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 15, 08:30 UTC
Use prebuilt FlashInfer x86_64 PyTorch 2.7 CUDA 12.8 wheel for CI (#18537)
Signed-off-by: Huy Do <huydhn@gmail.com>
Huy Do committed
May 23, 2025 at 14:17 UTC
1645b601960a62e466b9fc6e98f50d465989c52e
1 file changed
+8
-9
docker/Dockerfile
+8
-9
@@ -257,18 +257,17 @@ RUN --mount=type=bind,from=build,src=/workspace/dist,target=/vllm-workspace/dist
257
RUN --mount=type=cache,target=/root/.cache/uv \
258
. /etc/environment && \
259
if [ "$TARGETPLATFORM" != "linux/arm64" ]; then \
260
- # uv pip install --system https://github.com/flashinfer-ai/flashinfer/releases/download/v0.2.4/flashinfer_python-0.2.4+cu124torch2.6-cp38-abi3-linux_x86_64.whl ; \
261
- # TESTING: install FlashInfer from source to test 2.7.0 final RC
260
+ # FlashInfer alreary has a wheel for PyTorch 2.7.0 and CUDA 12.8. This is enough for CI use
261
if [[ "$CUDA_VERSION" == 12.8* ]]; then \
263
- export TORCH_CUDA_ARCH_LIST='7.5 8.0 8.9 9.0 10.0+PTX'; \
262
+ uv pip install --system https://download.pytorch.org/whl/cu128/flashinfer/flashinfer_python-0.2.5%2Bcu128torch2.7-cp38-abi3-linux_x86_64.whl; \
263
else \
264
export TORCH_CUDA_ARCH_LIST='7.5 8.0 8.9 9.0+PTX'; \
266
- fi; \
267
- CUDA_MAJOR="${CUDA_VERSION%%.*}"; \
268
- if [ "$CUDA_MAJOR" -lt 12 ]; then \
269
- export FLASHINFER_ENABLE_SM90=0; \
270
- fi; \
271
- uv pip install --system --no-build-isolation "git+https://github.com/flashinfer-ai/flashinfer@21ea1d2545f74782b91eb8c08fd503ac4c0743fc" ; \
265
+ CUDA_MAJOR="${CUDA_VERSION%%.*}"; \
266
+ if [ "$CUDA_MAJOR" -lt 12 ]; then \
267
+ export FLASHINFER_ENABLE_SM90=0; \
268
+ fi; \
269
+ uv pip install --system --no-build-isolation "git+https://github.com/flashinfer-ai/flashinfer@21ea1d2545f74782b91eb8c08fd503ac4c0743fc" ; \
270
+ fi \
271
fi
272
COPY examples examples
273
COPY benchmarks benchmarks