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
Bump Flashinfer to 0.3.1 (#24868)
Signed-off-by: bbartels <benjamin@bartels.dev>
Benjamin Bartels committed
Sep 15, 2025 at 20:45 UTC
94b03f88dd4d69ccf65e52439f4d2dacff97862f
2 files changed
+2
-2
docker/Dockerfile
+1
-1
@@ -375,7 +375,7 @@ RUN --mount=type=bind,from=build,src=/workspace/dist,target=/vllm-workspace/dist
375
# Install FlashInfer from source
376
ARG FLASHINFER_GIT_REPO="https://github.com/flashinfer-ai/flashinfer.git"
377
# Keep this in sync with "flashinfer" extra in setup.py
378
-ARG FLASHINFER_GIT_REF="v0.3.0"
378
+ARG FLASHINFER_GIT_REF="v0.3.1"
379
# Flag to control whether to compile FlashInfer AOT kernels
380
# Set to "true" to enable AOT compilation:
381
# docker build --build-arg FLASHINFER_AOT_COMPILE=true ...
setup.py
+1
-1
@@ -664,7 +664,7 @@ setup(
664
"mistral_common[audio]"], # Required for audio processing
665
"video": [], # Kept for backwards compatibility
666
# FlashInfer should be updated together with the Dockerfile
667
- "flashinfer": ["flashinfer-python==0.3.0"],
667
+ "flashinfer": ["flashinfer-python==0.3.1"],
668
# Optional deps for AMD FP4 quantization support
669
"petit-kernel": ["petit-kernel"],
670
},