gered/ollama
1
0
Fork 0
Code Issues Pull requests Actions Packages Projects Releases Wiki Activity

Changed to bild for rx580 GPU, this uses 5.7.1 rocm libraries

Browse source
This commit is contained in:
Matt 2024-09-15 14:59:52 +01:00
parent 56318fb365
commit 3449201ce4
6 changed files with 76 additions and 74 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

138
Dockerfile
View file

@ -4,7 +4,7 @@ ARG CUDA_VERSION_11=11.3.1
ARG CUDA_V11_ARCHITECTURES="50;52;53;60;61;62;70;72;75;80;86" ARG CUDA_V11_ARCHITECTURES="50;52;53;60;61;62;70;72;75;80;86"
ARG CUDA_VERSION_12=12.4.0 ARG CUDA_VERSION_12=12.4.0
ARG CUDA_V12_ARCHITECTURES="60;61;62;70;72;75;80;86;87;89;90;90a" ARG CUDA_V12_ARCHITECTURES="60;61;62;70;72;75;80;86;87;89;90;90a"
ARG ROCM_VERSION=6.1.2 ARG ROCM_VERSION=5.7.1
# Copy the minimal context we need to run the generate scripts # Copy the minimal context we need to run the generate scripts
FROM scratch AS llm-code FROM scratch AS llm-code
@ -47,39 +47,39 @@ RUN --mount=type=cache,target=/root/.ccache \
OLLAMA_CUSTOM_CUDA_DEFS="-DGGML_CUDA_USE_GRAPHS=on" \ OLLAMA_CUSTOM_CUDA_DEFS="-DGGML_CUDA_USE_GRAPHS=on" \
bash gen_linux.sh bash gen_linux.sh
FROM --platform=linux/arm64 nvidia/cuda:$CUDA_VERSION_11-devel-rockylinux8 AS cuda-11-build-server-arm64 #FROM --platform=linux/arm64 nvidia/cuda:$CUDA_VERSION_11-devel-rockylinux8 AS cuda-11-build-server-arm64
ARG CMAKE_VERSION #ARG CMAKE_VERSION
COPY ./scripts/rh_linux_deps.sh / #COPY ./scripts/rh_linux_deps.sh /
RUN CMAKE_VERSION=${CMAKE_VERSION} sh /rh_linux_deps.sh #RUN CMAKE_VERSION=${CMAKE_VERSION} sh /rh_linux_deps.sh
ENV PATH /opt/rh/gcc-toolset-10/root/usr/bin:$PATH #ENV PATH /opt/rh/gcc-toolset-10/root/usr/bin:$PATH
COPY --from=llm-code / /go/src/github.com/ollama/ollama/ #COPY --from=llm-code / /go/src/github.com/ollama/ollama/
WORKDIR /go/src/github.com/ollama/ollama/llm/generate #WORKDIR /go/src/github.com/ollama/ollama/llm/generate
ARG CGO_CFLAGS #ARG CGO_CFLAGS
ARG CUDA_V11_ARCHITECTURES #ARG CUDA_V11_ARCHITECTURES
ENV GOARCH arm64 #ENV GOARCH arm64
RUN OLLAMA_SKIP_STATIC_GENERATE=1 \ #RUN OLLAMA_SKIP_STATIC_GENERATE=1 \
OLLAMA_SKIP_CPU_GENERATE=1 \ # OLLAMA_SKIP_CPU_GENERATE=1 \
CMAKE_CUDA_ARCHITECTURES="${CUDA_V11_ARCHITECTURES}" \ # CMAKE_CUDA_ARCHITECTURES="${CUDA_V11_ARCHITECTURES}" \
CUDA_VARIANT="_v11" \ # CUDA_VARIANT="_v11" \
bash gen_linux.sh # bash gen_linux.sh
FROM --platform=linux/arm64 nvidia/cuda:$CUDA_VERSION_12-devel-rockylinux8 AS cuda-12-build-server-arm64 #FROM --platform=linux/arm64 nvidia/cuda:$CUDA_VERSION_12-devel-rockylinux8 AS cuda-12-build-server-arm64
ARG CMAKE_VERSION #ARG CMAKE_VERSION
COPY ./scripts/rh_linux_deps.sh / #COPY ./scripts/rh_linux_deps.sh /
RUN CMAKE_VERSION=${CMAKE_VERSION} sh /rh_linux_deps.sh #RUN CMAKE_VERSION=${CMAKE_VERSION} sh /rh_linux_deps.sh
ENV PATH /opt/rh/gcc-toolset-10/root/usr/bin:$PATH #ENV PATH /opt/rh/gcc-toolset-10/root/usr/bin:$PATH
COPY --from=llm-code / /go/src/github.com/ollama/ollama/ #COPY --from=llm-code / /go/src/github.com/ollama/ollama/
WORKDIR /go/src/github.com/ollama/ollama/llm/generate #WORKDIR /go/src/github.com/ollama/ollama/llm/generate
