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, 08:30 UTC
[Hardware][Intel-Gaudi] [CI/Build] Fix multiple containers using the same name in run-hpu-test.sh (#18752)
Signed-off-by: Lukasz Durejko <ldurejko@habana.ai>
Łukasz Durejko committed
May 27, 2025 at 09:10 UTC
bbd9a84dc55882a95974449008c275794309599f
1 file changed
+6
-6
.buildkite/scripts/hardware_ci/run-hpu-test.sh
+6
-6
@@ -10,17 +10,17 @@ docker build -t hpu-test-env -f docker/Dockerfile.hpu .
10
# Setup cleanup
11
# certain versions of HPU software stack have a bug that can
12
# override the exit code of the script, so we need to use
13
-# separate remove_docker_container and remove_docker_container_and_exit
13
+# separate remove_docker_containers and remove_docker_containers_and_exit
14
# functions, while other platforms only need one remove_docker_container
15
# function.
16
EXITCODE=1
17
-remove_docker_container() { docker rm -f hpu-test || true; }
18
-remove_docker_container_and_exit() { remove_docker_container; exit $EXITCODE; }
19
-trap remove_docker_container_and_exit EXIT
20
-remove_docker_container
17
+remove_docker_containers() { docker rm -f hpu-test || true; docker rm -f hpu-test-tp2 || true; }
18
+remove_docker_containers_and_exit() { remove_docker_containers; exit $EXITCODE; }
19
+trap remove_docker_containers_and_exit EXIT
20
+remove_docker_containers
21
22
# Run the image and launch offline inference
23
docker run --runtime=habana --name=hpu-test --network=host -e HABANA_VISIBLE_DEVICES=all -e VLLM_SKIP_WARMUP=true --entrypoint="" hpu-test-env python3 examples/offline_inference/basic/generate.py --model facebook/opt-125m
24
-docker run --runtime=habana --name=hpu-test --network=host -e HABANA_VISIBLE_DEVICES=all -e VLLM_SKIP_WARMUP=true --entrypoint="" hpu-test-env python3 examples/offline_inference/basic/generate.py --model facebook/opt-125m --tensor-parallel-size 2
24
+docker run --runtime=habana --name=hpu-test-tp2 --network=host -e HABANA_VISIBLE_DEVICES=all -e VLLM_SKIP_WARMUP=true --entrypoint="" hpu-test-env python3 examples/offline_inference/basic/generate.py --model facebook/opt-125m --tensor-parallel-size 2
25
26
EXITCODE=$?