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 17, 06:45 UTC
Dockerfile: Upgrade Cuda to 12.1 (#1609)
GhaziSyed committed
Nov 9, 2023 at 20:49 UTC
aebfcb262a2c6a66f96d8a82efc4ac4c35092222
1 file changed
+4
-4
Dockerfile
+4
-4
@@ -1,4 +1,4 @@
1
-FROM nvidia/cuda:11.8.0-devel-ubuntu22.04 AS dev
1
+FROM nvidia/cuda:12.1.0-devel-ubuntu22.04 AS dev
2
3
RUN apt-get update -y \
4
&& apt-get install -y python3-pip
@@ -9,7 +9,7 @@ WORKDIR /workspace
9
COPY requirements.txt requirements.txt
10
RUN --mount=type=cache,target=/root/.cache/pip \
11
pip install -r requirements.txt
12
-
12
+
13
# install development dependencies
14
COPY requirements-dev.txt requirements-dev.txt
15
RUN --mount=type=cache,target=/root/.cache/pip \
@@ -26,7 +26,7 @@ COPY pyproject.toml pyproject.toml
26
COPY vllm/__init__.py vllm/__init__.py
27
28
# max jobs used by Ninja to build extensions
29
-ENV MAX_JOBS=$max_jobs
29
+ENV MAX_JOBS=$max_jobs
30
RUN python3 setup.py build_ext --inplace
31
32
# image to run unit testing suite
@@ -41,7 +41,7 @@ COPY vllm vllm
41
ENTRYPOINT ["python3", "-m", "pytest", "tests"]
42
43
# use CUDA base as CUDA runtime dependencies are already installed via pip
44
-FROM nvidia/cuda:11.8.0-base-ubuntu22.04 AS vllm-base
44
+FROM nvidia/cuda:12.1.0-base-ubuntu22.04 AS vllm-base
45
46
# libnccl required for ray
47
RUN apt-get update -y \