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, 20:45 UTC
[CI/Build] Build on Ubuntu 20.04 instead of 22.04 (#6517)
Tyler Michael Smith committed
Jul 18, 2024 at 20:29 UTC
1689219ebf01c750de492271832e27c39df38648
4 files changed
+31
-13
.buildkite/test-pipeline.yaml
+2
-2
@@ -21,7 +21,7 @@ steps:
21
fast_check: true
22
fast_check_only: true
23
commands:
24
- - apt-get install curl libsodium23 && pytest -v -s tensorizer_loader # Tensorizer
24
+ - apt-get install -y curl libsodium23 && pytest -v -s tensorizer_loader # Tensorizer
25
- pytest -v -s metrics # Metrics
26
- "pip install \
27
opentelemetry-sdk \
@@ -221,7 +221,7 @@ steps:
221
- label: Tensorizer Test
222
#mirror_hardwares: [amd]
223
commands:
224
- - apt-get install curl libsodium23
224
+ - apt-get install -y curl libsodium23
225
- export VLLM_WORKER_MULTIPROC_METHOD=spawn
226
- pytest -v -s tensorizer_loader
227
Dockerfile
+27
-9
@@ -8,10 +8,10 @@
8
ARG CUDA_VERSION=12.4.1
9
#################### BASE BUILD IMAGE ####################
10
# prepare basic build environment
11
-FROM nvidia/cuda:${CUDA_VERSION}-devel-ubuntu22.04 AS base
11
+FROM nvidia/cuda:${CUDA_VERSION}-devel-ubuntu20.04 AS base
12
13
ARG CUDA_VERSION=12.4.1
14
-ARG PYTHON_VERSION=3
14
+ARG PYTHON_VERSION=3.10
15
16
ENV DEBIAN_FRONTEND=noninteractive
17
@@ -21,13 +21,16 @@ RUN echo 'tzdata tzdata/Areas select America' | debconf-set-selections \
21
&& apt-get install -y ccache software-properties-common \
22
&& add-apt-repository ppa:deadsnakes/ppa \
23
&& apt-get update -y \
24
- && apt-get install -y python${PYTHON_VERSION} python${PYTHON_VERSION}-dev python${PYTHON_VERSION}-venv python3-pip \
24
+ && apt-get install -y python${PYTHON_VERSION} python${PYTHON_VERSION}-dev python${PYTHON_VERSION}-venv \
25
&& if [ "${PYTHON_VERSION}" != "3" ]; then update-alternatives --install /usr/bin/python3 python3 /usr/bin/python${PYTHON_VERSION} 1; fi \
26
- && python3 --version \
27
- && python3 -m pip --version
26
+ && python3 --version
27
28
RUN apt-get update -y \
30
- && apt-get install -y python3-pip git curl sudo
29
+ && apt-get install -y git curl sudo
30
+
31
+# Install pip s.t. it will be compatible with our PYTHON_VERSION
32
+RUN curl -sS https://bootstrap.pypa.io/get-pip.py | python${PYTHON_VERSION}
33
+RUN python3 -m pip --version
34
35
# Workaround for https://github.com/openai/triton/issues/2507 and
36
# https://github.com/pytorch/pytorch/issues/107960 -- hopefully
@@ -58,7 +61,7 @@ ENV TORCH_CUDA_ARCH_LIST=${torch_cuda_arch_list}
61
#################### WHEEL BUILD IMAGE ####################
62
FROM base AS build
63
61
-ARG PYTHON_VERSION=3
64
+ARG PYTHON_VERSION=3.10
65
66
# install build dependencies
67
COPY requirements-build.txt requirements-build.txt
@@ -149,12 +152,27 @@ RUN pip --verbose wheel -r requirements-mamba.txt \
152
153
#################### vLLM installation IMAGE ####################
154
# image with vLLM installed
152
-FROM nvidia/cuda:${CUDA_VERSION}-base-ubuntu22.04 AS vllm-base
155
+FROM nvidia/cuda:${CUDA_VERSION}-base-ubuntu20.04 AS vllm-base
156
ARG CUDA_VERSION=12.4.1
157
+ARG PYTHON_VERSION=3.10
158
WORKDIR /vllm-workspace
159
160
+RUN echo 'tzdata tzdata/Areas select America' | debconf-set-selections \
161
+ && echo 'tzdata tzdata/Zones/America select Los_Angeles' | debconf-set-selections \
162
+ && apt-get update -y \
163
+ && apt-get install -y ccache software-properties-common \
164
+ && add-apt-repository ppa:deadsnakes/ppa \
165
+ && apt-get update -y \
166
+ && apt-get install -y python${PYTHON_VERSION} python${PYTHON_VERSION}-dev python${PYTHON_VERSION}-venv \
167
+ && if [ "${PYTHON_VERSION}" != "3" ]; then update-alternatives --install /usr/bin/python3 python3 /usr/bin/python${PYTHON_VERSION} 1; fi \
168
+ && python3 --version
169
+
170
RUN apt-get update -y \
157
- && apt-get install -y python3-pip git vim
171
+ && apt-get install -y python3-pip git curl
172
+
173
+# Install pip s.t. it will be compatible with our PYTHON_VERSION
174
+RUN curl -sS https://bootstrap.pypa.io/get-pip.py | python${PYTHON_VERSION}
175
+RUN python3 -m pip --version
176
177
# Workaround for https://github.com/openai/triton/issues/2507 and
178
# https://github.com/pytorch/pytorch/issues/107960 -- hopefully
Dockerfile.openvino
+1
-1
@@ -1,7 +1,7 @@
1
# The vLLM Dockerfile is used to construct vLLM image that can be directly used
2
# to run the OpenAI compatible server.
3
4
-FROM ubuntu:22.04 AS dev
4
+FROM ubuntu:20.04 AS dev
5
6
RUN apt-get update -y && \
7
apt-get install -y python3-pip git
Dockerfile.xpu
+1
-1
@@ -1,4 +1,4 @@
1
-FROM intel/oneapi-basekit:2024.1.0-devel-ubuntu22.04
1
+FROM intel/oneapi-basekit:2024.1.0-devel-ubuntu20.04
2
3
RUN wget -O- https://apt.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB | gpg --dearmor | tee /usr/share/keyrings/intel-oneapi-archive-keyring.gpg > /dev/null && \
4
echo "deb [signed-by=/usr/share/keyrings/intel-oneapi-archive-keyring.gpg] https://apt.repos.intel.com/oneapi all main " | tee /etc/apt/sources.list.d/oneAPI.list && \