nikola/llama_cpp
1
0
Fork 0
Code Issues Pull Requests Actions 32 Packages Projects Releases Wiki Activity

ci : move most slim jobs to self-hosted runners (#23619)

Browse Source 
* ci : remove tag from build-self-hosted.yml

* ci : slim -> self-hosted

* ci : prevent heavy CPU jobs from running on fast runners

* ci : prevent cmake pkg to run on dedicated fast runners

* ci : try to bump 3.11 -> 3.13

* ci : move lint back to 3.11

* ci : back to 3.11

* ci : add comment about UI jobs

* ci : move python requirements check to CPU runners

this job is a bit slow for a dedicated "fast" runner

* ci : add self-hosted ui workflow

* ci : fix UI naming

* tmp to check if arm64 fast is compatible with all jobs

* revert last commit
This commit is contained in:
Georgi Gerganov
2026-05-25 08:11:19 +03:00
committed by GitHub
parent 549b9d8433
commit 28123a3937
13 changed files with 140 additions and 74 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.github/workflows/build-self-hosted.yml
+4 -54
View File
@@ -55,24 +55,7 @@ env:
LLAMA_LOG_TIMESTAMPS: 1
jobs:
determine-tag:
name: Determine tag name
runs-on: [self-hosted, fast]
outputs:
tag_name: ${{ steps.tag.outputs.name }}
steps:
- name: Clone
uses: actions/checkout@v6
with:
fetch-depth: 0
- name: Determine tag name
id: tag
uses: ./.github/actions/get-tag-name
env:
BRANCH_NAME: ${{ github.head_ref || github.ref_name }}
ggml-ci-nvidia-cuda:
needs: determine-tag
runs-on: [self-hosted, Linux, NVIDIA]
steps:
@@ -82,14 +65,11 @@ jobs:
- name: Test
id: ggml-ci
env:
HF_UI_VERSION: ${{ needs.determine-tag.outputs.tag_name }}
run: |
nvidia-smi
GG_BUILD_CUDA=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
ggml-ci-nvidia-vulkan-cm:
needs: determine-tag
runs-on: [self-hosted, Linux, NVIDIA]
steps:
@@ -99,14 +79,11 @@ jobs:
- name: Test
id: ggml-ci
env:
HF_UI_VERSION: ${{ needs.determine-tag.outputs.tag_name }}
run: |
vulkaninfo --summary
GG_BUILD_VULKAN=1 GGML_VK_DISABLE_COOPMAT2=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
ggml-ci-nvidia-vulkan-cm2:
needs: determine-tag
runs-on: [self-hosted, Linux, NVIDIA, COOPMAT2]
steps:
@@ -116,14 +93,11 @@ jobs:
- name: Test
id: ggml-ci
env:
HF_UI_VERSION: ${{ needs.determine-tag.outputs.tag_name }}
run: |
vulkaninfo --summary
GG_BUILD_VULKAN=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
ggml-ci-nvidia-webgpu:
needs: determine-tag
runs-on: [self-hosted, Linux, NVIDIA, X64]
steps:
@@ -154,7 +128,6 @@ jobs:
# TODO: provision AMX-compatible machine
#ggml-ci-cpu-amx:
# needs: determine-tag
# runs-on: [self-hosted, Linux, CPU, AMX]
# steps:
@@ -169,7 +142,6 @@ jobs:
# TODO: provision AMD GPU machine
# ggml-ci-amd-vulkan:
# needs: determine-tag
# runs-on: [self-hosted, Linux, AMD]
# steps:
@@ -185,7 +157,6 @@ jobs:
# TODO: provision AMD GPU machine
# ggml-ci-amd-rocm:
# needs: determine-tag
# runs-on: [self-hosted, Linux, AMD]
# steps:
@@ -200,7 +171,6 @@ jobs:
# GG_BUILD_ROCM=1 GG_BUILD_AMDGPU_TARGETS="gfx1101" bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
ggml-ci-mac-metal:
needs: determine-tag
runs-on: [self-hosted, macOS, ARM64]
steps:
@@ -210,13 +180,10 @@ jobs:
- name: Test
id: ggml-ci
env:
HF_UI_VERSION: ${{ needs.determine-tag.outputs.tag_name }}
run: |
GG_BUILD_METAL=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
ggml-ci-mac-webgpu:
needs: determine-tag
runs-on: [self-hosted, macOS, ARM64]
steps:
@@ -239,14 +206,11 @@ jobs:
- name: Test
id: ggml-ci
env:
HF_UI_VERSION: ${{ needs.determine-tag.outputs.tag_name }}
run: |
GG_BUILD_WEBGPU=1 GG_BUILD_WEBGPU_DAWN_PREFIX="$GITHUB_WORKSPACE/dawn" \
bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
ggml-ci-mac-vulkan:
needs: determine-tag
runs-on: [self-hosted, macOS, ARM64]
steps:
@@ -256,14 +220,11 @@ jobs:
- name: Test
id: ggml-ci
env:
HF_UI_VERSION: ${{ needs.determine-tag.outputs.tag_name }}
run: |
vulkaninfo --summary
GG_BUILD_VULKAN=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
ggml-ci-linux-intel-vulkan:
needs: determine-tag
runs-on: [self-hosted, Linux, Intel]
steps:
@@ -275,14 +236,11 @@ jobs:
- name: Test
id: ggml-ci
env:
HF_UI_VERSION: ${{ needs.determine-tag.outputs.tag_name }}
run: |
vulkaninfo --summary
GG_BUILD_VULKAN=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
ggml-ci-win-intel-vulkan:
needs: determine-tag
runs-on: [self-hosted, Windows, X64, Intel]
steps:
@@ -297,7 +255,6 @@ jobs:
MSYSTEM: UCRT64
CHERE_INVOKING: 1
PATH: C:\msys64\ucrt64\bin;C:\msys64\usr\bin;C:\Windows\System32;${{ env.PATH }}
HF_UI_VERSION: ${{ needs.determine-tag.outputs.tag_name }}
run: |
vulkaninfo --summary
# Skip python related tests with GG_BUILD_LOW_PERF=1 since Windows MSYS2 UCRT64 currently fails to create
@@ -305,7 +262,6 @@ jobs:
LLAMA_FATAL_WARNINGS=OFF GG_BUILD_NINJA=1 GG_BUILD_VULKAN=1 GG_BUILD_LOW_PERF=1 ./ci/run.sh ./results/llama.cpp ./mnt/llama.cpp
ggml-ci-intel-openvino-gpu-low-perf:
needs: determine-tag
runs-on: [self-hosted, Linux, Intel, OpenVINO]
concurrency:
@@ -337,15 +293,12 @@ jobs:
- name: Test
id: ggml-ci
env:
HF_UI_VERSION: ${{ needs.determine-tag.outputs.tag_name }}
run: |
source ./openvino_toolkit/setupvars.sh
GG_BUILD_OPENVINO=1 GGML_OPENVINO_DEVICE=GPU GG_BUILD_LOW_PERF=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
ggml-ci-arm64-cpu-low-perf:
needs: determine-tag
runs-on: [self-hosted, Linux, ARM64]
runs-on: [self-hosted, Linux, ARM64, CPU]
steps:
- name: Clone
@@ -358,8 +311,7 @@ jobs:
LLAMA_ARG_THREADS=$(nproc) GG_BUILD_LOW_PERF=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
ggml-ci-arm64-cpu-high-perf:
needs: determine-tag
runs-on: [self-hosted, Linux, ARM64]
runs-on: [self-hosted, Linux, ARM64, CPU]
steps:
- name: Clone
@@ -378,8 +330,7 @@ jobs:
# if we resolve this, we should be able to offload these jobs to the self-hosted runners
#
# ggml-ci-arm64-cpu-high-perf-sve:
# needs: determine-tag
# runs-on: [self-hosted, Linux, NVIDIA, ARM64]
# runs-on: [self-hosted, Linux, ARM64, CPU]
#
# steps:
# - name: Clone
@@ -392,8 +343,7 @@ jobs:
# LLAMA_ARG_THREADS=$(nproc) GG_BUILD_NO_BF16=1 GG_BUILD_EXTRA_TESTS_0=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
#
# ggml-ci-arm64-cpu-kleidiai:
# needs: determine-tag
# runs-on: [self-hosted, Linux, NVIDIA, ARM64]
# runs-on: [self-hosted, Linux, ARM64, CPU]
#
# steps:
# - name: Clone