ARG CGO_CFLAGS #ARG CGO_CFLAGS
ARG CUDA_V12_ARCHITECTURES #ARG CUDA_V12_ARCHITECTURES
ENV GOARCH arm64 #ENV GOARCH arm64
RUN --mount=type=cache,target=/root/.ccache \ #RUN --mount=type=cache,target=/root/.ccache \
OLLAMA_SKIP_STATIC_GENERATE=1 \ # OLLAMA_SKIP_STATIC_GENERATE=1 \
OLLAMA_SKIP_CPU_GENERATE=1 \ # OLLAMA_SKIP_CPU_GENERATE=1 \
CMAKE_CUDA_ARCHITECTURES="${CUDA_V12_ARCHITECTURES}" \ # CMAKE_CUDA_ARCHITECTURES="${CUDA_V12_ARCHITECTURES}" \
CUDA_VARIANT="_v12" \ # CUDA_VARIANT="_v12" \
OLLAMA_CUSTOM_CUDA_DEFS="-DGGML_CUDA_USE_GRAPHS=on" \ # OLLAMA_CUSTOM_CUDA_DEFS="-DGGML_CUDA_USE_GRAPHS=on" \
bash gen_linux.sh # bash gen_linux.sh
FROM --platform=linux/amd64 rocm/dev-centos-7:${ROCM_VERSION}-complete AS rocm-build-amd64 FROM --platform=linux/amd64 rocm/dev-centos-7:${ROCM_VERSION}-complete AS rocm-build-amd64
@ -123,24 +123,24 @@ FROM --platform=linux/amd64 cpu-builder-amd64 AS cpu_avx2-build-amd64
RUN --mount=type=cache,target=/root/.ccache \ RUN --mount=type=cache,target=/root/.ccache \
OLLAMA_SKIP_STATIC_GENERATE=1 OLLAMA_CPU_TARGET="cpu_avx2" bash gen_linux.sh OLLAMA_SKIP_STATIC_GENERATE=1 OLLAMA_CPU_TARGET="cpu_avx2" bash gen_linux.sh
FROM --platform=linux/arm64 rockylinux:8 AS cpu-builder-arm64 #FROM --platform=linux/arm64 rockylinux:8 AS cpu-builder-arm64
ARG CMAKE_VERSION #ARG CMAKE_VERSION
ARG GOLANG_VERSION #ARG GOLANG_VERSION
COPY ./scripts/rh_linux_deps.sh / #COPY ./scripts/rh_linux_deps.sh /
RUN CMAKE_VERSION=${CMAKE_VERSION} GOLANG_VERSION=${GOLANG_VERSION} sh /rh_linux_deps.sh #RUN CMAKE_VERSION=${CMAKE_VERSION} GOLANG_VERSION=${GOLANG_VERSION} sh /rh_linux_deps.sh
ENV PATH /opt/rh/gcc-toolset-10/root/usr/bin:$PATH #ENV PATH /opt/rh/gcc-toolset-10/root/usr/bin:$PATH
COPY --from=llm-code / /go/src/github.com/ollama/ollama/ #COPY --from=llm-code / /go/src/github.com/ollama/ollama/
ARG OLLAMA_CUSTOM_CPU_DEFS #ARG OLLAMA_CUSTOM_CPU_DEFS
ARG CGO_CFLAGS #ARG CGO_CFLAGS
ENV GOARCH arm64 #ENV GOARCH arm64
WORKDIR /go/src/github.com/ollama/ollama/llm/generate #WORKDIR /go/src/github.com/ollama/ollama/llm/generate
FROM --platform=linux/arm64 cpu-builder-arm64 AS static-build-arm64 #FROM --platform=linux/arm64 cpu-builder-arm64 AS static-build-arm64
RUN --mount=type=cache,target=/root/.ccache \ #RUN --mount=type=cache,target=/root/.ccache \
OLLAMA_CPU_TARGET="static" bash gen_linux.sh # OLLAMA_CPU_TARGET="static" bash gen_linux.sh
FROM --platform=linux/arm64 cpu-builder-arm64 AS cpu-build-arm64 #FROM --platform=linux/arm64 cpu-builder-arm64 AS cpu-build-arm64
RUN --mount=type=cache,target=/root/.ccache \ #RUN --mount=type=cache,target=/root/.ccache \
OLLAMA_SKIP_STATIC_GENERATE=1 OLLAMA_CPU_TARGET="cpu" bash gen_linux.sh # OLLAMA_SKIP_STATIC_GENERATE=1 OLLAMA_CPU_TARGET="cpu" bash gen_linux.sh
# Intermediate stage used for ./scripts/build_linux.sh # Intermediate stage used for ./scripts/build_linux.sh
@ -163,20 +163,20 @@ RUN --mount=type=cache,target=/root/.ccache \
go build -trimpath -o dist/linux-amd64/bin/ollama . go build -trimpath -o dist/linux-amd64/bin/ollama .
