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
Upgrade flashinfer to 0.3.1 (#24470)
Signed-off-by: Lu Fang <lufang@fb.com> Co-authored-by: Michael Goin <mgoin64@gmail.com>
Lu Fang committed
Sep 15, 2025 at 19:36 UTC
0af3ce13558056b3637dd4661c89f4b037faa85a
1 file changed
+2
-2
docker/Dockerfile.nightly_torch
+2
-2
@@ -246,7 +246,7 @@ RUN pip install setuptools==75.6.0 packaging==23.2 ninja==1.11.1.3 build==1.2.2.
246
247
248
# build flashinfer for torch nightly from source around 10 mins
249
-# release version: v0.2.2.post1
249
+# release version: v0.3.1
250
# todo(elainewy): cache flashinfer build result for faster build
251
ENV CCACHE_DIR=/root/.cache/ccache
252
RUN --mount=type=cache,target=/root/.cache/ccache \
@@ -254,7 +254,7 @@ RUN --mount=type=cache,target=/root/.cache/ccache \
254
echo "git clone flashinfer..." \
255
&& git clone --recursive https://github.com/flashinfer-ai/flashinfer.git \
256
&& cd flashinfer \
257
- && git checkout v0.2.2.post1 \
257
+ && git checkout v0.3.1 \
258
&& git submodule update --init --recursive \
259
&& echo "finish git clone flashinfer..." \
260
&& rm -rf build \