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, 13:30 UTC
Updating builkite job for IBM Power (#17111)
Signed-off-by: Aaruni Aggarwal <aaruniagg@gmail.com>
Aaruni Aggarwal committed
Apr 24, 2025 at 22:36 UTC
5adf6f6b7f07b53d1beff47975528b6e637c0470
1 file changed
+11
-4
.buildkite/scripts/hardware_ci/run-cpu-test-ppc64le.sh
+11
-4
@@ -5,7 +5,12 @@
5
set -ex
6
7
# Setup cleanup
8
-remove_docker_container() { podman rm -f cpu-test-ubi9-ppc || true; podman system prune -f; }
8
+remove_docker_container() {
9
+ if [[ -n "$container_id" ]]; then
10
+ podman rm -f "$container_id" || true
11
+ fi
12
+ podman system prune -f
13
+}
14
trap remove_docker_container EXIT
15
remove_docker_container
16
@@ -13,17 +18,17 @@ remove_docker_container
18
podman build -t cpu-test-ubi9-ppc -f docker/Dockerfile.ppc64le .
19
20
# 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
21
+container_id=$(podman run -itd --entrypoint /bin/bash -v /tmp/:/root/.cache/huggingface --privileged=true --network host -e HF_TOKEN cpu-test-ubi9-ppc)
22
23
function cpu_tests() {
24
25
# offline inference
21
- podman exec cpu-test-ubi9-ppc bash -c "
26
+ podman exec -it "$container_id" bash -c "
27
set -e
28
python3 examples/offline_inference/basic/generate.py --model facebook/opt-125m"
29
30
# Run basic model test
26
- podman exec cpu-test-ubi9-ppc bash -c "
31
+ podman exec -it "$container_id" bash -c "
32
set -e
33
pip install pytest pytest-asyncio einops peft Pillow soundfile transformers_stream_generator matplotlib
34
pip install sentence-transformers datamodel_code_generator
@@ -33,6 +38,8 @@ function cpu_tests() {
38
}
39
40
# All of CPU tests are expected to be finished less than 40 mins.
41
+
42
+export container_id
43
export -f cpu_tests
44
timeout 40m bash -c cpu_tests
45