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 14, 05:30 UTC
[CI] Raise VLLM_MAX_SIZE_MB to 500 due to failing Build wheel - CUDA 12.9 (#26722)
Signed-off-by: mgoin <mgoin64@gmail.com>
Michael Goin committed
Oct 14, 2025 at 13:52 UTC
04b5f9802da90f27636549677ec280d34236a51d
2 files changed
+3
-3
.buildkite/check-wheel-size.py
+2
-2
@@ -5,11 +5,11 @@ import os
5
import sys
6
import zipfile
7
8
-# Read the VLLM_MAX_SIZE_MB environment variable, defaulting to 450 MiB
8
+# Read the VLLM_MAX_SIZE_MB environment variable, defaulting to 500 MiB
9
# Note that we have 800 MiB quota, please use it wisely.
10
# See https://github.com/pypi/support/issues/6326 .
11
# Please also sync the value with the one in Dockerfile.
12
-VLLM_MAX_SIZE_MB = int(os.environ.get("VLLM_MAX_SIZE_MB", 450))
12
+VLLM_MAX_SIZE_MB = int(os.environ.get("VLLM_MAX_SIZE_MB", 500))
13
14
15
def print_top_10_largest_files(zip_file):
docker/Dockerfile
+1
-1
@@ -229,7 +229,7 @@ RUN --mount=type=cache,target=/root/.cache/ccache \
229
# Check the size of the wheel if RUN_WHEEL_CHECK is true
230
COPY .buildkite/check-wheel-size.py check-wheel-size.py
231
# sync the default value with .buildkite/check-wheel-size.py
232
-ARG VLLM_MAX_SIZE_MB=450
232
+ARG VLLM_MAX_SIZE_MB=500
233
ENV VLLM_MAX_SIZE_MB=$VLLM_MAX_SIZE_MB
234
ARG RUN_WHEEL_CHECK=true
235
RUN if [ "$RUN_WHEEL_CHECK" = "true" ]; then \