first crack at optimized op_where

ghstack-source-id: ebe8bee4e3ca4184e91058bd7033e69e130644da
ghstack-comment-id: 2691805026
Pull Request resolved: #8861

Author: swolchok

backends/xnnpack/CMakeLists.txt

@@ -139,7 +139,7 @@ if(NOT CMAKE_TOOLCHAIN_FILE MATCHES ".*(iOS|ios\.toolchain)\.cmake$")
   endif()
 
   target_link_libraries(
-    xnn_executor_runner gflags portable_ops_lib ${xnn_executor_runner_libs}
+    xnn_executor_runner gflags optimized_native_cpu_ops_lib ${xnn_executor_runner_libs}
   )
   target_compile_options(xnn_executor_runner PUBLIC ${_common_compile_options})
 endif()

kernels/optimized/cpu/targets.bzl

@@ -95,6 +95,12 @@ _OPTIMIZED_ATEN_OPS = (
             "//executorch/kernels/portable/cpu/util:broadcast_util",
         ],
     ),
+    op_target(
+        name = "op_where",
+        deps = [
+            "//executorch/kernels/portable/cpu/util:elementwise_util",
+        ],
+    ),
 )
 
 

kernels/optimized/optimized.yaml

@@ -101,3 +101,8 @@
   kernels:
     - arg_meta: null
       kernel_name: torch::executor::opt_sub_scalar_out
+
+- op: where.self_out
+  kernels:
+    - arg_meta: null
+      kernel_name: torch::executor::opt_where_out

kernels/test/CMakeLists.txt

@@ -275,6 +275,7 @@ set(_optimized_kernels_test_sources
     "op_native_layer_norm_test.cpp"
     "op_neg_test.cpp"
     "op_sub_test.cpp"
+    "op_where_test.cpp"
     "UnaryUfuncRealHBBF16ToFloatHBF16Test.cpp"
     ${CMAKE_CURRENT_BINARY_DIR}/include/optimized/executorch/kernels/test/supported_features.cpp
 )