Skip to content

Commit f732bc4

Browse files
wwoodsTMwwoodsTM
committed
Merge branch 'master' into dry-sampling-post-refactor
2 parents 4dd25e8 + 7eee341 commit f732bc4

File tree

142 files changed

+8198
-7290
lines changed

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

142 files changed

+8198
-7290
lines changed

‎.devops/full-musa.Dockerfile‎

Lines changed: 26 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,26 @@
1+
ARG UBUNTU_VERSION=22.04
2+
# This needs to generally match the container host's environment.
3+
ARG MUSA_VERSION=rc3.1.0
4+
# Target the MUSA build image
5+
ARG BASE_MUSA_DEV_CONTAINER=mthreads/musa:${MUSA_VERSION}-devel-ubuntu${UBUNTU_VERSION}
6+
7+
FROM ${BASE_MUSA_DEV_CONTAINER} AS build
8+
9+
RUN apt-get update && \
10+
apt-get install -y build-essential cmake python3 python3-pip git libcurl4-openssl-dev libgomp1
11+
12+
COPY requirements.txt requirements.txt
13+
COPY requirements requirements
14+
15+
RUN pip install --upgrade pip setuptools wheel \
16+
&& pip install -r requirements.txt
17+
18+
WORKDIR /app
19+
20+
COPY . .
21+
22+
RUN cmake -B build -DGGML_MUSA=ON -DLLAMA_CURL=ON ${CMAKE_ARGS} -DCMAKE_EXE_LINKER_FLAGS=-Wl,--allow-shlib-undefined . && \
23+
cmake --build build --config Release -j$(nproc) && \
24+
cp build/bin/* .
25+
26+
ENTRYPOINT ["/app/.devops/tools.sh"]

‎.devops/full-rocm.Dockerfile‎

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -11,7 +11,7 @@ FROM ${BASE_ROCM_DEV_CONTAINER} AS build
1111
# Unless otherwise specified, we make a fat build.
1212
# List from https://github.com/ggerganov/llama.cpp/pull/1087#issuecomment-1682807878
1313
# This is mostly tied to rocBLAS supported archs.
14-
ARG ROCM_DOCKER_ARCH=\
14+
ARG ROCM_DOCKER_ARCH="\
1515
gfx803 \
1616
gfx900 \
1717
gfx906 \
@@ -21,7 +21,7 @@ ARG ROCM_DOCKER_ARCH=\
2121
gfx1030 \
2222
gfx1100 \
2323
gfx1101 \
24-
gfx1102
24+
gfx1102"
2525

2626
COPY requirements.txt requirements.txt
2727
COPY requirements requirements
@@ -34,7 +34,7 @@ WORKDIR /app
3434
COPY . .
3535

3636
# Set nvcc architecture
37-
ENV GPU_TARGETS=${ROCM_DOCKER_ARCH}
37+
ENV AMDGPU_TARGETS=${ROCM_DOCKER_ARCH}
3838
# Enable ROCm
3939
ENV GGML_HIPBLAS=1
4040
ENV CC=/opt/rocm/llvm/bin/clang

‎.devops/llama-cli-musa.Dockerfile‎

Lines changed: 30 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,30 @@
1+
ARG UBUNTU_VERSION=22.04
2+
# This needs to generally match the container host's environment.
3+
ARG MUSA_VERSION=rc3.1.0
4+
# Target the MUSA build image
5+
ARG BASE_MUSA_DEV_CONTAINER=mthreads/musa:${MUSA_VERSION}-devel-ubuntu${UBUNTU_VERSION}
6+
# Target the MUSA runtime image
7+
ARG BASE_MUSA_RUN_CONTAINER=mthreads/musa:${MUSA_VERSION}-runtime-ubuntu${UBUNTU_VERSION}
8+
9+
FROM ${BASE_MUSA_DEV_CONTAINER} AS build
10+
11+
RUN apt-get update && \
12+
apt-get install -y build-essential git cmake
13+
14+
WORKDIR /app
15+
16+
COPY . .
17+
18+
RUN cmake -B build -DGGML_MUSA=ON ${CMAKE_ARGS} -DCMAKE_EXE_LINKER_FLAGS=-Wl,--allow-shlib-undefined . && \
19+
cmake --build build --config Release --target llama-cli -j$(nproc)
20+
21+
FROM ${BASE_MUSA_RUN_CONTAINER} AS runtime
22+
23+
RUN apt-get update && \
24+
apt-get install -y libgomp1
25+
26+
COPY --from=build /app/build/ggml/src/libggml.so /libggml.so
27+
COPY --from=build /app/build/src/libllama.so /libllama.so
28+
COPY --from=build /app/build/bin/llama-cli /llama-cli
29+
30+
ENTRYPOINT [ "/llama-cli" ]

