Skip to content

Commit

Permalink
Merge branch 'master' of github.com:ggerganov/llama.cpp into mfalcon_…
Browse files Browse the repository at this point in the history
…mamba_cuda
  • Loading branch information
piDack committed Aug 29, 2024
2 parents 316a049 + 9fe94cc commit 99f2ac1
Show file tree
Hide file tree
Showing 5 changed files with 45 additions and 54 deletions.
25 changes: 11 additions & 14 deletions .devops/full-cuda.Dockerfile
Original file line number Diff line number Diff line change
@@ -1,18 +1,16 @@
ARG UBUNTU_VERSION=22.04

# This needs to generally match the container host's environment.
ARG CUDA_VERSION=11.7.1

ARG CUDA_VERSION=12.6.0
# Target the CUDA build image
ARG BASE_CUDA_DEV_CONTAINER=nvidia/cuda:${CUDA_VERSION}-devel-ubuntu${UBUNTU_VERSION}

FROM ${BASE_CUDA_DEV_CONTAINER} AS build

# Unless otherwise specified, we make a fat build.
ARG CUDA_DOCKER_ARCH=all
# CUDA architecture to build for (defaults to all supported archs)
ARG CUDA_DOCKER_ARCH=default

RUN apt-get update && \
apt-get install -y build-essential python3 python3-pip git libcurl4-openssl-dev libgomp1
apt-get install -y build-essential cmake python3 python3-pip git libcurl4-openssl-dev libgomp1

COPY requirements.txt requirements.txt
COPY requirements requirements
Expand All @@ -24,13 +22,12 @@ WORKDIR /app

COPY . .

# Set nvcc architecture
ENV CUDA_DOCKER_ARCH=${CUDA_DOCKER_ARCH}
# Enable CUDA
ENV GGML_CUDA=1
# Enable cURL
ENV LLAMA_CURL=1

RUN make -j$(nproc)
# Use the default CUDA archs if not specified
RUN if [ "${CUDA_DOCKER_ARCH}" != "default" ]; then \
export CMAKE_ARGS="-DCMAKE_CUDA_ARCHITECTURES=${CUDA_DOCKER_ARCH}"; \
fi && \
cmake -B build -DGGML_CUDA=ON -DLLAMA_CURL=ON ${CMAKE_ARGS} -DCMAKE_EXE_LINKER_FLAGS=-Wl,--allow-shlib-undefined . && \
cmake --build build --config Release --target llama-cli -j$(nproc) && \
cp build/bin/* .

ENTRYPOINT ["/app/.devops/tools.sh"]
24 changes: 13 additions & 11 deletions .devops/llama-cli-cuda.Dockerfile
Original file line number Diff line number Diff line change
@@ -1,35 +1,37 @@
ARG UBUNTU_VERSION=22.04
# This needs to generally match the container host's environment.
ARG CUDA_VERSION=11.7.1
ARG CUDA_VERSION=12.6.0
# Target the CUDA build image
ARG BASE_CUDA_DEV_CONTAINER=nvidia/cuda:${CUDA_VERSION}-devel-ubuntu${UBUNTU_VERSION}
# Target the CUDA runtime image
ARG BASE_CUDA_RUN_CONTAINER=nvidia/cuda:${CUDA_VERSION}-runtime-ubuntu${UBUNTU_VERSION}

FROM ${BASE_CUDA_DEV_CONTAINER} AS build

# Unless otherwise specified, we make a fat build.
ARG CUDA_DOCKER_ARCH=all
# CUDA architecture to build for (defaults to all supported archs)
ARG CUDA_DOCKER_ARCH=default

RUN apt-get update && \
apt-get install -y build-essential git
apt-get install -y build-essential git cmake

WORKDIR /app

COPY . .

# Set nvcc architecture
ENV CUDA_DOCKER_ARCH=${CUDA_DOCKER_ARCH}
# Enable CUDA
ENV GGML_CUDA=1

RUN make -j$(nproc) llama-cli
# Use the default CUDA archs if not specified
RUN if [ "${CUDA_DOCKER_ARCH}" != "default" ]; then \
export CMAKE_ARGS="-DCMAKE_CUDA_ARCHITECTURES=${CUDA_DOCKER_ARCH}"; \
fi && \
cmake -B build -DGGML_CUDA=ON ${CMAKE_ARGS} -DCMAKE_EXE_LINKER_FLAGS=-Wl,--allow-shlib-undefined . && \
cmake --build build --config Release --target llama-cli -j$(nproc)

FROM ${BASE_CUDA_RUN_CONTAINER} AS runtime

RUN apt-get update && \
apt-get install -y libgomp1

COPY --from=build /app/llama-cli /llama-cli
COPY --from=build /app/build/ggml/src/libggml.so /libggml.so
COPY --from=build /app/build/src/libllama.so /libllama.so
COPY --from=build /app/build/bin/llama-cli /llama-cli

