commit 082630ba4e2de697b6d372dd35ffc1be6a42c346
author Anitha Raj <anitha.raj@arm.com> Tue Aug 22 15:46:27 2023 +0100
committer Anitha Raj <Anitha.Raj@arm.com> Wed Aug 23 08:44:11 2023 +0000
tree 8d08824bf8d30d8bc3ade4493ed43bfbcae97bae
parent 1b2ee3eb6991efcbd09cde84ab0c383f15648738

Update CpuGemmConv2d and CpuFlatten to use CpuReshape operator

- Following CpuReshapeKernel Optimizations, update the CpuGemmConv2D and CpuFlatten
  to use CpuReshape operator instead of CpuReshapeKernel
- Minor changes to comment in NEReorgLayerKernel.h

Resolves COMPMID-6504

Signed-off-by: Anitha Raj <anitha.raj@arm.com>
Change-Id: Ib6ee1fdc313d91249f9fe41c81e73324031c1ff4
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/10186
Reviewed-by: Jakub Sujak <jakub.sujak@arm.com>
Reviewed-by: Gunes Bayir <gunes.bayir@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>

src/cpu/operators/CpuGemmConv2d.cpp

diff --git a/src/cpu/operators/CpuGemmConv2d.cpp b/src/cpu/operators/CpuGemmConv2d.cpp
index 7c0e58b..d11e4f0 100644
--- a/src/cpu/operators/CpuGemmConv2d.cpp
+++ b/src/cpu/operators/CpuGemmConv2d.cpp
@@ -35,11 +35,11 @@
 #include "src/core/helpers/MemoryHelpers.h"
 #include "src/cpu/kernels/CpuCol2ImKernel.h"
 #include "src/cpu/kernels/CpuIm2ColKernel.h"
-#include "src/cpu/kernels/CpuReshapeKernel.h"
 #include "src/cpu/kernels/CpuWeightsReshapeKernel.h"
 #include "src/cpu/operators/CpuGemm.h"
 #include "src/cpu/operators/CpuGemmLowpMatrixMultiplyCore.h"
 #include "src/cpu/operators/CpuGemmLowpOutputStage.h"
+#include "src/cpu/operators/CpuReshape.h"
 #include "src/cpu/utils/CpuAuxTensorHandler.h"
 
 #include <set>
@@ -92,7 +92,7 @@
 }
 
 CpuGemmConv2d::CpuGemmConv2d()
-    : _weights_reshape_kernel(nullptr), _im2col_kernel(), _mm_gemm(), _mm_gemmlowp(), _col2im_kernel(), _reshape_kernel(), _im2col_output(), _weights_reshaped(), _gemm_output(), _gemm_output_3d(),
+    : _weights_reshape_kernel(nullptr), _im2col_kernel(), _mm_gemm(), _mm_gemmlowp(), _col2im_kernel(), _reshape(), _im2col_output(), _weights_reshaped(), _gemm_output(), _gemm_output_3d(),
       _data_layout(DataLayout::NCHW), _skip_im2col(false), _skip_col2im(false), _is_quantized(false), _is_prepared(false), _aux_mem(AuxTensorIdx::Count)
 {
 }
@@ -379,8 +379,8 @@
     else
     {
         // Configure reshape layer
-        _reshape_kernel = std::make_unique<kernels::CpuReshapeKernel>();
-        _reshape_kernel->configure(gemm_output_to_use, dst);
+        _reshape = std::make_unique<CpuReshape>();
+        _reshape->configure(gemm_output_to_use, dst);
     }
 
     // Check if GEMM transforms weights
@@ -642,7 +642,7 @@
                 { TensorType::ACL_SRC, gemm_output_to_use },
                 { TensorType::ACL_DST, dst }
             };
-            NEScheduler::get().schedule_op(_reshape_kernel.get(), Window::DimY, _reshape_kernel->window(), pack);
+            _reshape->run(pack);
         }
     }
     else if(out_has_padding)
@@ -652,7 +652,7 @@
             { TensorType::ACL_SRC, gemm_output_to_use },
             { TensorType::ACL_DST, dst }
         };
-        NEScheduler::get().schedule_op(_reshape_kernel.get(), Window::DimY, _reshape_kernel->window(), pack);
+        _reshape->run(pack);
     }
 }