‎.devops/llama-cli-rocm.Dockerfile‎

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -11,7 +11,7 @@ FROM ${BASE_ROCM_DEV_CONTAINER} AS build
1111
# Unless otherwise specified, we make a fat build.
1212
# List from https://github.com/ggerganov/llama.cpp/pull/1087#issuecomment-1682807878
1313
# This is mostly tied to rocBLAS supported archs.
14-
ARG ROCM_DOCKER_ARCH=\
14+
ARG ROCM_DOCKER_ARCH="\
1515
gfx803 \
1616
gfx900 \
1717
gfx906 \
@@ -21,7 +21,7 @@ ARG ROCM_DOCKER_ARCH=\
2121
gfx1030 \
2222
gfx1100 \
2323
gfx1101 \
24-
gfx1102
24+
gfx1102"
2525

2626
COPY requirements.txt requirements.txt
2727
COPY requirements requirements
@@ -34,7 +34,7 @@ WORKDIR /app
3434
COPY . .
3535

3636
# Set nvcc architecture
37-
ENV GPU_TARGETS=${ROCM_DOCKER_ARCH}
37+
ENV AMDGPU_TARGETS=${ROCM_DOCKER_ARCH}
3838
# Enable ROCm
3939
ENV GGML_HIPBLAS=1
4040
ENV CC=/opt/rocm/llvm/bin/clang

‎.devops/llama-server-musa.Dockerfile‎

Lines changed: 35 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,35 @@
1+
ARG UBUNTU_VERSION=22.04
2+
# This needs to generally match the container host's environment.
3+
ARG MUSA_VERSION=rc3.1.0
4+
# Target the MUSA build image
5+
ARG BASE_MUSA_DEV_CONTAINER=mthreads/musa:${MUSA_VERSION}-devel-ubuntu${UBUNTU_VERSION}
6+
# Target the MUSA runtime image
7+
ARG BASE_MUSA_RUN_CONTAINER=mthreads/musa:${MUSA_VERSION}-runtime-ubuntu${UBUNTU_VERSION}
8+
9+
FROM ${BASE_MUSA_DEV_CONTAINER} AS build
10+
11+
RUN apt-get update && \
12+
apt-get install -y build-essential git cmake libcurl4-openssl-dev
13+
14+
WORKDIR /app
15+
16+
COPY . .
17+
18+
RUN cmake -B build -DGGML_MUSA=ON -DLLAMA_CURL=ON ${CMAKE_ARGS} -DCMAKE_EXE_LINKER_FLAGS=-Wl,--allow-shlib-undefined . && \
19+
cmake --build build --config Release --target llama-server -j$(nproc)
20+
21+
FROM ${BASE_MUSA_RUN_CONTAINER} AS runtime
22+
23+
RUN apt-get update && \
24+
apt-get install -y libcurl4-openssl-dev libgomp1 curl
25+
26+
COPY --from=build /app/build/ggml/src/libggml.so /libggml.so
27+
COPY --from=build /app/build/src/libllama.so /libllama.so
28+
COPY --from=build /app/build/bin/llama-server /llama-server
29+
30+
# Must be set to 0.0.0.0 so it can listen to requests from host machine
31+
ENV LLAMA_ARG_HOST=0.0.0.0
32+
33+
HEALTHCHECK CMD [ "curl", "-f", "http://localhost:8080/health" ]
34+
35+
ENTRYPOINT [ "/llama-server" ]

‎.devops/llama-server-rocm.Dockerfile‎

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -11,7 +11,7 @@ FROM ${BASE_ROCM_DEV_CONTAINER} AS build
1111
# Unless otherwise specified, we make a fat build.
1212
# List from https://github.com/ggerganov/llama.cpp/pull/1087#issuecomment-1682807878
1313
# This is mostly tied to rocBLAS supported archs.
14-
ARG ROCM_DOCKER_ARCH=\
14+
ARG ROCM_DOCKER_ARCH="\
1515
gfx803 \
1616
gfx900 \
1717
gfx906 \
@@ -21,7 +21,7 @@ ARG ROCM_DOCKER_ARCH=\
2121
gfx1030 \
2222
gfx1100 \
2323
gfx1101 \
24-
gfx1102
24+
gfx1102"
2525

2626
COPY requirements.txt requirements.txt
2727
COPY requirements requirements
@@ -34,7 +34,7 @@ WORKDIR /app
3434
COPY . .
3535

3636
# Set nvcc architecture
37-
ENV GPU_TARGETS=${ROCM_DOCKER_ARCH}
37+
ENV AMDGPU_TARGETS=${ROCM_DOCKER_ARCH}
3838
# Enable ROCm
3939
ENV GGML_HIPBLAS=1
4040
ENV CC=/opt/rocm/llvm/bin/clang

