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 15, 14:45 UTC
Adding vllm buildkite job for IBM Power (#16679)
Signed-off-by: Aaruni Aggarwal <aaruniagg@gmail.com>
Aaruni Aggarwal committed
Apr 17, 2025 at 08:17 UTC
3c776dcefb3f327d252a6950fd6d5e990856a11e
1 file changed
+26
-2
.buildkite/scripts/hardware_ci/run-cpu-test-ppc64le.sh
+26
-2
@@ -5,10 +5,34 @@
5
set -ex
6
7
# Setup cleanup
8
-remove_docker_container() { docker rm -f cpu-test || true; docker system prune -f; }
8
+remove_docker_container() { podman rm -f cpu-test-ubi9-ppc || true; podman system prune -f; }
9
trap remove_docker_container EXIT
10
remove_docker_container
11
12
# Try building the docker image
13
-docker build -t cpu-test -f docker/Dockerfile.ppc64le .
13
+podman build -t cpu-test-ubi9-ppc -f docker/Dockerfile.ppc64le .
14
+
15
+# Run the image
16
+podman run -itd --entrypoint /bin/bash -v /tmp/:/root/.cache/huggingface --privileged=true --network host -e HF_TOKEN --name cpu-test-ubi9-ppc cpu-test-ubi9-ppc
17
+
18
+function cpu_tests() {
19
+
20
+ # offline inference
21
+ podman exec cpu-test-ubi9-ppc bash -c "
22
+ set -e
23
+ python3 examples/offline_inference/basic/generate.py --model facebook/opt-125m"
24
+
25
+ # Run basic model test
26
+ podman exec cpu-test-ubi9-ppc bash -c "
27
+ set -e
28
+ pip install pytest pytest-asyncio einops peft Pillow soundfile transformers_stream_generator matplotlib
29
+ pip install sentence-transformers datamodel_code_generator
30
+ pytest -v -s tests/models/embedding/language/test_cls_models.py::test_classification_models[float-jason9693/Qwen2.5-1.5B-apeach]
31
+ pytest -v -s tests/models/embedding/language/test_embedding.py::test_models[half-BAAI/bge-base-en-v1.5]
32
+ pytest -v -s tests/models/encoder_decoder/language -m cpu_model"
33
+}
34
+
35
+# All of CPU tests are expected to be finished less than 40 mins.
36
+export -f cpu_tests
37
+timeout 40m bash -c cpu_tests
38