@tduncan / Nvidia-Optimization-Stack-VLLM Mirror · read-only / commits / 2f7024987e
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/Build][Bugfix] Fix CPU CI image clean up (#11836)

Signed-off-by: jiang1.li <jiang1.li@intel.com>

Li, Jiang committed Jan 8, 2025 at 23:18 UTC 2f7024987e582b85b280909b87287668cd97c92f
2 files changed +3 -6
.buildkite/run-cpu-test.sh
+2 -5
@@ -13,7 +13,7 @@ numactl -C "$CORE_RANGE" -N "$NUMA_NODE" docker build -t cpu-test-"$BUILDKITE_BU
13 numactl -C "$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-"$BUILDKITE_BUILD_NUMBER"-"$NUMA_NODE" cpu-test-"$BUILDKITE_BUILD_NUMBER"-avx2-"$NUMA_NODE" || true; }
16 +remove_docker_container() { set -e; 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
@@ -35,10 +35,7 @@ function cpu_tests() {
35 # Run basic model test
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 \
40 - transformers_stream_generator matplotlib datamodel_code_generator
41 - pip install torchvision --index-url https://download.pytorch.org/whl/cpu
38 + pip install -r vllm/requirements-test.txt
39 pytest -v -s tests/models/decoder_only/language -m cpu_model
40 pytest -v -s tests/models/embedding/language -m cpu_model
41 pytest -v -s tests/models/encoder_decoder/language -m cpu_model
vllm/model_executor/layers/activation.py
+1 -1
@@ -61,7 +61,7 @@ class SiluAndMul(CustomOp):
61
62 def __init__(self):
63 super().__init__()
64 - if current_platform.is_cuda_alike():
64 + if current_platform.is_cuda_alike() or current_platform.is_cpu():
65 self.op = torch.ops._C.silu_and_mul
66 elif current_platform.is_xpu():
67 import intel_extension_for_pytorch as ipex