[cadence] add reference quantized_fully_connected_out

Differential Revision: D70723811

Pull Request resolved: #9020

Author: zonglinpeng

backends/cadence/aot/functions.yaml

@@ -238,3 +238,13 @@
   kernels:
     - arg_meta: null
       kernel_name: impl::reference::quantized_conv_per_tensor_out
+
+- func: cadence::quantized_fully_connected.out(Tensor src, Tensor weight, Tensor bias, int src_zero_point, Tensor weight_zero_point, Tensor out_multiplier, Tensor out_shift, int out_zero_point, Tensor? offset, *, Tensor(a!) out) -> Tensor(a!)
+  kernels:
+    - arg_meta: null
+      kernel_name: impl::reference::quantized_fully_connected_out
+
+- func: cadence::quantized_fully_connected.per_tensor_out(Tensor src, Tensor weight, Tensor bias, int src_zero_point, int weight_zero_point, int out_multiplier, int out_shift, int out_zero_point, Tensor? offset, *, Tensor(a!) out) -> Tensor(a!)
+  kernels:
+    - arg_meta: null
+      kernel_name: impl::reference::quantized_fully_connected_per_tensor_out

backends/cadence/reference/operators/CMakeLists.txt

@@ -87,6 +87,7 @@ add_library(
   "quantized_relu_out.cpp"
   "quantized_layer_norm.cpp"
   "quantize_per_tensor.cpp"
+  "quantized_fully_connected_out.cpp"
   "dequantize_per_tensor.cpp"
   "quantized_matmul_out.cpp"
   "im2row_out.cpp"

backends/cadence/reference/operators/quantized_fully_connected_out.cpp

@@ -0,0 +1,97 @@
+/*
+ * Copyright (c) Meta Platforms, Inc. and affiliates.
+ * All rights reserved.
+ *
+ * This source code is licensed under the BSD-style license found in the
+ * LICENSE file in the root directory of this source tree.
+ */
+#include <executorch/backends/cadence/reference/kernels/kernels.h>
+#include <executorch/backends/cadence/reference/operators/operators.h>
+#include <executorch/backends/cadence/reference/operators/quantized_ops.h>
+
+namespace impl {
+namespace reference {
+namespace native {
+
+using ::executorch::aten::optional;
+using ::executorch::aten::ScalarType;
+using ::executorch::aten::Tensor;
+using ::executorch::runtime::KernelRuntimeContext;
+
+void quantized_fully_connected_out(
+    __ET_UNUSED KernelRuntimeContext& ctx,
+    const Tensor& in,
+    const Tensor& weight,
+    const Tensor& bias,
+    int64_t in_zero_point,
+    const Tensor& weight_zero_point_t,
+    const Tensor& out_multiplier,
+    const Tensor& out_shift,
+    int64_t out_zero_point,
+    __ET_UNUSED const optional<Tensor>& offset,
+    Tensor& out) {
+#define typed_quantized_linear(ctype, dtype) \
+  case ScalarType::dtype: {                  \
+    quantized_linear_<ctype>(                \
+        in,                                  \
+        weight,                              \
+        bias,                                \
+        in_zero_point,                       \
+        weight_zero_point_t,                 \
+        out_multiplier,                      \
+        out_shift,                           \
+        out_zero_point,                      \
+        out);                                \
+    break;                                   \
+  }
+
+  ScalarType dtype = out.scalar_type();
+  switch (dtype) {
+    ET_FORALL_CADENCE_QUANTIZED_TYPES(typed_quantized_linear);
+    default:
+      ET_DCHECK_MSG(
+          false, "Unhandled dtype %s", torch::executor::toString(dtype));
+  }
+#undef typed_quantized_linear
+}
+
+void quantized_fully_connected_per_tensor_out(
+    __ET_UNUSED KernelRuntimeContext& ctx,
+    const Tensor& in,
+    const Tensor& weight,
+    const Tensor& bias,
+    int64_t in_zero_point,
+    int64_t weight_zero_point,
+    int64_t out_multiplier,
+    int64_t out_shift,
+    int64_t out_zero_point,
+    __ET_UNUSED const optional<Tensor>& offset,
+    Tensor& out) {
+#define typed_quantized_linear(ctype, dtype) \
+  case ScalarType::dtype: {                  \
+    quantized_linear_per_tensor_<ctype>(     \
+        in,                                  \
+        weight,                              \
+        bias,                                \
+        in_zero_point,                       \
+        weight_zero_point,                   \
+        out_multiplier,                      \
+        out_shift,                           \
+        out_zero_point,                      \
+        out);                                \
+    break;                                   \
+  }
+
+  ScalarType dtype = out.scalar_type();
+  switch (dtype) {
+    ET_FORALL_CADENCE_QUANTIZED_TYPES(typed_quantized_linear);
+    default:
+      ET_DCHECK_MSG(
+          false, "Unhandled dtype %s", torch::executor::toString(dtype));
+  }
+#undef typed_quantized_linear
+}
+
+}; // namespace native
+}; // namespace reference
+}; // namespace impl