13.0.3) CUINDEX=130 ;; \
|
||||
*) echo "Unsupported CUDA version: $CUDA_VERSION" && exit 1 ;; \
|
||||
esac \
|
||||
&& mkdir -p /flashinfer_jit_output \
|
||||
@@ -616,10 +610,10 @@ RUN --mount=type=cache,target=/root/.cache/pip \
|
||||
# to avoid shipping wrong-CUDA libs on cu13 images.
|
||||
RUN --mount=type=cache,target=/root/.cache/pip if [ "${CUDA_VERSION%%.*}" = "12" ]; then \
|
||||
python3 -m pip install nixl nixl-cu12 --no-deps ; \
|
||||
python3 -m pip install cuda-python==12.9 ; \
|
||||
python3 -m pip install "cuda-python>=12,<13" ; \
|
||||
elif [ "${CUDA_VERSION%%.*}" = "13" ]; then \
|
||||
python3 -m pip install nixl nixl-cu13 --no-deps ; \
|
||||
python3 -m pip install cuda-python==13.2.0 ; \
|
||||
python3 -m pip install "cuda-python>=13,<14" ; \
|
||||
fi
|
||||
|
||||
# Add yank script
|
||||
|
||||
@@ -9,7 +9,7 @@
|
||||
# 2. DeepEP patch + rebuild:
|
||||
# topk 11->16, SWITCH_HIDDEN += 3584, EP>8 SourceMeta alignment,
|
||||
# and cross-node timeout headroom; rebuilt for sm_90 and sm_100a only
|
||||
# 3. DeepGEMM upgrade to 0.1.5.post1:
|
||||
# 3. DeepGEMM upgrade to 0.1.5.post2:
|
||||
# official MegaMoE runtime-JIT header with Kimi-K3 SiTU support
|
||||
# 4. FlashInfer CuTeDSL MLA DCP patch:
|
||||
# apply the seven runtime-file diffs; exclude tests absent from the wheel
|
||||
@@ -25,7 +25,7 @@
|
||||
|
||||
FROM lmsysorg/sglang:v0.5.16-cu129 AS base
|
||||
|
||||
ARG SGL_DEEP_GEMM_VERSION="0.1.5.post1"
|
||||
ARG SGL_DEEP_GEMM_VERSION="0.1.5.post2"
|
||||
|
||||
# Current Kimi-K3 source auto-discovers and builds its PyO3 extensions.
|
||||
ARG RUST_VERSION="1.90.0"
|
||||
|
||||
@@ -10,7 +10,7 @@
|
||||
# topk 11->16, SWITCH_HIDDEN += 3584, EP>8 SourceMeta alignment,
|
||||
# cross-node timeout headroom, CUDA-13 cccl include; rebuilt for
|
||||
# sm_90, sm_100a, and sm_103a
|
||||
# 3. DeepGEMM upgrade to 0.1.5.post1:
|
||||
# 3. DeepGEMM upgrade to 0.1.5.post2:
|
||||
# official MegaMoE runtime-JIT header with Kimi-K3 SiTU support
|
||||
# 4. FlashInfer CuTeDSL MLA DCP patch:
|
||||
# apply the seven runtime-file diffs; exclude tests absent from the wheel
|
||||
@@ -26,7 +26,7 @@
|
||||
|
||||
FROM lmsysorg/sglang:v0.5.16 AS base
|
||||
|
||||
ARG SGL_DEEP_GEMM_VERSION="0.1.5.post1"
|
||||
ARG SGL_DEEP_GEMM_VERSION="0.1.5.post2"
|
||||
|
||||
# Current Kimi-K3 source auto-discovers and builds its PyO3 extensions.
|
||||
ARG RUST_VERSION="1.90.0"
|
||||
|
||||
@@ -7,7 +7,7 @@ ARG ARCH=x86_64
|
||||
ARG CUDA_VERSION=13.0
|
||||
ARG PYTHON_VERSION=3.12
|
||||
ARG PYTHON_TAG=cp312-cp312
|
||||
ARG TORCH_VER=2.11.0
|
||||
ARG TORCH_VER=2.13.0
|
||||
ARG TVM_FFI_VER=0.1.11
|
||||
ARG PIP_DEFAULT_INDEX=https://pypi.python.org/simple
|
||||
ARG PYTORCH_MIRROR=download.pytorch.org
|
||||
|
||||
Reference in New Issue
Block a user