software-engineering/spark-vllm-docker
2
0
Fork 0
Code Issues Actions Packages Projects Releases Wiki Activity

Added cubins and jit-cache

Browse Source
This commit is contained in:
Eugene Rakhmatulin
2026-01-29 11:42:04 -08:00
parent 36e3b7af27
commit b58ba7b19a
1 changed files with 10 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
Dockerfile.mxfp4
View File

@@ -98,8 +98,8 @@ ARG CUTLASS_REPO=https://github.com/christopherowen/cutlass.git
ARG FLASHINFER_SHA=f349e52496a72a00d8c4ac02c7a1e38523ff7194
ARG CUTLASS_SHA=11af7f02ab52c9130e422eeb4b44042fbd60c083
# RUN --mount=type=cache,id=uv-cache,target=/root/.cache/uv \
# uv pip install "apache-tvm-ffi<0.2" nvidia-ml-py requests wheel
RUN --mount=type=cache,id=uv-cache,target=/root/.cache/uv \
uv pip install nvidia-nvshmem-cu13
# Clone FlashInfer (cached for faster rebuilds)
RUN --mount=type=cache,id=git-flashinfer,target=/git-cache/flashinfer \
@@ -147,15 +147,15 @@ RUN --mount=type=cache,id=uv-cache,target=/root/.cache/uv \
uv pip install --no-build-isolation . -v
# flashinfer-cubin
# RUN --mount=type=cache,id=uv-cache,target=/root/.cache/uv \
# --mount=type=cache,id=ccache,target=/root/.ccache \
# cd flashinfer-cubin && uv pip install --no-build-isolation . -v
RUN --mount=type=cache,id=uv-cache,target=/root/.cache/uv \
--mount=type=cache,id=ccache,target=/root/.ccache \
cd flashinfer-cubin && uv pip install --no-build-isolation . -v
# # flashinfer-jit-cache
# RUN --mount=type=cache,id=uv-cache,target=/root/.cache/uv \
# --mount=type=cache,id=ccache,target=/root/.ccache \
# cd flashinfer-jit-cache && \
# uv pip install --no-build-isolation . -v
# flashinfer-jit-cache
RUN --mount=type=cache,id=uv-cache,target=/root/.cache/uv \
--mount=type=cache,id=ccache,target=/root/.ccache \
cd flashinfer-jit-cache && \
uv pip install --no-build-isolation . -v
# --- VLLM SOURCE CACHE BUSTER ---
# Change THIS argument to force a fresh git clone and rebuild of vLLM