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 2, 03:30 UTC
[CI/Build] Bump flashinfer to v0.6.12 (#44036)
Signed-off-by: Vadim Gimpelson <vadim.gimpelson@gmail.com> Co-authored-by: Jee Jee Li <pandaleefree@gmail.com>
Vadim Gimpelson committed
Jun 3, 2026 at 02:19 UTC
8b3b71ee9db164dd236f20250e5b34e84b75de80
4 files changed
+6
-6
docker/Dockerfile
+1
-1
@@ -757,7 +757,7 @@ RUN --mount=type=cache,target=/opt/uv/cache \
757
# Install FlashInfer JIT cache (requires CUDA-version-specific index URL)
758
# https://docs.flashinfer.ai/installation.html
759
# From versions.json: .flashinfer.version
760
-ARG FLASHINFER_VERSION=0.6.11.post2
760
+ARG FLASHINFER_VERSION=0.6.12
761
RUN --mount=type=cache,target=/opt/uv/cache \
762
uv pip install --system flashinfer-jit-cache==${FLASHINFER_VERSION} \
763
--extra-index-url https://flashinfer.ai/whl/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.')
docker/Dockerfile.nightly_torch
+2
-2
@@ -256,13 +256,13 @@ RUN pip install setuptools==75.6.0 packaging==23.2 ninja==1.11.1.3 build==1.2.2.
256
257
258
# build flashinfer for torch nightly from source around 10 mins
259
-# release version: v0.6.11.post2
259
+# release version: v0.6.12
260
# todo(elainewy): cache flashinfer build result for faster build
261
ENV CCACHE_DIR=/root/.cache/ccache
262
RUN --mount=type=cache,target=/root/.cache/ccache \
263
--mount=type=cache,target=/root/.cache/uv \
264
echo "git clone flashinfer..." \
265
- && git clone --depth 1 --branch v0.6.11.post2 --recursive https://github.com/flashinfer-ai/flashinfer.git \
265
+ && git clone --depth 1 --branch v0.6.12 --recursive https://github.com/flashinfer-ai/flashinfer.git \
266
&& cd flashinfer \
267
&& git submodule update --init --recursive \
268
&& echo "finish git clone flashinfer..." \
docker/versions.json
+1
-1
@@ -68,7 +68,7 @@
68
"default": "true"
69
},
70
"FLASHINFER_VERSION": {
71
- "default": "0.6.11.post2"
71
+ "default": "0.6.12"
72
},
73
"GDRCOPY_CUDA_VERSION": {
74
"default": "12.8"
requirements/cuda.txt
+2
-2
@@ -9,8 +9,8 @@ torchaudio==2.11.0
9
# These must be updated alongside torch
10
torchvision==0.26.0 # Required for phi3v processor. See https://github.com/pytorch/vision?tab=readme-ov-file#installation for corresponding version
11
# FlashInfer should be updated together with the Dockerfile
12
-flashinfer-python==0.6.11.post2
13
-flashinfer-cubin==0.6.11.post2
12
+flashinfer-python==0.6.12
13
+flashinfer-cubin==0.6.12
14
apache-tvm-ffi==0.1.9
15
tilelang==0.1.9
16
# Cap nvidia-cudnn-frontend (transitive dep of flashinfer) due to