A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 14, 05:30 UTC
update to latest deepgemm for dsv3.2 (#25871)
Signed-off-by: youkaichao <youkaichao@gmail.com>
youkaichao committed
Sep 29, 2025 at 17:51 UTC
9360d34fa18fcba15a0e4dccb2ef3d781bcfba7c
2 files changed
+2
-2
docker/Dockerfile
+1
-1
@@ -456,7 +456,7 @@ RUN --mount=type=cache,target=/root/.cache/uv \
456
ARG DEEPGEMM_GIT_REF
457
COPY tools/install_deepgemm.sh /tmp/install_deepgemm.sh
458
RUN --mount=type=cache,target=/root/.cache/uv \
459
- VLLM_DOCKER_BUILD_CONTEXT=1 /tmp/install_deepgemm.sh --cuda-version "${CUDA_VERSION}" ${DEEPGEMM_GIT_REF:+--ref "$DEEPGEMM_GIT_REF"}
459
+ VLLM_DOCKER_BUILD_CONTEXT=1 TORCH_CUDA_ARCH_LIST="9.0a 10.0a" /tmp/install_deepgemm.sh --cuda-version "${CUDA_VERSION}" ${DEEPGEMM_GIT_REF:+--ref "$DEEPGEMM_GIT_REF"}
460
461
COPY tools/install_gdrcopy.sh install_gdrcopy.sh
462
RUN set -eux; \
tools/install_deepgemm.sh
+1
-1
@@ -6,7 +6,7 @@ set -e
6
7
# Default values
8
DEEPGEMM_GIT_REPO="https://github.com/deepseek-ai/DeepGEMM.git"
9
-DEEPGEMM_GIT_REF="ea9c5d9270226c5dd7a577c212e9ea385f6ef048"
9
+DEEPGEMM_GIT_REF="594953acce41793ae00a1233eb516044d604bcb6"
10
11
# Parse command line arguments
12
while [[ $# -gt 0 ]]; do