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, 17:00 UTC
Fix ppc64le buildkite job (#8309)
sumitd2 committed
Sep 11, 2024 at 01:28 UTC
02751a7a42c18454030ff35e350afab31e26f51d
2 files changed
+4
-4
.buildkite/run-cpu-test-ppc64le.sh
+2
-1
@@ -11,8 +11,9 @@ trap remove_docker_container EXIT
11
remove_docker_container
12
13
# Run the image, setting --shm-size=4g for tensor parallel.
14
+source /etc/environment
15
#docker run -itd --entrypoint /bin/bash -v ~/.cache/huggingface:/root/.cache/huggingface --privileged=true --network host -e HF_TOKEN --env VLLM_CPU_KVCACHE_SPACE=4 --shm-size=4g --name cpu-test cpu-test
15
-docker run -itd --entrypoint /bin/bash -v ~/.cache/huggingface:/root/.cache/huggingface --privileged=true --network host -e HF_TOKEN --name cpu-test cpu-test
16
+docker run -itd --entrypoint /bin/bash -v ~/.cache/huggingface:/root/.cache/huggingface --privileged=true --network host -e HF_TOKEN=$HF_TOKEN --name cpu-test cpu-test
17
18
# Run basic model test
19
docker exec cpu-test bash -c "
Dockerfile.ppc64le
+2
-3
@@ -4,7 +4,7 @@ USER root
4
5
ENV PATH="/usr/local/cargo/bin:$PATH:/opt/conda/bin/"
6
7
-RUN apt-get update -y && apt-get install -y git wget vim libnuma-dev libsndfile-dev libprotobuf-dev build-essential
7
+RUN apt-get update -y && apt-get install -y git wget curl vim libnuma-dev libsndfile-dev libprotobuf-dev build-essential
8
9
# Some packages in requirements-cpu are installed here
10
# IBM provides optimized packages for ppc64le processors in the open-ce project for mamba
@@ -16,7 +16,7 @@ COPY ./ /workspace/vllm
16
WORKDIR /workspace/vllm
17
18
# These packages will be in rocketce eventually
19
-RUN pip install -v cmake torch==2.3.1 uvloop==0.20.0 -r requirements-cpu.txt --prefer-binary --extra-index-url https://repo.fury.io/mgiessing
19
+RUN pip install -v cmake xformers torch==2.3.1 uvloop==0.20.0 -r requirements-cpu.txt --prefer-binary --extra-index-url https://repo.fury.io/mgiessing
20
21
RUN VLLM_TARGET_DEVICE=cpu python3 setup.py install
22
@@ -25,4 +25,3 @@ WORKDIR /workspace/
25
RUN ln -s /workspace/vllm/tests && ln -s /workspace/vllm/examples && ln -s /workspace/vllm/benchmarks
26
27
ENTRYPOINT ["python3", "-m", "vllm.entrypoints.openai.api_server"]
28
-