RogerSik/ollama-intel-gpu
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update compute-runtine, oneAPI, and now use upstream ollama

Browse Source
This commit is contained in:
Matt Curfman 2024-07-31 21:45:50 -07:00
parent 65184dc2a3
commit dadaf4000c
3 changed files with 23 additions and 49 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

69
Dockerfile
View File

@ -1,77 +1,54 @@
FROM ubuntu:22.04
ENV DEBIAN_FRONTEND noninteractive
ENV DEBIAN_FRONTEND=noninteractive
ENV TZ=america/los_angeles
# #1 install base packages
#-----------------------
# Base packages
RUN apt update && \
apt install --no-install-recommends -q -y \
software-properties-common \
ca-certificates \
gnupg \
wget \
python3 \
python3-pip
curl \
ocl-icd-libopencl1
# #2 Install Intel GPU compute user-space drivers
#-----------------------
RUN wget -qO - https://repositories.intel.com/gpu/intel-graphics.key | \
gpg --dearmor --output /usr/share/keyrings/intel-graphics.gpg && \
echo "deb [arch=amd64 signed-by=/usr/share/keyrings/intel-graphics.gpg] https://repositories.intel.com/gpu/ubuntu jammy client" | \
tee /etc/apt/sources.list.d/intel-gpu-jammy.list
# apt update && \
# apt install --no-install-recommends -q -y \
# intel-opencl-icd intel-level-zero-gpu level-zero
# *Defect/Workaround*: Intel's apt repo does not contain the latest GPU runtime supporting kernel 6.8, so install the packages directly from their compute
# runtime repo
# https://github.com/intel/compute-runtime/issues/710
RUN apt update && \
apt install --no-install-recommends -q -y \
ocl-icd-libopencl1 \
clinfo && \
cd /tmp && \
wget https://github.com/intel/intel-graphics-compiler/releases/download/igc-1.0.16510.2/intel-igc-core_1.0.16510.2_amd64.deb && \
wget https://github.com/intel/intel-graphics-compiler/releases/download/igc-1.0.16510.2/intel-igc-opencl_1.0.16510.2_amd64.deb && \
wget https://github.com/intel/compute-runtime/releases/download/24.13.29138.7/intel-level-zero-gpu_1.3.29138.7_amd64.deb && \
wget https://github.com/intel/compute-runtime/releases/download/24.13.29138.7/intel-opencl-icd_24.13.29138.7_amd64.deb && \
wget https://github.com/intel/compute-runtime/releases/download/24.13.29138.7/libigdgmm12_22.3.18_amd64.deb && \
# Intel GPU compute user-space drivers
RUN mkdir -p /tmp/gpu && \
cd /tmp/gpu && \
wget https://github.com/oneapi-src/level-zero/releases/download/v1.17.19/level-zero_1.17.19+u22.04_amd64.deb && \
wget https://github.com/intel/intel-graphics-compiler/releases/download/igc-1.0.17193.4/intel-igc-core_1.0.17193.4_amd64.deb && \
wget https://github.com/intel/intel-graphics-compiler/releases/download/igc-1.0.17193.4/intel-igc-opencl_1.0.17193.4_amd64.deb && \
wget https://github.com/intel/compute-runtime/releases/download/24.26.30049.6/intel-level-zero-gpu_1.3.30049.6_amd64.deb && \
wget https://github.com/intel/compute-runtime/releases/download/24.26.30049.6/intel-opencl-icd_24.26.30049.6_amd64.deb && \
wget https://github.com/intel/compute-runtime/releases/download/24.26.30049.6/libigdgmm12_22.3.20_amd64.deb && \
dpkg -i *.deb && \
apt install --no-install-recommends -q -y \
level-zero
rm *.deb
# Required compute runtime level-zero variables
ENV ZES_ENABLE_SYSMAN=1
# #3 Install oneAPI
#-----------------------
# *Defect/Workaround*: Intel's oneAPI MKL changed the linkage model, breaking pytorch wheel. Downgrade to oneAPI 2024.0 instead
# Source: https://github.com/pytorch/pytorch/issues/123097
# oneAPI
RUN wget -qO - https://apt.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRODUCTS.PUB | \
gpg --dearmor --output /usr/share/keyrings/oneapi-archive-keyring.gpg && \
echo "deb [signed-by=/usr/share/keyrings/oneapi-archive-keyring.gpg] https://apt.repos.intel.com/oneapi all main" | \
tee /etc/apt/sources.list.d/oneAPI.list && \
apt update && \
apt install --no-install-recommends -q -y \
intel-oneapi-common-vars=2024.0.0-49406 \
intel-oneapi-common-oneapi-vars=2024.0.0-49406 \
intel-oneapi-mkl=2024.0.0-49656 \
intel-oneapi-tcm-1.0=1.0.0-435 \
intel-oneapi-dnnl=2024.0.0-49521
intel-oneapi-runtime-libs
# Required oneAPI environment variables
ENV USE_XETLA=OFF
ENV SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
ENV SYCL_CACHE_PERSISTENT=1
COPY _init.sh /usr/share/lib/init_workspace.sh
COPY _run.sh /usr/share/lib/run_workspace.sh
# #3 Ollama specific dependencies
#-----------------------
RUN pip3 install --pre --upgrade ipex-llm[cpp]
# Ollama
RUN curl -fsSL https://ollama.com/install.sh | sh
ENV OLLAMA_NUM_GPU=999
ENV OLLAMA_HOST 0.0.0.0:11434
ENV OLLAMA_HOST=0.0.0.0:11434
ENTRYPOINT ["/bin/bash", "/usr/share/lib/run_workspace.sh"]
# https://github.com/ollama/ollama/issues/1590
ENV OLLAMA_INTEL_GPU=1
ENTRYPOINT ["/usr/local/bin/ollama", "serve"]

1
_init.sh
View File

@ -1 +0,0 @@
source /opt/intel/oneapi/setvars.sh

2
_run.sh
View File

@ -1,2 +0,0 @@
source /usr/share/lib/init_workspace.sh
/usr/local/lib/python3.10/dist-packages/bigdl/cpp/libs/ollama serve