ENTRYPOINT [ "/llama-cli" ]
31 changes: 16 additions & 15 deletions .devops/llama-server-cuda.Dockerfile
Original file line number Diff line number Diff line change
@@ -1,40 +1,41 @@
ARG UBUNTU_VERSION=22.04
# This needs to generally match the container host's environment.
ARG CUDA_VERSION=11.7.1
ARG CUDA_VERSION=12.6.0
# Target the CUDA build image
ARG BASE_CUDA_DEV_CONTAINER=nvidia/cuda:${CUDA_VERSION}-devel-ubuntu${UBUNTU_VERSION}
# Target the CUDA runtime image
ARG BASE_CUDA_RUN_CONTAINER=nvidia/cuda:${CUDA_VERSION}-runtime-ubuntu${UBUNTU_VERSION}

FROM ${BASE_CUDA_DEV_CONTAINER} AS build

# Unless otherwise specified, we make a fat build.
ARG CUDA_DOCKER_ARCH=all
# CUDA architecture to build for (defaults to all supported archs)
ARG CUDA_DOCKER_ARCH=default

RUN apt-get update && \
apt-get install -y build-essential git libcurl4-openssl-dev
apt-get install -y build-essential git cmake libcurl4-openssl-dev

WORKDIR /app

COPY . .

# Set nvcc architecture
ENV CUDA_DOCKER_ARCH=${CUDA_DOCKER_ARCH}
# Enable CUDA
ENV GGML_CUDA=1
# Enable cURL
ENV LLAMA_CURL=1
# Must be set to 0.0.0.0 so it can listen to requests from host machine
ENV LLAMA_ARG_HOST=0.0.0.0

RUN make -j$(nproc) llama-server
# Use the default CUDA archs if not specified
RUN if [ "${CUDA_DOCKER_ARCH}" != "default" ]; then \
export CMAKE_ARGS="-DCMAKE_CUDA_ARCHITECTURES=${CUDA_DOCKER_ARCH}"; \
fi && \
cmake -B build -DGGML_CUDA=ON -DLLAMA_CURL=ON ${CMAKE_ARGS} -DCMAKE_EXE_LINKER_FLAGS=-Wl,--allow-shlib-undefined . && \
cmake --build build --config Release --target llama-server -j$(nproc)

FROM ${BASE_CUDA_RUN_CONTAINER} AS runtime

RUN apt-get update && \
apt-get install -y libcurl4-openssl-dev libgomp1 curl

COPY --from=build /app/llama-server /llama-server
COPY --from=build /app/build/ggml/src/libggml.so /libggml.so
COPY --from=build /app/build/src/libllama.so /libllama.so
COPY --from=build /app/build/bin/llama-server /llama-server

# Must be set to 0.0.0.0 so it can listen to requests from host machine
ENV LLAMA_ARG_HOST=0.0.0.0

HEALTHCHECK CMD [ "curl", "-f", "http://localhost:8080/health" ]

Expand Down
15 changes: 3 additions & 12 deletions .github/workflows/docker.yml
Original file line number Diff line number Diff line change
Expand Up @@ -96,21 +96,12 @@ jobs:
env:
GITHUB_REPOSITORY_OWNER: '${{ github.repository_owner }}'

- name: Build and push Docker image (versioned)
- name: Build and push Docker image (tagged + versioned)
if: github.event_name == 'push'
uses: docker/build-push-action@v4
uses: docker/build-push-action@v6
with:
context: .
push: true
platforms: ${{ matrix.config.platforms }}
tags: "ghcr.io/${{ env.repository_owner_lowercase }}/llama.cpp:${{ matrix.config.tag }}-${{ env.COMMIT_SHA }}"
file: ${{ matrix.config.dockerfile }}

- name: Build and push Docker image (tagged)
uses: docker/build-push-action@v4
with:
context: .
push: ${{ github.event_name == 'push' }}
platforms: ${{ matrix.config.platforms }}
tags: "ghcr.io/${{ env.repository_owner_lowercase }}/llama.cpp:${{ matrix.config.tag }},ghcr.io/${{ env.repository_owner_lowercase }}/llama.cpp:${{ matrix.config.tag }}-${{ steps.tag.outputs.name }}"
tags: "ghcr.io/${{ env.repository_owner_lowercase }}/llama.cpp:${{ matrix.config.tag }}-${{ env.COMMIT_SHA }},ghcr.io/${{ env.repository_owner_lowercase }}/llama.cpp:${{ matrix.config.tag }},ghcr.io/${{ env.repository_owner_lowercase }}/llama.cpp:${{ matrix.config.tag }}-${{ steps.tag.outputs.name }}"
file: ${{ matrix.config.dockerfile }}
4 changes: 2 additions & 2 deletions docs/docker.md
Original file line number Diff line number Diff line change
Expand Up @@ -66,8 +66,8 @@ You may want to pass in some different `ARGS`, depending on the CUDA environment

The defaults are:

- `CUDA_VERSION` set to `11.7.1`
- `CUDA_DOCKER_ARCH` set to `all`
- `CUDA_VERSION` set to `12.6.0`
- `CUDA_DOCKER_ARCH` set to the cmake build default, which includes all the supported architectures

The resulting images, are essentially the same as the non-CUDA images:

Expand Down

0 comments on commit 99f2ac1

Please sign in to comment.