‎.github/workflows/bench.yml.disabled‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -27,10 +27,10 @@ on:
2727
push:
2828
branches:
2929
- master
30-
paths: ['llama.cpp', 'ggml.c', 'ggml-backend.c', 'ggml-quants.c', '**/*.cu', 'examples/server/*.h*', 'examples/server/*.cpp']
30+
paths: ['llama.cpp', 'ggml.c', 'ggml-backend.cpp', 'ggml-quants.c', '**/*.cu', 'examples/server/*.h*', 'examples/server/*.cpp']
3131
pull_request_target:
3232
types: [opened, synchronize, reopened]
33-
paths: ['llama.cpp', 'ggml.c', 'ggml-backend.c', 'ggml-quants.c', '**/*.cu', 'examples/server/*.h*', 'examples/server/*.cpp']
33+
paths: ['llama.cpp', 'ggml.c', 'ggml-backend.cpp', 'ggml-quants.c', '**/*.cu', 'examples/server/*.h*', 'examples/server/*.cpp']
3434
schedule:
3535
- cron: '04 2 * * *'
3636

‎.github/workflows/build.yml‎

Lines changed: 6 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -19,6 +19,11 @@ concurrency:
1919
group: ${{ github.workflow }}-${{ github.head_ref && github.ref || github.run_id }}
2020
cancel-in-progress: true
2121

22+
# Fine-grant permission
23+
# https://docs.github.com/en/actions/security-for-github-actions/security-guides/automatic-token-authentication#modifying-the-permissions-for-the-github_token
24+
permissions:
25+
contents: write # for creating release
26+
2227
env:
2328
BRANCH_NAME: ${{ github.head_ref || github.ref_name }}
2429
GGML_NLOOP: 3
@@ -1032,7 +1037,7 @@ jobs:
10321037
run: |
10331038
$env:HIP_PATH=$(Resolve-Path 'C:\Program Files\AMD\ROCm\*\bin\clang.exe' | split-path | split-path)
10341039
$env:CMAKE_PREFIX_PATH="${env:HIP_PATH}"
1035-
cmake -G "Unix Makefiles" -B build -S . -DCMAKE_C_COMPILER="${env:HIP_PATH}\bin\clang.exe" -DCMAKE_CXX_COMPILER="${env:HIP_PATH}\bin\clang++.exe" -DGGML_HIPBLAS=ON -DCMAKE_BUILD_TYPE=Release -DGPU_TARGETS=${{ matrix.gpu_target }} -DGGML_RPC=ON
1040+
cmake -G "Unix Makefiles" -B build -S . -DCMAKE_C_COMPILER="${env:HIP_PATH}\bin\clang.exe" -DCMAKE_CXX_COMPILER="${env:HIP_PATH}\bin\clang++.exe" -DGGML_HIPBLAS=ON -DCMAKE_BUILD_TYPE=Release -DAMDGPU_TARGETS=${{ matrix.gpu_target }} -DGGML_RPC=ON
10361041
cmake --build build -j ${env:NUMBER_OF_PROCESSORS}
10371042
md "build\bin\rocblas\library\"
10381043
cp "${env:HIP_PATH}\bin\hipblas.dll" "build\bin\"

‎.github/workflows/close-issue.yml‎

Lines changed: 5 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -3,6 +3,11 @@ on:
33
schedule:
44
- cron: "42 0 * * *"
55

6+
# Fine-grant permission
7+
# https://docs.github.com/en/actions/security-for-github-actions/security-guides/automatic-token-authentication#modifying-the-permissions-for-the-github_token
8+
permissions:
9+
issues: write
10+
611
jobs:
712
close-issues:
813
runs-on: ubuntu-latest

‎.github/workflows/docker.yml‎

Lines changed: 3 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -43,6 +43,9 @@ jobs:
4343
- { tag: "light-cuda", dockerfile: ".devops/llama-cli-cuda.Dockerfile", platforms: "linux/amd64" }
4444
- { tag: "server-cuda", dockerfile: ".devops/llama-server-cuda.Dockerfile", platforms: "linux/amd64" }
4545
- { tag: "full-cuda", dockerfile: ".devops/full-cuda.Dockerfile", platforms: "linux/amd64" }
46+
- { tag: "light-musa", dockerfile: ".devops/llama-cli-musa.Dockerfile", platforms: "linux/amd64" }
47+
- { tag: "server-musa", dockerfile: ".devops/llama-server-musa.Dockerfile", platforms: "linux/amd64" }
48+
- { tag: "full-musa", dockerfile: ".devops/full-musa.Dockerfile", platforms: "linux/amd64" }
4649
# Note: the rocm images are failing due to a compiler error and are disabled until this is fixed to allow the workflow to complete
4750
#- { tag: "light-rocm", dockerfile: ".devops/llama-cli-rocm.Dockerfile", platforms: "linux/amd64,linux/arm64" }
4851
#- { tag: "server-rocm", dockerfile: ".devops/llama-server-rocm.Dockerfile", platforms: "linux/amd64,linux/arm64" }

0 commit comments

Comments
 (0)