COMPMID-903: Implements NEPermute for NHWC conversions

Change-Id: I4083e8d16bb23933634f229a1408dfd0e8f2922a
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/120069
Tested-by: Jenkins <bsgcomp@arm.com>
Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>

4 files changed, 75 insertions, 5 deletions

diff --git a/arm_compute/runtime/NEON/NEFunctions.h b/arm_compute/runtime/NEON/NEFunctions.h
index d09fcb280c..077cf577e7 100644
--- a/arm_compute/runtime/NEON/NEFunctions.h
+++ b/arm_compute/runtime/NEON/NEFunctions.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2016, 2018 ARM Limited.
+ * Copyright (c) 2016-2018 ARM Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -88,6 +88,7 @@
 #include "arm_compute/runtime/NEON/functions/NENonMaximaSuppression3x3.h"
 #include "arm_compute/runtime/NEON/functions/NENormalizationLayer.h"
 #include "arm_compute/runtime/NEON/functions/NEOpticalFlow.h"
+#include "arm_compute/runtime/NEON/functions/NEPermute.h"
 #include "arm_compute/runtime/NEON/functions/NEPhase.h"
 #include "arm_compute/runtime/NEON/functions/NEPixelWiseMultiplication.h"
 #include "arm_compute/runtime/NEON/functions/NEPoolingLayer.h"
diff --git a/arm_compute/runtime/NEON/functions/NEConvolutionLayer.h b/arm_compute/runtime/NEON/functions/NEConvolutionLayer.h
index f65e7ef36b..f80f67d944 100644
--- a/arm_compute/runtime/NEON/functions/NEConvolutionLayer.h
+++ b/arm_compute/runtime/NEON/functions/NEConvolutionLayer.h
@@ -146,6 +146,14 @@ private:
      *                     except for input of QASYMM8 type where output should be of S32 type.
      */
     void configure_mm(const ITensor *input, const ITensor *weights, ITensor *output);
+    /** Prepare the appropriate assembly optimized kernel
+     *
+     * @param[in] ci CPU information
+     * @param[in] M  M parameter of matrix multiplication
+     * @param[in] N  N parameter of matrix multiplication
+     * @param[in] K  K parameter of matrix multiplication
+     */
+    void configure_asm_mm(const struct CPUInfo &ci, int M, int N, int K);
 
 private:
     MemoryGroup                                         _memory_group;
diff --git a/arm_compute/runtime/NEON/functions/NEDepthwiseConvolutionLayer.h b/arm_compute/runtime/NEON/functions/NEDepthwiseConvolutionLayer.h
index 682effe84b..1c65b3ccf5 100644
--- a/arm_compute/runtime/NEON/functions/NEDepthwiseConvolutionLayer.h
+++ b/arm_compute/runtime/NEON/functions/NEDepthwiseConvolutionLayer.h
@@ -32,10 +32,10 @@
 #include "arm_compute/core/NEON/kernels/NEFillBorderKernel.h"
 #include "arm_compute/core/NEON/kernels/NEGEMMMatrixVectorMultiplyKernel.h"
 #include "arm_compute/core/Types.h"
-#include "arm_compute/runtime/CPP/functions/CPPPermute.h"
 #include "arm_compute/runtime/IFunction.h"
 #include "arm_compute/runtime/IMemoryManager.h"
 #include "arm_compute/runtime/MemoryGroup.h"
+#include "arm_compute/runtime/NEON/functions/NEPermute.h"
 #include "arm_compute/runtime/Tensor.h"
 
 namespace arm_compute
@@ -71,9 +71,9 @@ private:
     NEDepthwiseConvolutionLayer3x3Kernel      _dwc_kernel;
     NEDirectConvolutionLayerOutputStageKernel _output_stage_kernel;
     NEFillBorderKernel                        _border_handler;
-    CPPPermute                                _permute_input;
-    CPPPermute                                _permute_weights;
-    CPPPermute                                _permute_output;
+    NEPermute                                 _permute_input;
+    NEPermute                                 _permute_weights;
+    NEPermute                                 _permute_output;
     Tensor                                    _accumulator;
     Tensor                                    _input_nhwc;
     Tensor                                    _weights_hwio;
diff --git a/arm_compute/runtime/NEON/functions/NEPermute.h b/arm_compute/runtime/NEON/functions/NEPermute.h
new file mode 100644
index 0000000000..58626cd2f2
--- /dev/null
+++ b/arm_compute/runtime/NEON/functions/NEPermute.h
@@ -0,0 +1,61 @@
+/*
+ * Copyright (c) 2018 ARM Limited.
+ *
+ * SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+ * deal in the Software without restriction, including without limitation the
+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+ * sell copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#ifndef __ARM_COMPUTE_NEPERMUTE_H__
+#define __ARM_COMPUTE_NEPERMUTE_H__
+
+#include "arm_compute/runtime/NEON/INESimpleFunction.h"
+
+#include "arm_compute/core/Types.h"
+
+namespace arm_compute
+{
+class ITensor;
+
+/** Basic function to run @ref NEPermuteKernel */
+class NEPermute : public INESimpleFunction
+{
+public:
+    /** Configure the permute NEON kernel
+     *
+     * @note Supported permutation vectors : [2, 0, 1], [1, 2, 0]
+     *
+     * @param[in]  input  The input tensor to permute. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/QS16/F16/U32/S32/F32
+     * @param[out] output The output tensor. Data types supported: Same as @p input
+     * @param[in]  perm   Permutation vector
+     */
+    void configure(const ITensor *input, ITensor *output, const PermutationVector &perm);
+    /** Static function to check if given info will lead to a valid configuration of @ref NEPermute
+     *
+     * @note Supported permutation vectors : [2, 0, 1], [1, 2, 0]
+     *
+     * @param[in] input  The input tensor to permute. Data types supported: U8/S8/QS8/QASYMM8/U16/S16/QS16/F16/U32/S32/F32
+     * @param[in] output The output tensor. Data types supported: Same as @p input
+     * @param[in] perm   Permutation vector
+     *
+     * @return a status
+     */
+    static Status validate(const ITensorInfo *input, const ITensorInfo *output, const PermutationVector &perm);
+};
+}
+#endif /* __ARM_COMPUTE_NEPERMUTE_H__ */