Update Neon™ depthwise kernel

- Reduce duplication and simplify overall structure.
- Improve multi-threaded performance by sharing more data
  in lower-level caches.

Partially Resolves: COMPMID-5054
Signed-off-by: Ramy Elgammal <ramy.elgammal@arm.com>
Change-Id: Iac747f39b21c540122fa75218762631c4d787911
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7449
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Andrew Mundy
Reviewed-by: Sheri Zhang <sheri.zhang@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

1 files changed, 10 insertions, 14 deletions

diff --git a/src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp32_nhwc_generic_output9_mla_depthfirst.hpp b/src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp32_nhwc_generic_output9_mla_depthfirst.hpp
index 0f6cecdc56..6fa02b781e 100644
--- a/src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp32_nhwc_generic_output9_mla_depthfirst.hpp
+++ b/src/core/NEON/kernels/arm_conv/depthwise/kernels/a64_fp32_nhwc_generic_output9_mla_depthfirst.hpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2021 Arm Limited.
+ * Copyright (c) 2021-2022 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -28,28 +28,24 @@
 
 #pragma once
 
+#if defined(__aarch64__)
+
 namespace arm_conv {
 namespace depthwise {
 
 void a64_fp32_nhwc_generic_output9_mla_depthfirst_impl(const float *const *const, float *const *const, const void *, const void *, const unsigned int, const unsigned int, const float, const float);
 
-struct a64_fp32_nhwc_generic_output9_mla_depthfirst
+class a64_fp32_nhwc_generic_output9_mla_depthfirst : public GenericDepthfirstKernelStrategy<float, float, float, float>
 {
-  typedef float bias_type;
-  typedef float input_type;
-  typedef float weight_type;
-  typedef float return_type;
-
-  typedef void (*kern_type)(const float *const *const, float *const *const, const void *, const void *, const unsigned int, const unsigned int, const float, const float);
-
-  constexpr static arm_gemm::VLType vl_type = arm_gemm::VLType::None;
+  KernelType kernel = a64_fp32_nhwc_generic_output9_mla_depthfirst_impl;
 
-  constexpr static unsigned int n_output_points = 9;
+  public:
+  a64_fp32_nhwc_generic_output9_mla_depthfirst(const CPUInfo *) : GenericDepthfirstKernelStrategy<float, float, float, float>(9, arm_gemm::VLType::None) {}
 
-  kern_type kernel = a64_fp32_nhwc_generic_output9_mla_depthfirst_impl;
-
-  a64_fp32_nhwc_generic_output9_mla_depthfirst(const CPUInfo *) {}
+  KernelType get_kernel() const override { return kernel; }
 };
 
 }  // namespace depthwise
 }  // namespace arm_conv
+
+#endif // defined(__aarch64__)