xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-01-23 13:14:34 +08:00
Code Issues Packages Projects Releases Wiki Activity

[cpu][fix] Fix Arm CI tests (#29552)

Browse Source 
Signed-off-by: Fadi Arafeh <fadi.arafeh@arm.com>
This commit is contained in:
Fadi Arafeh 2025-11-27 05:09:41 +00:00 committed by GitHub
parent da8e1a1bf9
commit ecb1952378
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 12 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
.buildkite/scripts/hardware_ci/run-cpu-test-arm.sh
View File

@ -7,53 +7,51 @@ set -ex
# allow to bind to different cores
CORE_RANGE=${CORE_RANGE:-0-16}
OMP_CORE_RANGE=${OMP_CORE_RANGE:-0-16}
NUMA_NODE=${NUMA_NODE:-0}
export CMAKE_BUILD_PARALLEL_LEVEL=32
export CMAKE_BUILD_PARALLEL_LEVEL=16
# Setup cleanup
remove_docker_container() {
set -e;
docker rm -f cpu-test-"$NUMA_NODE" || true;
docker rm -f cpu-test || true;
}
trap remove_docker_container EXIT
remove_docker_container
# Try building the docker image
numactl -C "$CORE_RANGE" -N "$NUMA_NODE" docker build --tag cpu-test-"$NUMA_NODE" --target vllm-test -f docker/Dockerfile.cpu .
docker build --tag cpu-test --target vllm-test -f docker/Dockerfile.cpu .
# Run the image, setting --shm-size=4g for tensor parallel.
docker run -itd --cpuset-cpus="$CORE_RANGE" --cpuset-mems="$NUMA_NODE" --entrypoint /bin/bash -v ~/.cache/huggingface:/root/.cache/huggingface --privileged=true -e HF_TOKEN --env VLLM_CPU_KVCACHE_SPACE=16 --env VLLM_CPU_CI_ENV=1 -e E2E_OMP_THREADS="$OMP_CORE_RANGE" --shm-size=4g --name cpu-test-"$NUMA_NODE" cpu-test-"$NUMA_NODE"
# Run the image
docker run -itd --cpuset-cpus="$CORE_RANGE" --entrypoint /bin/bash -v ~/.cache/huggingface:/root/.cache/huggingface -e HF_TOKEN --env VLLM_CPU_KVCACHE_SPACE=16 --env VLLM_CPU_CI_ENV=1 -e E2E_OMP_THREADS="$OMP_CORE_RANGE" --shm-size=4g --name cpu-test cpu-test
function cpu_tests() {
set -e
export NUMA_NODE=$2
docker exec cpu-test-"$NUMA_NODE" bash -c "
docker exec cpu-test bash -c "
set -e
pip list"
# offline inference
docker exec cpu-test-"$NUMA_NODE" bash -c "
docker exec cpu-test bash -c "
set -e
python3 examples/offline_inference/basic/generate.py --model facebook/opt-125m"
# Run kernel tests
docker exec cpu-test-"$NUMA_NODE" bash -c "
docker exec cpu-test bash -c "
set -e
pytest -x -v -s tests/kernels/test_onednn.py
pytest -x -v -s tests/kernels/attention/test_cpu_attn.py"
# basic online serving
docker exec cpu-test-"$NUMA_NODE" bash -c '
docker exec cpu-test bash -c '
set -e
VLLM_CPU_OMP_THREADS_BIND=$E2E_OMP_THREADS vllm serve meta-llama/Llama-3.2-3B-Instruct --max-model-len 2048 &
VLLM_CPU_OMP_THREADS_BIND=$E2E_OMP_THREADS vllm serve Qwen/Qwen3-0.6B --max-model-len 2048 &
server_pid=$!
timeout 600 bash -c "until curl localhost:8000/v1/models; do sleep 1; done" || exit 1
vllm bench serve \
--backend vllm \
--dataset-name random \
--model meta-llama/Llama-3.2-3B-Instruct \
--model Qwen/Qwen3-0.6B \
--num-prompts 20 \
--endpoint /v1/completions
kill -s SIGTERM $server_pid &'
@ -61,4 +59,4 @@ function cpu_tests() {
# All of CPU tests are expected to be finished less than 40 mins.
export -f cpu_tests
timeout 2h bash -c "cpu_tests $CORE_RANGE $NUMA_NODE"
timeout 2h bash -c cpu_tests