56 lines
1.5 KiB
Plaintext
56 lines
1.5 KiB
Plaintext
# Stage 1: Build Ollama with Vulkan support
|
|
FROM ubuntu:26.04 AS builder
|
|
|
|
ENV DEBIAN_FRONTEND=noninteractive
|
|
|
|
RUN apt update && apt upgrade -y && apt install -y \
|
|
build-essential \
|
|
git \
|
|
curl \
|
|
wget \
|
|
golang-go \
|
|
libvulkan-dev \
|
|
vulkan-tools \
|
|
glslang-tools \
|
|
ninja-build \
|
|
spirv-headers \
|
|
spirv-tools \
|
|
ca-certificates \
|
|
&& rm -rf /var/lib/apt/lists/*
|
|
|
|
# Directly install glslc via Debian arm64 package
|
|
RUN wget http://ftp.debian.org/debian/pool/main/s/shaderc/glslc_2023.2-1_arm64.deb \
|
|
&& apt-get update && apt-get install -y ./glslc_2023.2-1_arm64.deb \
|
|
&& rm glslc_2023.2-1_arm64.deb && rm -rf /var/lib/apt/lists/*
|
|
|
|
WORKDIR /app
|
|
RUN git clone --recursive https://github.com/ollama/ollama.git .
|
|
|
|
# Configure and safely build with constrained parallelism to protect 96GB RAM
|
|
RUN cmake -B build . -DOLLAMA_LLAMA_BACKENDS="vulkan" -DCMAKE_BUILD_TYPE=Release
|
|
RUN cmake --build build --parallel 16
|
|
|
|
# Build the main Go application binary
|
|
RUN go build -o ollama .
|
|
|
|
# Stage 2: Minimal Runtime Image
|
|
FROM ubuntu:24.04
|
|
|
|
ENV DEBIAN_FRONTEND=noninteractive
|
|
|
|
RUN apt-get update && apt-get install -y \
|
|
libvulkan1 \
|
|
mesa-vulkan-drivers \
|
|
vulkan-tools \
|
|
ca-certificates \
|
|
&& rm -rf /var/lib/apt/lists/*
|
|
|
|
# Copy the compiled Go binary and generated native backend libraries
|
|
COPY --from=builder /app/ollama /usr/local/bin/ollama
|
|
COPY --from=builder /app/build/lib/ollama /usr/local/lib/ollama
|
|
|
|
EXPOSE 11434
|
|
ENV OLLAMA_HOST=0.0.0.0:11434
|
|
ENV OLLAMA_MODELS=/root/.ollama/models
|
|
|
|
ENTRYPOINT ["ollama", "serve"] |