# Intermediate stage used for ./scripts/build_linux.sh # Intermediate stage used for ./scripts/build_linux.sh
FROM --platform=linux/arm64 cpu-build-arm64 AS build-arm64 #FROM --platform=linux/arm64 cpu-build-arm64 AS build-arm64
ENV CGO_ENABLED 1 #ENV CGO_ENABLED 1
ARG GOLANG_VERSION #ARG GOLANG_VERSION
WORKDIR /go/src/github.com/ollama/ollama #WORKDIR /go/src/github.com/ollama/ollama
COPY . . #COPY . .
COPY --from=static-build-arm64 /go/src/github.com/ollama/ollama/llm/build/linux/ llm/build/linux/ #COPY --from=static-build-arm64 /go/src/github.com/ollama/ollama/llm/build/linux/ llm/build/linux/
COPY --from=cuda-11-build-server-arm64 /go/src/github.com/ollama/ollama/dist/ dist/ #COPY --from=cuda-11-build-server-arm64 /go/src/github.com/ollama/ollama/dist/ dist/
COPY --from=cuda-11-build-server-arm64 /go/src/github.com/ollama/ollama/llm/build/linux/ llm/build/linux/ #COPY --from=cuda-11-build-server-arm64 /go/src/github.com/ollama/ollama/llm/build/linux/ llm/build/linux/
COPY --from=cuda-12-build-server-arm64 /go/src/github.com/ollama/ollama/dist/ dist/ #COPY --from=cuda-12-build-server-arm64 /go/src/github.com/ollama/ollama/dist/ dist/
COPY --from=cuda-12-build-server-arm64 /go/src/github.com/ollama/ollama/llm/build/linux/ llm/build/linux/ #COPY --from=cuda-12-build-server-arm64 /go/src/github.com/ollama/ollama/llm/build/linux/ llm/build/linux/
ARG GOFLAGS #ARG GOFLAGS
ARG CGO_CFLAGS #ARG CGO_CFLAGS
RUN --mount=type=cache,target=/root/.ccache \ #RUN --mount=type=cache,target=/root/.ccache \
go build -trimpath -o dist/linux-arm64/bin/ollama . # go build -trimpath -o dist/linux-arm64/bin/ollama .
# Strip out ROCm dependencies to keep the primary image lean # Strip out ROCm dependencies to keep the primary image lean
FROM --platform=linux/amd64 ubuntu:22.04 as amd64-libs-without-rocm FROM --platform=linux/amd64 ubuntu:22.04 as amd64-libs-without-rocm
@ -190,11 +190,11 @@ RUN apt-get update && apt-get install -y ca-certificates && \
apt-get clean && rm -rf /var/lib/apt/lists/* apt-get clean && rm -rf /var/lib/apt/lists/*
COPY --from=build-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64/bin/ /bin/ COPY --from=build-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64/bin/ /bin/
FROM --platform=linux/arm64 ubuntu:22.04 as runtime-arm64 #FROM --platform=linux/arm64 ubuntu:22.04 as runtime-arm64
COPY --from=build-arm64 /go/src/github.com/ollama/ollama/dist/linux-arm64/lib/ /lib/ #COPY --from=build-arm64 /go/src/github.com/ollama/ollama/dist/linux-arm64/lib/ /lib/
RUN apt-get update && apt-get install -y ca-certificates && \ #RUN apt-get update && apt-get install -y ca-certificates && \
apt-get clean && rm -rf /var/lib/apt/lists/* # apt-get clean && rm -rf /var/lib/apt/lists/*
COPY --from=build-arm64 /go/src/github.com/ollama/ollama/dist/linux-arm64/bin/ /bin/ #COPY --from=build-arm64 /go/src/github.com/ollama/ollama/dist/linux-arm64/bin/ /bin/
# Radeon images are much larger so we keep it distinct from the CPU/CUDA image # Radeon images are much larger so we keep it distinct from the CPU/CUDA image
FROM --platform=linux/amd64 rocm/dev-centos-7:${ROCM_VERSION}-complete as runtime-rocm FROM --platform=linux/amd64 rocm/dev-centos-7:${ROCM_VERSION}-complete as runtime-rocm

