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 16, 22:00 UTC
Move release wheel env var to Dockerfile instead (#6163)
Simon Mo committed
Jul 5, 2024 at 17:19 UTC
bc96d5c330e079fa501eee05e97bf15009c9a094
2 files changed
+3
-2
.buildkite/release-pipeline.yaml
+2
-2
@@ -1,13 +1,13 @@
1
steps:
2
- block: "Build wheels"
3
4
- - label: "Build wheel - Python {{matrix.python_version}}, CUDA {{matrix.cuda_version}}"
4
+ - label: "Build wheel - Python {{matrix.python_version}}, CUDA {{matrix.cuda_version}}"
5
agents:
6
queue: cpu_queue
7
commands:
8
- "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg USE_SCCACHE=1 --build-arg CUDA_VERSION={{matrix.cuda_version}} --build-arg PYTHON_VERSION={{matrix.python_version}} --tag vllm-ci:build-image --target build --progress plain ."
9
- "mkdir artifacts"
10
- - "docker run --rm -v $(pwd)/artifacts:/artifacts_host -e CMAKE_BUILD_TYPE=Release vllm-ci:build-image cp -r dist /artifacts_host"
10
+ - "docker run --rm -v $(pwd)/artifacts:/artifacts_host vllm-ci:build-image cp -r dist /artifacts_host"
11
- "aws s3 cp --recursive artifacts/dist s3://vllm-wheels/$BUILDKITE_COMMIT/"
12
matrix:
13
setup:
Dockerfile
+1
@@ -99,6 +99,7 @@ RUN --mount=type=cache,target=/root/.cache/pip \
99
&& rm -rf sccache.tar.gz sccache-v0.8.1-x86_64-unknown-linux-musl \
100
&& export SCCACHE_BUCKET=vllm-build-sccache \
101
&& export SCCACHE_REGION=us-west-2 \
102
+ && export CMAKE_BUILD_TYPE=Release \
103
&& sccache --show-stats \
104
&& python3 setup.py bdist_wheel --dist-dir=dist \
105
&& sccache --show-stats; \