Document and refine MoE hot-cache PP optimization path

.github/workflows/ai-issues.yml

@@ -6,7 +6,7 @@ on:
 
 jobs:
   find-related:
-    if: github.event.action == 'opened'
+    if: ${{ github.event.action == 'opened' && (github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp') }}
     runs-on: [self-hosted, opencode]
 
     permissions:

.github/workflows/build-cmake-pkg.yml

@@ -5,6 +5,7 @@ on:
 
 jobs:
   linux:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, Linux, CPU]
     steps:
       - uses: actions/checkout@v6

.github/workflows/build-ibm.yml

@@ -36,6 +36,7 @@ env:
 jobs:
 
   ubuntu-24-s390x:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: ubuntu-24.04-s390x
 
     steps:
@@ -95,6 +96,7 @@ jobs:
           ./bin/llama-completion -m stories260K-be.gguf -p "One day, Lily met a Shoggoth" -n 500 -c 256
 
   ubuntu-24-ppc64le:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: ubuntu-24.04-ppc64le
 
     steps:

.github/workflows/build-openvino.yml

@@ -35,6 +35,7 @@ env:
 
 jobs:
   ubuntu-24-openvino:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     name: ubuntu-24-openvino-${{ matrix.openvino_device }}
 
     concurrency:

.github/workflows/build-riscv.yml

@@ -35,6 +35,7 @@ env:
 
 jobs:
   ubuntu-cpu-riscv64-native:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: ubuntu-24.04-riscv
 
     steps:
@@ -108,6 +109,7 @@ jobs:
           ./bin/llama-completion -m stories260K.gguf -p "One day, Lily met a Shoggoth" -n 500 -c 256
 
   ubuntu-riscv64-native-sanitizer:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: ubuntu-24.04-riscv
 
     continue-on-error: true

.github/workflows/build-sanitize.yml

@@ -28,6 +28,7 @@ env:
 
 jobs:
   ctest:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, X64, CPU, Linux]
 
     continue-on-error: true

.github/workflows/build-self-hosted.yml

@@ -56,6 +56,7 @@ env:
 
 jobs:
   gpu-cuda:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, Linux, NVIDIA]
 
     steps:
@@ -70,6 +71,7 @@ jobs:
           GG_BUILD_CUDA=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 
   gpu-vulkan-nvidia-cm:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, Linux, NVIDIA]
 
     steps:
@@ -84,6 +86,7 @@ jobs:
           GG_BUILD_VULKAN=1 GGML_VK_DISABLE_COOPMAT2=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 
   gpu-vulkan-nvidia-cm2:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, Linux, NVIDIA, COOPMAT2]
 
     steps:
@@ -98,6 +101,7 @@ jobs:
           GG_BUILD_VULKAN=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 
   gpu-webgpu-nvidia:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, Linux, NVIDIA, X64]
 
     steps:
@@ -171,6 +175,7 @@ jobs:
   #         GG_BUILD_ROCM=1 GG_BUILD_AMDGPU_TARGETS="gfx1101" bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 
   gpu-metal:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, macOS, ARM64]
 
     steps:
@@ -184,6 +189,7 @@ jobs:
           GG_BUILD_METAL=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 
   gpu-webgpu-apple:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, macOS, ARM64]
 
     steps:
@@ -211,6 +217,7 @@ jobs:
             bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 
   gpu-vulkan:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, macOS, ARM64]
 
     steps:
@@ -225,6 +232,7 @@ jobs:
           GG_BUILD_VULKAN=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 
   gpu-vulkan-intel-linux:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, Linux, Intel]
 
     steps:
@@ -241,6 +249,7 @@ jobs:
           GG_BUILD_VULKAN=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 
   gpu-vulkan-intel-windows:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, Windows, X64, Intel]
 
     steps:
@@ -262,6 +271,7 @@ 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
 
   cpu-openvino-low-perf:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, Linux, Intel, OpenVINO]
 
     concurrency:
@@ -298,6 +308,7 @@ jobs:
           GG_BUILD_OPENVINO=1 GGML_OPENVINO_DEVICE=GPU GG_BUILD_LOW_PERF=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 
   cpu-any-low-perf:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, CPU]
 
     steps:
@@ -311,6 +322,7 @@ jobs:
           LLAMA_ARG_THREADS=$(nproc) GG_BUILD_LOW_PERF=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 
   cpu-any-high-perf:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, CPU]
 
     steps:
@@ -324,6 +336,7 @@ jobs:
           LLAMA_ARG_THREADS=$(nproc) GG_BUILD_HIGH_PERF=1 GG_BUILD_NO_SVE=1 GG_BUILD_NO_BF16=1 GG_BUILD_EXTRA_TESTS_0=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 
   cpu-arm64-graviton4:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: ah-ubuntu_22_04-c8g_8x
 
     steps:
@@ -363,6 +376,7 @@ 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
 
   cpu-arm64-graviton4-kleidiai:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: ah-ubuntu_22_04-c8g_8x
 
     steps:

.github/workflows/build.yml

@@ -58,6 +58,7 @@ env:
 
 jobs:
   build-cmake-pkg:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     uses: ./.github/workflows/build-cmake-pkg.yml
 
   ubuntu-cpu:

.github/workflows/check-vendor.yml

@@ -19,6 +19,7 @@ on:
 
 jobs:
   check-vendor:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, fast]
 
     steps:

.github/workflows/code-style.yml

@@ -15,6 +15,7 @@ concurrency:
 
 jobs:
   model-naming:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, fast]
     steps:
       - uses: actions/checkout@v6

.github/workflows/editorconfig.yml

@@ -15,6 +15,7 @@ concurrency:
 
 jobs:
   editorconfig:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, fast]
     steps:
       - uses: actions/checkout@v6

.github/workflows/labeler.yml

@@ -4,6 +4,7 @@ on:
 
 jobs:
   labeler:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     permissions:
       contents: read
       pull-requests: write

.github/workflows/pre-tokenizer-hashes.yml

@@ -12,6 +12,7 @@ on:
 
 jobs:
     pre-tokenizer-hashes:
+        if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
         runs-on: [self-hosted, fast]
 
         steps:

.github/workflows/python-check-requirements.yml

@@ -20,6 +20,7 @@ concurrency:
 
 jobs:
   python-check-requirements:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, CPU, fast]
     name: check-requirements
     steps:

.github/workflows/python-lint.yml

@@ -21,6 +21,7 @@ concurrency:
 
 jobs:
   flake8-lint:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, fast]
     name: Lint
     steps:

.github/workflows/python-type-check.yml

@@ -22,6 +22,7 @@ concurrency:
 
 jobs:
   python-type-check:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, fast]
     name: python type-check
     steps:

.github/workflows/release.yml

@@ -38,6 +38,7 @@ env:
 jobs:
 
   check_release:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, fast]
 
     outputs:

.github/workflows/server-sanitize.yml

@@ -37,6 +37,7 @@ concurrency:
 
 jobs:
   server:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, CPU, Linux, llama-server]
 
     strategy:

.github/workflows/server-self-hosted.yml

@@ -40,6 +40,7 @@ concurrency:
 
 jobs:
   server-metal:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, llama-server, macOS, ARM64]
 
     name: server-metal (${{ matrix.wf_name }})
@@ -92,6 +93,7 @@ jobs:
           pytest -v -x -m "not slow"
 
   server-cuda:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, llama-server, Linux, NVIDIA]
 
     name: server-cuda (${{ matrix.wf_name }})
@@ -131,6 +133,7 @@ jobs:
           pytest -v -x -m "not slow"
 
   server-kleidiai:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: ah-ubuntu_22_04-c8g_8x
 
     name: server-kleidiai (${{ matrix.wf_name }})

.github/workflows/ui-build.yml

@@ -5,6 +5,7 @@ on:
 
 jobs:
   build:
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     runs-on: [self-hosted, fast]
     env:
       BRANCH_NAME: ${{ github.head_ref || github.ref_name }}

.github/workflows/ui-self-hosted.yml

@@ -42,10 +42,12 @@ concurrency:
 jobs:
   ui-build:
     name: Build static output
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     uses: ./.github/workflows/ui-build.yml
 
   ui-checks:
     name: Checks
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     needs: ui-build
     runs-on: [self-hosted, PLAYWRIGHT]
     continue-on-error: true
@@ -83,6 +85,7 @@ jobs:
 
   e2e-tests:
     name: E2E Tests
+    if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
     needs: ui-build
     runs-on: [self-hosted, PLAYWRIGHT]
     steps:

.github/workflows/update-ops-docs.yml

@@ -16,6 +16,7 @@ on:
 
 jobs:
     update-ops-docs:
+        if: ${{ github.repository == 'ggml-org/llama.cpp' || github.repository == 'ggerganov/llama.cpp' }}
         runs-on: [self-hosted, fast, ARM64]
 
         steps: