Merge branch 'pytorch:main' into integrating-vision-kernels

Author: cad-rlc

.Package.swift/kernels_custom/dummy.swift .Package.swift/kernels_llm/dummy.swift.Package.swift/kernels_custom/dummy.swift renamed to .Package.swift/kernels_llm/dummy.swift

@@ -1 +1 @@
-7cda4017ddda554752e89069ae205be5e8388f59
+90f1e7bed15ca5e48c61c5b6dc5ad4810524f82f

…e.swift/kernels_custom_debug/dummy.swift …kage.swift/kernels_llm_debug/dummy.swift.Package.swift/kernels_custom_debug/dummy.swift renamed to .Package.swift/kernels_llm_debug/dummy.swift .Package.swift/kernels_custom_debug/dummy.swift renamed to .Package.swift/kernels_llm_debug/dummy.swift

@@ -0,0 +1,13 @@
+
+set -ex
+
+install_emscripten() {
+    git clone https://github.com/emscripten-core/emsdk.git
+    pushd emsdk || return
+    ./emsdk install 4.0.10
+    ./emsdk activate 4.0.10
+    source ./emsdk_env.sh
+    popd || return
+}
+
+install_emscripten

.ci/docker/ci_commit_pins/pytorch.txt

@@ -35,15 +35,15 @@ cmake -DPYTHON_EXECUTABLE=python \
     -DEXECUTORCH_BUILD_XNNPACK=OFF \
     -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON \
     -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \
-    -DEXECUTORCH_BUILD_KERNELS_CUSTOM=ON \
+    -DEXECUTORCH_BUILD_KERNELS_LLM=ON \
     -Bcmake-out .
 cmake --build cmake-out -j16 --target install --config Release
 
 # Install llama runner with torchao
 cmake -DPYTHON_EXECUTABLE=python \
     -DBUILD_TESTING=OFF \
     -DCMAKE_BUILD_TYPE=Release \
-    -DEXECUTORCH_BUILD_KERNELS_CUSTOM=ON \
+    -DEXECUTORCH_BUILD_KERNELS_LLM=ON \
     -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \
     -DEXECUTORCH_BUILD_XNNPACK=OFF \
     -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON \

.ci/scripts/setup-emscripten.sh

@@ -39,7 +39,7 @@ EXECUTORCH_COMMON_CMAKE_ARGS="                      \
         -DEXECUTORCH_BUILD_EXTENSION_DATA_LOADER=ON \
         -DEXECUTORCH_BUILD_EXTENSION_FLAT_TENSOR=ON \
         -DEXECUTORCH_BUILD_EXTENSION_TENSOR=ON      \
-        -DEXECUTORCH_BUILD_KERNELS_CUSTOM=ON        \
+        -DEXECUTORCH_BUILD_KERNELS_LLM=ON        \
         -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON     \
         -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON     \
         -DEXECUTORCH_BUILD_XNNPACK=ON               \
@@ -69,7 +69,7 @@ LLAVA_COMMON_CMAKE_ARGS="                        \
         -DPYTHON_EXECUTABLE="$PYTHON_EXECUTABLE" \
         -DCMAKE_INSTALL_PREFIX=${BUILD_DIR}      \
         -DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE}   \
-        -DEXECUTORCH_BUILD_KERNELS_CUSTOM=ON     \
+        -DEXECUTORCH_BUILD_KERNELS_LLM=ON     \
         -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON  \
         -DEXECUTORCH_BUILD_XNNPACK=ON"
 

.ci/scripts/test_llama_torchao_lowbit.sh

@@ -22,31 +22,14 @@ NPROC=8
 if hash nproc &> /dev/null; then NPROC=$(nproc); fi
 
 cmake_install_executorch_libraries() {
-  cmake -DPYTHON_EXECUTABLE=python \
-      -DCMAKE_INSTALL_PREFIX=${BUILD_DIR} \
-      -DEXECUTORCH_ENABLE_LOGGING=1 \
-      -DCMAKE_BUILD_TYPE=${BUILD_TYPE} \
-      -DEXECUTORCH_BUILD_EXTENSION_DATA_LOADER=ON \
-      -DEXECUTORCH_BUILD_EXTENSION_FLAT_TENSOR=ON \
-      -DEXECUTORCH_BUILD_EXTENSION_MODULE=ON \
-      -DEXECUTORCH_BUILD_EXTENSION_TENSOR=ON \
-      -DEXECUTORCH_BUILD_XNNPACK=ON \
-      -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON \
-      -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \
-      -DEXECUTORCH_BUILD_KERNELS_CUSTOM=ON \
-      -B${BUILD_DIR} .
-
-  cmake --build ${BUILD_DIR} -j${NPROC} --target install --config ${BUILD_TYPE}
+  rm -rf cmake-out
+  cmake --preset llm -DCMAKE_INSTALL_PREFIX=cmake-out -DCMAKE_BUILD_TYPE=${BUILD_TYPE}
+  cmake --build cmake-out -j16 --target install --config ${BUILD_TYPE}
 }
 
 cmake_build_phi_3_mini() {
-  cmake -DPYTHON_EXECUTABLE=$PYTHON_EXECUTABLE \
-      -DCMAKE_INSTALL_PREFIX=${BUILD_DIR} \
+  cmake -DCMAKE_PREFIX_PATH=${BUILD_DIR} \
       -DCMAKE_BUILD_TYPE=${BUILD_TYPE} \
-      -DEXECUTORCH_BUILD_KERNELS_CUSTOM=ON \
-      -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON \
-      -DEXECUTORCH_BUILD_XNNPACK=ON \
-      -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON \
       -B${BUILD_DIR}/${MODEL_DIR} \
       ${MODEL_DIR}
 
@@ -81,7 +64,7 @@ run_and_verify() {
     ${BUILD_DIR}/${MODEL_DIR}/phi_3_mini_runner \
     --model_path=phi-3-mini.pte \
     --tokenizer_path=tokenizer.bin \
-    --seq_len=128 \
+    --seq_len=60 \
     --temperature=0 \
     --prompt="<|system|>
 You are a helpful assistant.<|end|>

.ci/scripts/test_llava.sh

@@ -10,7 +10,7 @@ set -eux
 # TODO: can't query cadence & vulkan backends
 # TODO: can't query //kernels/prim_ops because of non-buckified stuff in OSS.
 buck2 query "//backends/apple/... + //backends/example/... + \
-//backends/mediatek/... + //backends/test/... + //backends/transforms/... + \
+//backends/mediatek/... + //backends/transforms/... + \
 //backends/xnnpack/... + //configurations/... + //kernels/aten/... + \
 //kernels/optimized/... + //kernels/portable/... + //kernels/quantized/... + \
 //kernels/test/... + //runtime/... + //schema/... + //test/... + //util/..."

.ci/scripts/test_phi_3_mini.sh

@@ -47,15 +47,13 @@ jobs:
     name: build-aar
     needs: check-if-aar-exists
     if: ${{ !github.event.pull_request.head.repo.fork }}
-    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@release/2.7
+    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
     secrets: inherit
     permissions:
       id-token: write
       contents: read
     with:
       secrets-env: EXECUTORCH_MAVEN_SIGNING_KEYID EXECUTORCH_MAVEN_SIGNING_PASSWORD EXECUTORCH_MAVEN_CENTRAL_PASSWORD EXECUTORCH_MAVEN_CENTRAL_USERNAME EXECUTORCH_MAVEN_SIGNING_GPG_KEY_CONTENTS
-      # As this job has access to Maven credential, run this on a fresh ephemeral runner
-      runner: ephemeral.linux.2xlarge
       docker-image: executorch-ubuntu-22.04-clang12-android
       submodules: 'recursive'
       ref: ${{ github.sha }}

.ci/scripts/unittest-buck2.sh

@@ -152,7 +152,7 @@ jobs:
           "backend_coreml"
           "backend_mps"
           "backend_xnnpack"
-          "kernels_custom"
+          "kernels_llm"
           "kernels_optimized"
           "kernels_quantized"
           "threadpool"