COMPMID-2847: Fuse output stage in GEMMLowpMatrixMultiplyReshapedOnlyRHS
Change-Id: Icd60eb368a34295434e8c141885b4666973a92a1
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/2732
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
diff --git a/arm_compute/core/CL/kernels/CLGEMMLowpMatrixMultiplyReshapedOnlyRHSKernel.h b/arm_compute/core/CL/kernels/CLGEMMLowpMatrixMultiplyReshapedOnlyRHSKernel.h
index 9dd5496..7845d24 100644
--- a/arm_compute/core/CL/kernels/CLGEMMLowpMatrixMultiplyReshapedOnlyRHSKernel.h
+++ b/arm_compute/core/CL/kernels/CLGEMMLowpMatrixMultiplyReshapedOnlyRHSKernel.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2019 ARM Limited.
+ * Copyright (c) 2019-2020 ARM Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -25,6 +25,7 @@
#define ARM_COMPUTE_CLGEMMLOWPMATRIXMULTIPLYRESHAPEDONLYRHSKERNEL_H
#include "arm_compute/core/CL/ICLKernel.h"
+#include "arm_compute/core/KernelDescriptors.h"
namespace arm_compute
{
@@ -33,6 +34,7 @@
/** OpenCL kernel to multiply matrices with QASYMM8 data type when only the input matrix RHS (input1) has been reshaped
*
* @note The input matrix input1 must be reshaped through @ref CLGEMMReshapeRHSMatrixKernel
+ * @note For fused output stage, only GEMMLowpOutputStageType::QUANTIZE_DOWN_FIXEDPOINT type is supported
*/
class CLGEMMLowpMatrixMultiplyReshapedOnlyRHSKernel : public ICLKernel
{
@@ -49,37 +51,59 @@
CLGEMMLowpMatrixMultiplyReshapedOnlyRHSKernel &operator=(CLGEMMLowpMatrixMultiplyReshapedOnlyRHSKernel &&) = default;
/** Initialise the kernel's input and output.
*
- * @param[in] input0 Input tensor containing the LHS matrix. Data type supported: QASYMM8/QASYMM8_SIGNED
- * @param[in] input1 Input tensor containing the RHS reshaped matrix. Data type supported: same as @p input0
- * @param[out] output Output tensor to store the result of matrix multiplication. Data type supported: S32
- * @param[in] lhs_info LHS matrix information used to retrieve the number of rows to be processed by each thread
- * lhs_info.m0: 2,3,4,5,6,7,8
- * lhs_info.k0: 2,3,4,8,16
- * @param[in] rhs_info RHS matrix information used for reshaping the input1 tensor. Only the following values are supported:
- * rhs_info.n0: 2,3,4,8,16
- * rhs_info.k0: 2,3,4,8,16
- * rhs_info.transpose: true
- * @param[in] gemm_info GEMM information used to retrieve the original dimensions of the input matrices
+ * @param[in] input0 Input tensor containing the LHS matrix. Data type supported: QASYMM8/QASYMM8_SIGNED
+ * @param[in] input1 Input tensor containing the RHS reshaped matrix. Data type supported: same as @p input0
+ * @param[out] output Output tensor. Data type supported: QASYMM8/QASYMM8_SIGNED/S32.
+ * @param[in] gemm_info GEMM information used to retrieve the original dimensions of the input matrices, output stage information and RHS/LHS info.
+ * Only the following values are supported for LHS info:
+ * lhs_info.m0: 2,3,4,5,6,7,8
+ * lhs_info.k0: 2,3,4,8,16
+ * Only the following values are supported for RHS info:
+ * rhs_info.n0: 2,3,4,8,16
+ * rhs_info.k0: same as lhs_info.k0
+ * rhs_info.transpose: true
+ * @param[in] vector_sum_col (Optional) Input row-vector of sums of all the entries in each column of matrix B.
+ * Note: vector_sum_col can be a nullptr in case a_offset = 0. Data type supported: S32
+ * @param[in] vector_sum_row (Optional) Input row-vector of sums of all the entries in each row of matrix A.
+ * Note: vector_sum_row can be a nullptr in case b_offset = 0. Data type supported: S32
+ * @param[in] bias (Optional) Biases tensor. Only shared biases supported and it can be a nullptr if the addition of biases is not required.
+ * Biases are 1D tensor with dimensions [OFM]. Data type supported: S32.
+ * @param[in] output_multipliers (Optional) Output multipliers tensor. In case of per-channel quantization, the number of multipliers must be equal to the number of filters (OFM).
+ * Supported data types: S32.
+ * @param[in] output_shifts (Optional) Output shifts tensor. In case of per-channel quantization, the number of multipliers must be equal to the number of filters (OFM).
+ * Supported data types: S32.
*/
- void configure(const ICLTensor *input0, const ICLTensor *input1, ICLTensor *output, const GEMMLHSMatrixInfo &lhs_info, const GEMMRHSMatrixInfo &rhs_info, const GEMMReshapeInfo &gemm_info);
+ void configure(const ICLTensor *input0, const ICLTensor *input1, ICLTensor *output, const GEMMKernelInfo &gemm_info, const ICLTensor *vector_sum_col = nullptr,
+ const ICLTensor *vector_sum_row = nullptr, const ICLTensor *bias = nullptr, const ICLTensor *output_multipliers = nullptr, const ICLTensor *output_shifts = nullptr);
/** Static function to check if given info will lead to a valid configuration of @ref CLGEMMLowpMatrixMultiplyReshapedOnlyRHSKernel
*
- * @param[in] input0 Input tensor info for the LHS matrix. Data type supported: QASYMM8/QASYMM8_SIGNED
- * @param[in] input1 Input tensor info for the RHS reshaped matrix. Data type supported: same as @p input0
- * @param[in] output Output tensor info. Data type supported: S32
- * @param[in] lhs_info LHS matrix information used to retrieve the number of rows to be processed by each thread
- * lhs_info.m0: 2,3,4,5,6,7,8
- * lhs_info.k0: 2,3,4,8,16
- * @param[in] rhs_info RHS matrix information used for reshaping the input1 tensor. Only the following values are supported:
- * rhs_info.n0: 2,3,4,8,16
- * rhs_info.k0: same as lhs_info.k0
- * rhs_info.transpose: true
- * @param[in] gemm_info GEMM information used to retrieve the original dimensions of the input matrices
+ * @param[in] input0 Input tensor info for the LHS matrix. Data type supported: QASYMM8/QASYMM8_SIGNED
+ * @param[in] input1 Input tensor info for the RHS reshaped matrix. Data type supported: same as @p input0
+ * @param[in] output Output tensor info. Data type supported: QASYMM8/QASYMM8_SIGNED/S32.
+ * @param[in] gemm_info GEMM information used to retrieve the original dimensions of the input matrices, output stage information and RHS/LHS info.
+ * Only the following values are supported for LHS info:
+ * lhs_info.m0: 2,3,4,5,6,7,8
+ * lhs_info.k0: 2,3,4,8,16
+ * Only the following values are supported for RHS info:
+ * rhs_info.n0: 2,3,4,8,16
+ * rhs_info.k0: same as lhs_info.k0
+ * rhs_info.transpose: true
+ * @param[in] vector_sum_col (Optional) Input row-vector info of sums of all the entries in each column of matrix B.
+ * Note: vector_sum_col can be a nullptr in case a_offset = 0. Data type supported: S32
+ * @param[in] vector_sum_row (Optional) Input row-vector info of sums of all the entries in each row of matrix A.
+ * Note: vector_sum_row can be a nullptr in case b_offset = 0. Data type supported: S32
+ * @param[in] bias (Optional) Biases tensor info. Only shared biases supported and it can be a nullptr if the addition of biases is not required.
+ * Biases are 1D tensor with dimensions [OFM]. Data type supported: S32.
+ * @param[in] output_multipliers (Optional) Output multipliers tensor info. In case of per-channel quantization, the number of multipliers must be equal to the number of filters (OFM).
+ * Supported data types: S32.
+ * @param[in] output_shifts (Optional) Output shifts tensor info. In case of per-channel quantization, the number of multipliers must be equal to the number of filters (OFM).
+ * Supported data types: S32.
*
* @return a status
*/
- static Status validate(const ITensorInfo *input0, const ITensorInfo *input1, const ITensorInfo *output, const GEMMLHSMatrixInfo &lhs_info, const GEMMRHSMatrixInfo &rhs_info,
- const GEMMReshapeInfo &gemm_info);
+ static Status validate(const ITensorInfo *input0, const ITensorInfo *input1, const ITensorInfo *output, const GEMMKernelInfo &gemm_info, const ITensorInfo *vector_sum_col = nullptr,
+ const ITensorInfo *vector_sum_row = nullptr, const ITensorInfo *bias = nullptr, const ITensorInfo *output_multipliers = nullptr,
+ const ITensorInfo *output_shifts = nullptr);
// Inherited methods overridden:
void run(const Window &window, cl::CommandQueue &queue) override;
@@ -88,10 +112,17 @@
const ICLTensor *_input0;
const ICLTensor *_input1;
ICLTensor *_output;
+ const ICLTensor *_vector_sum_col;
+ const ICLTensor *_vector_sum_row;
+ const ICLTensor *_bias;
+ const ICLTensor *_output_multipliers;
+ const ICLTensor *_output_shifts;
bool _slide_matrix_b;
bool _reinterpret_input_as_3d;
bool _reinterpret_output_as_3d;
bool _use_dummy_work_items;
+ bool _is_quantized_per_channel;
+ bool _fuse_output_stage;
};
} // namespace arm_compute
#endif /* ARM_COMPUTE_CLGEMMLOWPMATRIXMULTIPLYRESHAPEDONLYRHSKERNEL_H */
\ No newline at end of file
diff --git a/arm_compute/core/KernelDescriptors.h b/arm_compute/core/KernelDescriptors.h
index 4b04beb..58400b1 100644
--- a/arm_compute/core/KernelDescriptors.h
+++ b/arm_compute/core/KernelDescriptors.h
@@ -54,14 +54,21 @@
/** Descriptor used by the GEMM kernels */
struct GEMMKernelInfo
{
- unsigned int m{ 0 }; /**< Number of LHS rows*/
- unsigned int n{ 0 }; /**< Number of RHS columns*/
- unsigned int k{ 0 }; /**< Number of LHS columns or RHS rows */
- unsigned int depth_output_gemm3d{ 0 }; /**< Depth of the output tensor in case is reinterpreted as 3D */
- bool reinterpret_input_as_3d{ false }; /**< Flag used to reinterpret the input as 3D */
- bool broadcast_bias{ false }; /**< Flag used to broadcase the bias addition */
- bool fp_mixed_precision{ false }; /**< Flag used to indicate wider accumulators (32 bit instead of 16 for FP16). */
- ActivationLayerInfo activation_info{}; /**< Activation function to perform after the matrix multiplication */
+ unsigned int m{ 0 }; /**< Number of LHS rows*/
+ unsigned int n{ 0 }; /**< Number of RHS columns*/
+ unsigned int k{ 0 }; /**< Number of LHS columns or RHS rows */
+ unsigned int depth_output_gemm3d{ 0 }; /**< Depth of the output tensor in case is reinterpreted as 3D */
+ bool reinterpret_input_as_3d{ false }; /**< Flag used to reinterpret the input as 3D */
+ bool broadcast_bias{ false }; /**< Flag used to broadcast the bias addition */
+ bool fp_mixed_precision{ false }; /**< Flag used to indicate wider accumulators (32 bit instead of 16 for FP16). */
+ ActivationLayerInfo activation_info{}; /**< Activation function to perform after the matrix multiplication */
+ int mult_transpose1xW_width{ 1 }; /**< Multiplication factor for the width of the 1xW transposed block */
+ int mult_interleave4x4_height{ 1 }; /**< Multiplication factor for the height of the 4x4 interleaved block */
+ GEMMLHSMatrixInfo lhs_info{}; /**< LHS matrix information used to retrieve the number of rows processed by each thread */
+ GEMMRHSMatrixInfo rhs_info{}; /**< RHS matrix information used for reshaping the RHS matrix */
+ int32_t a_offset{ 0 }; /**< Offset to be added to each element of the matrix A */
+ int32_t b_offset{ 0 }; /**< Offset to be added to each element of the matrix B */
+ GEMMLowpOutputStageInfo output_stage{}; /**< GEMMLowp output stage information */
};
/** Descriptor used by the depthwise convolution kernels */