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, 15:45 UTC
[misc] small qol fixes for release process (#8517)
Simon Mo committed
Sep 16, 2024 at 15:11 UTC
5ce45eb54d3fb870f1fb6865c67aac05ec9bf555
2 files changed
+5
-1
Dockerfile
+2
@@ -82,6 +82,7 @@ ENV BUILDKITE_COMMIT=${buildkite_commit}
82
ARG USE_SCCACHE
83
ARG SCCACHE_BUCKET_NAME=vllm-build-sccache
84
ARG SCCACHE_REGION_NAME=us-west-2
85
+ARG SCCACHE_S3_NO_CREDENTIALS=0
86
# if USE_SCCACHE is set, use sccache to speed up compilation
87
RUN --mount=type=cache,target=/root/.cache/pip \
88
if [ "$USE_SCCACHE" = "1" ]; then \
@@ -92,6 +93,7 @@ RUN --mount=type=cache,target=/root/.cache/pip \
93
&& rm -rf sccache.tar.gz sccache-v0.8.1-x86_64-unknown-linux-musl \
94
&& export SCCACHE_BUCKET=${SCCACHE_BUCKET_NAME} \
95
&& export SCCACHE_REGION=${SCCACHE_REGION_NAME} \
96
+ && export SCCACHE_S3_NO_CREDENTIALS=${SCCACHE_S3_NO_CREDENTIALS} \
97
&& export SCCACHE_IDLE_TIMEOUT=0 \
98
&& export CMAKE_BUILD_TYPE=Release \
99
&& sccache --show-stats \
setup.py
+3
-1
@@ -371,7 +371,9 @@ def get_vllm_version() -> str:
371
cuda_version = str(get_nvcc_cuda_version())
372
if cuda_version != MAIN_CUDA_VERSION:
373
cuda_version_str = cuda_version.replace(".", "")[:3]
374
- version += f"+cu{cuda_version_str}"
374
+ # skip this for source tarball, required for pypi
375
+ if "sdist" not in sys.argv:
376
+ version += f"+cu{cuda_version_str}"
377
elif _is_hip():
378
# Get the HIP version
379
hipcc_version = get_hipcc_rocm_version()