4
gpu/amd_linux.go
View file

@ -41,7 +41,7 @@ const (
var ( var (
// Used to validate if the given ROCm lib is usable // Used to validate if the given ROCm lib is usable
ROCmLibGlobs = []string{"libhipblas.so.2*", "rocblas"} // TODO - probably include more coverage of files here... ROCmLibGlobs = []string{"libhipblas.so.1*", "rocblas"} // TODO - probably include more coverage of files here...
RocmStandardLocations = []string{"/opt/rocm/lib", "/usr/lib64"} RocmStandardLocations = []string{"/opt/rocm/lib", "/usr/lib64"}
) )
@ -386,7 +386,7 @@ func AMDValidateLibDir() (string, error) {
} }
// Well known ollama installer path // Well known ollama installer path
installedRocmDir := "/usr/share/ollama/lib/rocm" installedRocmDir := "/opt/rocm-5.7.1"
if rocmLibUsable(installedRocmDir) { if rocmLibUsable(installedRocmDir) {
return installedRocmDir, nil return installedRocmDir, nil
} }

2
gpu/gpu.go
View file

@ -59,7 +59,7 @@ var (
// With our current CUDA compile flags, older than 5.0 will not work properly // With our current CUDA compile flags, older than 5.0 will not work properly
var CudaComputeMin = [2]C.int{5, 0} var CudaComputeMin = [2]C.int{5, 0}
var RocmComputeMin = 9 var RocmComputeMin = 8
// TODO find a better way to detect iGPU instead of minimum memory // TODO find a better way to detect iGPU instead of minimum memory
const IGPUMemLimit = 1 * format.GibiByte // 512G is what they typically report, so anything less than 1G must be iGPU const IGPUMemLimit = 1 * format.GibiByte // 512G is what they typically report, so anything less than 1G must be iGPU

1
llm/generate/gen_linux.sh
View file

@ -22,6 +22,7 @@ amdGPUs() {
return return
fi fi
GPU_LIST=( GPU_LIST=(
"gfx803"
"gfx900" "gfx900"
"gfx906:xnack-" "gfx906:xnack-"
"gfx908:xnack-" "gfx908:xnack-"

2
scripts/build_docker.sh
View file

@ -13,7 +13,7 @@ DOCKER_ORG=${DOCKER_ORG:-"ollama"}
RELEASE_IMAGE_REPO=${RELEASE_IMAGE_REPO:-"${DOCKER_ORG}/release"} RELEASE_IMAGE_REPO=${RELEASE_IMAGE_REPO:-"${DOCKER_ORG}/release"}
FINAL_IMAGE_REPO=${FINAL_IMAGE_REPO:-"${DOCKER_ORG}/ollama"} FINAL_IMAGE_REPO=${FINAL_IMAGE_REPO:-"${DOCKER_ORG}/ollama"}
BUILD_ARCH=${BUILD_ARCH:-"amd64 arm64"} BUILD_ARCH=${BUILD_ARCH:-"amd64"}
# Set PUSH to a non-empty string to trigger push instead of load # Set PUSH to a non-empty string to trigger push instead of load
PUSH=${PUSH:-""} PUSH=${PUSH:-""}

3
scripts/build_linux.sh
View file

@ -6,7 +6,8 @@ export VERSION=${VERSION:-$(git describe --tags --first-parent --abbrev=7 --long
export GOFLAGS="'-ldflags=-w -s \"-X=github.com/ollama/ollama/version.Version=$VERSION\" \"-X=github.com/ollama/ollama/server.mode=release\"'" export GOFLAGS="'-ldflags=-w -s \"-X=github.com/ollama/ollama/version.Version=$VERSION\" \"-X=github.com/ollama/ollama/server.mode=release\"'"
GZIP=$(which pigz 2>/dev/null || echo "gzip") GZIP=$(which pigz 2>/dev/null || echo "gzip")
BUILD_ARCH=${BUILD_ARCH:-"amd64 arm64"} #BUILD_ARCH=${BUILD_ARCH:-"amd64 arm64"}
BUILD_ARCH=${BUILD_ARCH:-"amd64"}
export AMDGPU_TARGETS=${AMDGPU_TARGETS:=""} export AMDGPU_TARGETS=${AMDGPU_TARGETS:=""}
mkdir -p dist mkdir -p dist