@tduncan / Nvidia-Optimization-Stack-VLLM Mirror · read-only / commits / e784c6b998
Fork 0

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, 00:45 UTC

[ci/build] sync default value for wheel size (#12398)

Signed-off-by: youkaichao <youkaichao@gmail.com>

youkaichao committed Jan 24, 2025 at 17:54 UTC e784c6b9984e8f8116f74000b863d941495acb0b
2 files changed +3 -2
.buildkite/check-wheel-size.py
+1
@@ -5,6 +5,7 @@ import zipfile
5 # Read the VLLM_MAX_SIZE_MB environment variable, defaulting to 300 MiB
6 # Note that we have 400 MiB quota, please use it wisely.
7 # See https://github.com/pypi/support/issues/3792 .
8 +# Please also sync the value with the one in Dockerfile.
9 VLLM_MAX_SIZE_MB = int(os.environ.get('VLLM_MAX_SIZE_MB', 300))
10
11
Dockerfile
+2 -2
@@ -126,8 +126,8 @@ RUN --mount=type=cache,target=/root/.cache/ccache \
126
127 # Check the size of the wheel if RUN_WHEEL_CHECK is true
128 COPY .buildkite/check-wheel-size.py check-wheel-size.py
129 -# Default max size of the wheel is 250MB
130 -ARG VLLM_MAX_SIZE_MB=250
129 +# sync the default value with .buildkite/check-wheel-size.py
130 +ARG VLLM_MAX_SIZE_MB=300
131 ENV VLLM_MAX_SIZE_MB=$VLLM_MAX_SIZE_MB
132 ARG RUN_WHEEL_CHECK=true
133 RUN if [ "$RUN_WHEEL_CHECK" = "true" ]; then \