@tduncan / Nvidia-Optimization-Stack-VLLM Mirror · read-only / commits / 1e4ce295ae
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, 03:15 UTC

[CI][CPU] adding build number to docker image name (#11788)

Signed-off-by: Yuan Zhou <yuan.zhou@intel.com>

Yuan committed Jan 7, 2025 at 15:28 UTC 1e4ce295ae70771f8e0eaa50962b3dda29c3f0d6
1 file changed +11 -11
.buildkite/run-cpu-test.sh
+11 -11
@@ -9,31 +9,31 @@ CORE_RANGE=${CORE_RANGE:-48-95}
9 NUMA_NODE=${NUMA_NODE:-1}
10
11 # Try building the docker image
12 -numactl -C "$CORE_RANGE" -N "$NUMA_NODE" docker build -t cpu-test -f Dockerfile.cpu .
13 -numactl -C "$CORE_RANGE" -N "$NUMA_NODE" docker build --build-arg VLLM_CPU_DISABLE_AVX512="true" -t cpu-test-avx2 -f Dockerfile.cpu .
12 +numactl -C "$CORE_RANGE" -N "$NUMA_NODE" docker build -t cpu-test-"$BUILDKITE_BUILD_NUMBER" -f Dockerfile.cpu .
13 +numactl -C "$CORE_RANGE" -N "$NUMA_NODE" docker build --build-arg VLLM_CPU_DISABLE_AVX512="true" -t cpu-test-"$BUILDKITE_BUILD_NUMBER"-avx2 -f Dockerfile.cpu .
14
15 # Setup cleanup
16 -remove_docker_container() { docker rm -f cpu-test-"$NUMA_NODE" cpu-test-avx2-"$NUMA_NODE" || true; }
16 +remove_docker_container() { docker rm -f cpu-test-"$BUILDKITE_BUILD_NUMBER"-"$NUMA_NODE" cpu-test-"$BUILDKITE_BUILD_NUMBER"-avx2-"$NUMA_NODE" || true; }
17 trap remove_docker_container EXIT
18 remove_docker_container
19
20 # Run the image, setting --shm-size=4g for tensor parallel.
21 docker run -itd --entrypoint /bin/bash -v ~/.cache/huggingface:/root/.cache/huggingface --cpuset-cpus="$CORE_RANGE" \
22 - --cpuset-mems="$NUMA_NODE" --privileged=true --network host -e HF_TOKEN --env VLLM_CPU_KVCACHE_SPACE=4 --shm-size=4g --name cpu-test-"$NUMA_NODE" cpu-test
22 + --cpuset-mems="$NUMA_NODE" --privileged=true --network host -e HF_TOKEN --env VLLM_CPU_KVCACHE_SPACE=4 --shm-size=4g --name cpu-test-"$BUILDKITE_BUILD_NUMBER"-"$NUMA_NODE" cpu-test-"$BUILDKITE_BUILD_NUMBER"
23 docker run -itd --entrypoint /bin/bash -v ~/.cache/huggingface:/root/.cache/huggingface --cpuset-cpus="$CORE_RANGE" \
24 - --cpuset-mems="$NUMA_NODE" --privileged=true --network host -e HF_TOKEN --env VLLM_CPU_KVCACHE_SPACE=4 --shm-size=4g --name cpu-test-avx2-"$NUMA_NODE" cpu-test-avx2
24 + --cpuset-mems="$NUMA_NODE" --privileged=true --network host -e HF_TOKEN --env VLLM_CPU_KVCACHE_SPACE=4 --shm-size=4g --name cpu-test-"$BUILDKITE_BUILD_NUMBER"-avx2-"$NUMA_NODE" cpu-test-"$BUILDKITE_BUILD_NUMBER"-avx2
25
26 function cpu_tests() {
27 set -e
28 export NUMA_NODE=$2
29
30 # offline inference
31 - docker exec cpu-test-avx2-"$NUMA_NODE" bash -c "
31 + docker exec cpu-test-"$BUILDKITE_BUILD_NUMBER"-avx2-"$NUMA_NODE" bash -c "
32 set -e
33 python3 examples/offline_inference.py"
34
35 # Run basic model test
36 - docker exec cpu-test-"$NUMA_NODE" bash -c "
36 + docker exec cpu-test-"$BUILDKITE_BUILD_NUMBER"-"$NUMA_NODE" bash -c "
37 set -e
38 pip install pytest pytest-asyncio \
39 decord einops librosa peft Pillow sentence-transformers soundfile \
@@ -46,26 +46,26 @@ function cpu_tests() {
46 pytest -v -s tests/models/decoder_only/vision_language -m cpu_model"
47
48 # Run compressed-tensor test
49 - docker exec cpu-test-"$NUMA_NODE" bash -c "
49 + docker exec cpu-test-"$BUILDKITE_BUILD_NUMBER"-"$NUMA_NODE" bash -c "
50 set -e
51 pytest -s -v \
52 tests/quantization/test_compressed_tensors.py::test_compressed_tensors_w8a8_static_setup \
53 tests/quantization/test_compressed_tensors.py::test_compressed_tensors_w8a8_dynamic_per_token"
54
55 # Run AWQ test
56 - docker exec cpu-test-"$NUMA_NODE" bash -c "
56 + docker exec cpu-test-"$BUILDKITE_BUILD_NUMBER"-"$NUMA_NODE" bash -c "
57 set -e
58 pytest -s -v \
59 tests/quantization/test_ipex_quant.py"
60
61 # Run chunked-prefill and prefix-cache test
62 - docker exec cpu-test-"$NUMA_NODE" bash -c "
62 + docker exec cpu-test-"$BUILDKITE_BUILD_NUMBER"-"$NUMA_NODE" bash -c "
63 set -e
64 pytest -s -v -k cpu_model \
65 tests/basic_correctness/test_chunked_prefill.py"
66
67 # online inference
68 - docker exec cpu-test-"$NUMA_NODE" bash -c "
68 + docker exec cpu-test-"$BUILDKITE_BUILD_NUMBER"-"$NUMA_NODE" bash -c "
69 set -e
70 export VLLM_CPU_KVCACHE_SPACE=10
71 export VLLM_CPU_OMP_THREADS_BIND=$1