From b6e13c4f8d23323c2288e50a0e37cbf7f96846b8 Mon Sep 17 00:00:00 2001
From: Devmate Bot <noreply+537391475639613@fb.com>
Date: Sun, 14 Dec 2025 11:05:09 -0800
Subject: [PATCH] Your diff, D88106671, broke one test

Reviewed By: hsharma35

Differential Revision: D89143173
---
 .../generic/operators/quantized_op_macros.h   | 30 +++++++++----------
 .../cadence/generic/operators/targets.bzl     |  1 +
 2 files changed, 16 insertions(+), 15 deletions(-)

diff --git a/backends/cadence/generic/operators/quantized_op_macros.h b/backends/cadence/generic/operators/quantized_op_macros.h
index eda6de2e8d7..a64aa3f2636 100644
--- a/backends/cadence/generic/operators/quantized_op_macros.h
+++ b/backends/cadence/generic/operators/quantized_op_macros.h
@@ -9,11 +9,9 @@
 #pragma once
 
 #include <executorch/backends/cadence/generic/operators/cadence_type_util.h>
+#include <executorch/kernels/portable/cpu/util/broadcast_util.h>
 #include <executorch/runtime/core/exec_aten/exec_aten.h>
 
-// Generate kernels that perform elementwise arithmetic on two quantized
-// tensors. The tensors are either the same size, or the second tensor is a
-// scalar.
 #define DECLARE_POINTWISE_TENSOR_QUANTIZED_BINARY_OP(BINARY_FUNC_NAME, OP) \
   template <typename T>                                                    \
   void BINARY_FUNC_NAME(                                                   \
@@ -26,17 +24,19 @@
       float out_scale,                                                     \
       int32_t out_zero_point,                                              \
       ::executorch::aten::Tensor& out) {                                   \
-    const T* __restrict__ X_data = X.const_data_ptr<T>();                  \
-    const T* __restrict__ Y_data = Y.const_data_ptr<T>();                  \
-    T* __restrict__ out_data = out.mutable_data_ptr<T>();                  \
     float inv_out_scale = 1.0f / out_scale;                                \
-    for (size_t i = 0, e = X.numel(); i < e; ++i) {                        \
-      float x = ::impl::generic::kernels::dequantize<T>(                   \
-          X_data[i], X_scale, X_zero_point);                               \
-      float y = ::impl::generic::kernels::dequantize<T>(                   \
-          Y_data[i], Y_scale, Y_zero_point);                               \
-      float z = x OP y;                                                    \
-      out_data[i] = ::impl::generic::kernels::quantize<T>(                 \
-          z, inv_out_scale, out_zero_point);                               \
-    }                                                                      \
+    ::torch::executor::apply_binary_elementwise_fn<T, T, T>(              \
+        [X_scale, X_zero_point, Y_scale, Y_zero_point, inv_out_scale,     \
+         out_zero_point](const T x_val, const T y_val) {                   \
+          float x = ::impl::generic::kernels::dequantize<T>(               \
+              x_val, X_scale, X_zero_point);                               \
+          float y = ::impl::generic::kernels::dequantize<T>(               \
+              y_val, Y_scale, Y_zero_point);                               \
+          float z = x OP y;                                                \
+          return ::impl::generic::kernels::quantize<T>(                    \
+              z, inv_out_scale, out_zero_point);                           \
+        },                                                                 \
+        X,                                                                 \
+        Y,                                                                 \
+        out);                                                              \
   }
diff --git a/backends/cadence/generic/operators/targets.bzl b/backends/cadence/generic/operators/targets.bzl
index 6990a24db50..15e2f592f2e 100644
--- a/backends/cadence/generic/operators/targets.bzl
+++ b/backends/cadence/generic/operators/targets.bzl
@@ -15,6 +15,7 @@ def define_common_targets():
         exported_headers = ["quantized_op_macros.h"],
         exported_deps = [
             ":cadence_type_util",
+            "//executorch/kernels/portable/cpu/util:broadcast_util",
             "//executorch/runtime/kernel:kernel_includes",
         ]
     )