mirrors/ollama-for-amd
1
0
Fork 0
mirror of https://github.com/likelovewant/ollama-for-amd.git synced 2025-12-21 22:33:56 +00:00
Code Issues Packages Projects Releases Wiki Activity
Files
bc8909fb38525c89dda842d4ecfc86a933089a99
ollama-for-amd/Dockerfile
Daniel Hiltgen bc8909fb38 Use runners for GPU discovery (#12090) 
This revamps how we discover GPUs in the system by leveraging the Ollama
runner.  This should eliminate inconsistency between our GPU discovery and the
runners capabilities at runtime, particularly for cases where we try to filter
out unsupported GPUs.  Now the runner does that implicitly based on the actual
device list.  In some cases free VRAM reporting can be unreliable which can
leaad to scheduling mistakes, so this also includes a patch to leverage more
reliable VRAM reporting libraries if available.

Automatic workarounds have been removed as only one GPU leveraged this, which
is now documented. This GPU will soon fall off the support matrix with the next
ROCm bump.

Additional cleanup of the scheduler and discovery packages can be done in the
future once we have switched on the new memory management code, and removed
support for the llama runner.
2025-10-01 15:12:32 -07:00

156 lines
6.5 KiB
Docker
Raw Blame History

# vim: filetype=dockerfile
ARG FLAVOR=${TARGETARCH}
ARG PARALLEL=8
ARG ROCMVERSION=6.3.3
ARG JETPACK5VERSION=r35.4.1
ARG JETPACK6VERSION=r36.4.0
ARG CMAKEVERSION=3.31.2
# We require gcc v10 minimum. v10.3 has regressions, so the rockylinux 8.5 AppStream has the latest compatible version
FROM --platform=linux/amd64 rocm/dev-almalinux-8:${ROCMVERSION}-complete AS base-amd64
RUN yum install -y yum-utils \
&& yum-config-manager --add-repo https://dl.rockylinux.org/vault/rocky/8.5/AppStream/\$basearch/os/ \
&& rpm --import https://dl.rockylinux.org/pub/rocky/RPM-GPG-KEY-Rocky-8 \
&& dnf install -y yum-utils ccache gcc-toolset-10-gcc-10.2.1-8.2.el8 gcc-toolset-10-gcc-c++-10.2.1-8.2.el8 gcc-toolset-10-binutils-2.35-11.el8 \
&& dnf install -y ccache \
&& yum-config-manager --add-repo https://developer.download.nvidia.com/compute/cuda/repos/rhel8/x86_64/cuda-rhel8.repo
ENV PATH=/opt/rh/gcc-toolset-10/root/usr/bin:$PATH
FROM --platform=linux/arm64 almalinux:8 AS base-arm64
# install epel-release for ccache
RUN yum install -y yum-utils epel-release \
&& dnf install -y clang ccache \
&& yum-config-manager --add-repo https://developer.download.nvidia.com/compute/cuda/repos/rhel8/sbsa/cuda-rhel8.repo
ENV CC=clang CXX=clang++
FROM base-${TARGETARCH} AS base
ARG CMAKEVERSION
RUN curl -fsSL https://github.com/Kitware/CMake/releases/download/v${CMAKEVERSION}/cmake-${CMAKEVERSION}-linux-$(uname -m).tar.gz | tar xz -C /usr/local --strip-components 1
COPY CMakeLists.txt CMakePresets.json .
COPY ml/backend/ggml/ggml ml/backend/ggml/ggml
ENV LDFLAGS=-s
FROM base AS cpu
RUN dnf install -y gcc-toolset-11-gcc gcc-toolset-11-gcc-c++
ENV PATH=/opt/rh/gcc-toolset-11/root/usr/bin:$PATH
ARG PARALLEL
RUN --mount=type=cache,target=/root/.ccache \
cmake --preset 'CPU' \
&& cmake --build --parallel ${PARALLEL} --preset 'CPU' \
&& cmake --install build --component CPU --strip --parallel ${PARALLEL}
FROM base AS cuda-11
ARG CUDA11VERSION=11.8
RUN dnf install -y cuda-toolkit-${CUDA11VERSION//./-}
ENV PATH=/usr/local/cuda-11/bin:$PATH
ARG PARALLEL
RUN --mount=type=cache,target=/root/.ccache \
cmake --preset 'CUDA 11' -DOLLAMA_RUNNER_DIR="cuda_v11" \
&& cmake --build --parallel ${PARALLEL} --preset 'CUDA 11' \
&& cmake --install build --component CUDA --strip --parallel ${PARALLEL}
FROM base AS cuda-12
ARG CUDA12VERSION=12.8
RUN dnf install -y cuda-toolkit-${CUDA12VERSION//./-}
ENV PATH=/usr/local/cuda-12/bin:$PATH
ARG PARALLEL
RUN --mount=type=cache,target=/root/.ccache \
cmake --preset 'CUDA 12' -DOLLAMA_RUNNER_DIR="cuda_v12"\
&& cmake --build --parallel ${PARALLEL} --preset 'CUDA 12' \
&& cmake --install build --component CUDA --strip --parallel ${PARALLEL}
FROM base AS cuda-13
ARG CUDA13VERSION=13.0
RUN dnf install -y cuda-toolkit-${CUDA13VERSION//./-}
ENV PATH=/usr/local/cuda-13/bin:$PATH
ARG PARALLEL
RUN --mount=type=cache,target=/root/.ccache \
cmake --preset 'CUDA 13' -DOLLAMA_RUNNER_DIR="cuda_v13" \
&& cmake --build --parallel ${PARALLEL} --preset 'CUDA 13' \
&& cmake --install build --component CUDA --strip --parallel ${PARALLEL}
FROM base AS rocm-6
ENV PATH=/opt/rocm/hcc/bin:/opt/rocm/hip/bin:/opt/rocm/bin:/opt/rocm/hcc/bin:$PATH
ARG PARALLEL
RUN --mount=type=cache,target=/root/.ccache \
cmake --preset 'ROCm 6' -DOLLAMA_RUNNER_DIR="rocm" \
&& cmake --build --parallel ${PARALLEL} --preset 'ROCm 6' \
&& cmake --install build --component HIP --strip --parallel ${PARALLEL}
FROM --platform=linux/arm64 nvcr.io/nvidia/l4t-jetpack:${JETPACK5VERSION} AS jetpack-5
ARG CMAKEVERSION
RUN apt-get update && apt-get install -y curl ccache \
&& curl -fsSL https://github.com/Kitware/CMake/releases/download/v${CMAKEVERSION}/cmake-${CMAKEVERSION}-linux-$(uname -m).tar.gz | tar xz -C /usr/local --strip-components 1
COPY CMakeLists.txt CMakePresets.json .
COPY ml/backend/ggml/ggml ml/backend/ggml/ggml
ARG PARALLEL
RUN --mount=type=cache,target=/root/.ccache \
cmake --preset 'JetPack 5' -DOLLAMA_RUNNER_DIR="cuda_jetpack5" \
&& cmake --build --parallel ${PARALLEL} --preset 'JetPack 5' \
&& cmake --install build --component CUDA --strip --parallel ${PARALLEL}
FROM --platform=linux/arm64 nvcr.io/nvidia/l4t-jetpack:${JETPACK6VERSION} AS jetpack-6
ARG CMAKEVERSION
RUN apt-get update && apt-get install -y curl ccache \
&& curl -fsSL https://github.com/Kitware/CMake/releases/download/v${CMAKEVERSION}/cmake-${CMAKEVERSION}-linux-$(uname -m).tar.gz | tar xz -C /usr/local --strip-components 1
COPY CMakeLists.txt CMakePresets.json .
COPY ml/backend/ggml/ggml ml/backend/ggml/ggml
ARG PARALLEL
RUN --mount=type=cache,target=/root/.ccache \
cmake --preset 'JetPack 6' -DOLLAMA_RUNNER_DIR="cuda_jetpack6" \
&& cmake --build --parallel ${PARALLEL} --preset 'JetPack 6' \
&& cmake --install build --component CUDA --strip --parallel ${PARALLEL}
FROM base AS build
WORKDIR /go/src/github.com/ollama/ollama
COPY go.mod go.sum .
RUN curl -fsSL https://golang.org/dl/go$(awk '/^go/ { print $2 }' go.mod).linux-$(case $(uname -m) in x86_64) echo amd64 ;; aarch64) echo arm64 ;; esac).tar.gz | tar xz -C /usr/local
ENV PATH=/usr/local/go/bin:$PATH
RUN go mod download
COPY . .
ARG GOFLAGS="'-ldflags=-w -s'"
ENV CGO_ENABLED=1
ARG CGO_CFLAGS
ARG CGO_CXXFLAGS
RUN --mount=type=cache,target=/root/.cache/go-build \
go build -trimpath -buildmode=pie -o /bin/ollama .
FROM --platform=linux/amd64 scratch AS amd64
# COPY --from=cuda-11 dist/lib/ollama/ /lib/ollama/
COPY --from=cuda-12 dist/lib/ollama /lib/ollama/
COPY --from=cuda-13 dist/lib/ollama/ /lib/ollama/
FROM --platform=linux/arm64 scratch AS arm64
# COPY --from=cuda-11 dist/lib/ollama/ /lib/ollama/
COPY --from=cuda-12 dist/lib/ollama /lib/ollama/
COPY --from=cuda-13 dist/lib/ollama/ /lib/ollama/
COPY --from=jetpack-5 dist/lib/ollama/ /lib/ollama/
COPY --from=jetpack-6 dist/lib/ollama/ /lib/ollama/
FROM scratch AS rocm
COPY --from=rocm-6 dist/lib/ollama /lib/ollama
FROM ${FLAVOR} AS archive
COPY --from=cpu dist/lib/ollama /lib/ollama
COPY --from=build /bin/ollama /bin/ollama
FROM ubuntu:24.04
RUN apt-get update \
&& apt-get install -y ca-certificates \
&& apt-get clean \
&& rm -rf /var/lib/apt/lists/*
COPY --from=archive /bin /usr/bin
ENV PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
COPY --from=archive /lib/ollama /usr/lib/ollama
ENV LD_LIBRARY_PATH=/usr/local/nvidia/lib:/usr/local/nvidia/lib64
ENV NVIDIA_DRIVER_CAPABILITIES=compute,utility
ENV NVIDIA_VISIBLE_DEVICES=all
ENV OLLAMA_HOST=0.0.0.0:11434
EXPOSE 11434
ENTRYPOINT ["/bin/ollama"]
CMD ["serve"]
Reference in New Issue View Git Blame Copy Permalink