From ab5b1a279284bed350d3bb75f3d9d3aec6edca0e Mon Sep 17 00:00:00 2001 From: Sang-Hoon Park Date: Tue, 15 Oct 2019 09:29:13 +0100 Subject: COMPMID-2744 [CL] add support for 9x9 quantized direct convolution Change-Id: I858ce5b9a530f8568e154f5d724d267e142ef9b2 Signed-off-by: Sang-Hoon Park Reviewed-on: https://review.mlplatform.org/c/2091 Comments-Addressed: Arm Jenkins Tested-by: Arm Jenkins Reviewed-by: Manuel Bottini Reviewed-by: Michele Di Giorgio Reviewed-by: Giuseppe Rossini --- .../CL/cl_kernels/direct_convolution_quantized.cl | 302 +++++++++++++++++++++ 1 file changed, 302 insertions(+) create mode 100644 src/core/CL/cl_kernels/direct_convolution_quantized.cl (limited to 'src/core/CL/cl_kernels/direct_convolution_quantized.cl') diff --git a/src/core/CL/cl_kernels/direct_convolution_quantized.cl b/src/core/CL/cl_kernels/direct_convolution_quantized.cl new file mode 100644 index 0000000000..1182428cd5 --- /dev/null +++ b/src/core/CL/cl_kernels/direct_convolution_quantized.cl @@ -0,0 +1,302 @@ +/* + * Copyright (c) 2017-2019 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. + */ +#include "helpers_asymm.h" + +#undef CONVERT_SAT + +#if defined(DATA_TYPE) && defined(STRIDE_X) && defined(WEIGHTS_DEPTH) + +#if KERNEL_SIZE == 9 + +#if STRIDE_X == 1 +#define CONVOLUTION1x9(acc, src_row_ptr, weights_row_ptr) CONVOLUTION1x9_STRIDE1(acc, src_row_ptr, weights_row_ptr) +#elif STRIDE_X == 2 +#define CONVOLUTION1x9(acc, src_row_ptr, weights_row_ptr) CONVOLUTION1x9_STRIDE2(acc, src_row_ptr, weights_row_ptr) +#else /* STRIDE_X not equals 1 or 2 */ +#error "STRIDE_X larger than 2 is not supported" +#endif /* STRIDE_X */ + +#define CONVOLUTION1x9_STRIDE1(acc, src_row_ptr, weights_row_ptr) \ + ({ \ + int8 weights_values0 = convert_int8(vload8(0, weights_row_ptr)); \ + int weights_value1 = convert_int(*(weights_row_ptr + 8)); \ + int16 src0 = convert_int16(vload16(0, src_row_ptr)); \ + acc += (src0.lo + input_offset) * ((int8)weights_values0.s0 + weight_offset); \ + acc += ((int8)(src0.s1234, src0.s5678) + input_offset) * ((int8)weights_values0.s1 + weight_offset); \ + acc += ((int8)(src0.s2345, src0.s6789) + input_offset) * ((int8)weights_values0.s2 + weight_offset); \ + acc += ((int8)(src0.s3456, src0.s789A) + input_offset) * ((int8)weights_values0.s3 + weight_offset); \ + acc += ((int8)(src0.s4567, src0.s89AB) + input_offset) * ((int8)weights_values0.s4 + weight_offset); \ + acc += ((int8)(src0.s5678, src0.s9ABC) + input_offset) * ((int8)weights_values0.s5 + weight_offset); \ + acc += ((int8)(src0.s6789, src0.sABCD) + input_offset) * ((int8)weights_values0.s6 + weight_offset); \ + acc += ((int8)(src0.s789A, src0.sBCDE) + input_offset) * ((int8)weights_values0.s7 + weight_offset); \ + acc += ((int8)(src0.s89AB, src0.sCDEF) + input_offset) * ((int8)weights_value1 + weight_offset); \ + }) + +#define CONVOLUTION1x9_STRIDE2(acc, src_row_ptr, weights_row_ptr) \ + ({ \ + int8 weights_values0 = convert_int8(vload8(0, weights_row_ptr)); \ + int weights_value1 = convert_int(*(weights_row_ptr + 8)); \ + int16 src0 = convert_int16(vload16(0, src_row_ptr)); \ + int8 src1 = convert_int8(vload8(0, src_row_ptr + 16)); \ + acc += (src0.even + input_offset) * ((int8)weights_values0.s0 + weight_offset); \ + acc += ((int8)(src0.s1357, src0.s9BDF) + input_offset) * ((int8)weights_values0.s1 + weight_offset); \ + acc += ((int8)(src0.s2468, src0.sACE, src1.s0) + input_offset) * ((int8)weights_values0.s2 + weight_offset); \ + acc += ((int8)(src0.s3579, src0.sBDF, src1.s1) + input_offset) * ((int8)weights_values0.s3 + weight_offset); \ + acc += ((int8)(src0.s468A, src0.sCE, src1.s02) + input_offset) * ((int8)weights_values0.s4 + weight_offset); \ + acc += ((int8)(src0.s579B, src0.sDF, src1.s13) + input_offset) * ((int8)weights_values0.s5 + weight_offset); \ + acc += ((int8)(src0.s68AC, src0.sE, src1.s024) + input_offset) * ((int8)weights_values0.s6 + weight_offset); \ + acc += ((int8)(src0.s79BD, src0.sF, src1.s135) + input_offset) * ((int8)weights_values0.s7 + weight_offset); \ + acc += ((int8)(src0.s8ACE, src1.s0246) + input_offset) * ((int8)weights_value1 + weight_offset); \ + }) + +#elif KERNEL_SIZE == 5 + +#if STRIDE_X == 1 +#define CONVOLUTION1x5(acc, src_row_ptr, weights_row_ptr) CONVOLUTION1x5_STRIDE1(acc, src_row_ptr, weights_row_ptr) +#elif STRIDE_X == 2 +#define CONVOLUTION1x5(acc, src_row_ptr, weights_row_ptr) CONVOLUTION1x5_STRIDE2(acc, src_row_ptr, weights_row_ptr) +#else /* STRIDE_X not equals 1 or 2 */ +#error "STRIDE_X larger than 2 is not supported" +#endif /* STRIDE_X */ + +#define CONVOLUTION1x5_STRIDE1(acc, src_row_ptr, weights_row_ptr) \ + ({ \ + int4 weights_values0 = convert_int4(vload4(0, weights_row_ptr)); \ + int weights_value1 = convert_int(*(weights_row_ptr + 4)); \ + int8 src0 = convert_int8(vload8(0, src_row_ptr)); \ + int4 src1 = convert_int4(vload4(0, src_row_ptr + 8)); \ + acc += (src0 + input_offset) * ((int8)weights_values0.s0 + weight_offset); \ + acc += ((int8)(src0.s1234, src0.s567, src1.s0) + input_offset) * ((int8)weights_values0.s1 + weight_offset); \ + acc += ((int8)(src0.s234, src0.s567, src1.s01) + input_offset) * ((int8)weights_values0.s2 + weight_offset); \ + acc += ((int8)(src0.s345, src0.s67, src1.s012) + input_offset) * ((int8)weights_values0.s3 + weight_offset); \ + acc += ((int8)(src0.s45, src0.s67, src1.s0123) + input_offset) * ((int8)weights_value1 + weight_offset); \ + }) + +#define CONVOLUTION1x5_STRIDE2(acc, src_row_ptr, weights_row_ptr) \ + ({ \ + int4 weights_values0 = convert_int4(vload4(0, weights_row_ptr)); \ + int weights_value1 = convert_int(*(weights_row_ptr + 4)); \ + int16 src0 = convert_int16(vload16(0, src_row_ptr)); \ + int4 src1 = convert_int4(vload4(0, src_row_ptr + 16)); \ + acc += (src0.even + input_offset) * ((int8)weights_values0.s0 + weight_offset); \ + acc += ((int8)(src0.s1357, src0.s9BDF) + input_offset) * ((int8)weights_values0.s1 + weight_offset); \ + acc += ((int8)(src0.s2468, src0.sACE, src1.s0) + input_offset) * ((int8)weights_values0.s2 + weight_offset); \ + acc += ((int8)(src0.s3579, src0.sBDF, src1.s1) + input_offset) * ((int8)weights_values0.s3 + weight_offset); \ + acc += ((int8)(src0.s468a, src0.sCE, src1.s02) + input_offset) * ((int8)weights_value1 + weight_offset); \ + }) + +#elif KERNEL_SIZE == 3 + +#if STRIDE_X == 1 +#define CONVOLUTION1x3(acc, src_row_ptr, weights_row_ptr) CONVOLUTION1x3_STRIDE1(acc, src_row_ptr, weights_row_ptr) +#elif STRIDE_X == 2 +#define CONVOLUTION1x3(acc, src_row_ptr, weights_row_ptr) CONVOLUTION1x3_STRIDE2(acc, src_row_ptr, weights_row_ptr) +#else /* STRIDE_X not equals 1 or 2 */ +#error "STRIDE_X larger than 2 is not supported" +#endif /* STRIDE_X */ + +#define CONVOLUTION1x3_STRIDE1(acc, src_row_ptr, weights_row_ptr) \ + ({ \ + int3 weights_values0 = convert_int3(vload3(0, weights_row_ptr)); \ + int8 src0 = convert_int8(vload8(0, src_row_ptr)); \ + int2 src1 = convert_int2(vload2(0, src_row_ptr + 8)); \ + acc += (src0 + input_offset) * ((int8)weights_values0.s0 + weight_offset); \ + acc += ((int8)(src0.s1234, src0.s567, src1.s0) + input_offset) * ((int8)weights_values0.s1 + weight_offset); \ + acc += ((int8)(src0.s234, src0.s567, src1.s01) + input_offset) * ((int8)weights_values0.s2 + weight_offset); \ + }) + +#define CONVOLUTION1x3_STRIDE2(acc, src_row_ptr, weights_row_ptr) \ + ({ \ + int3 weights_values0 = convert_int3(vload3(0, weights_row_ptr)); \ + int16 src0 = convert_int16(vload16(0, src_row_ptr)); \ + int src1 = convert_int(*(src_row_ptr + 16)); \ + acc += (src0.even + input_offset) * ((int8)weights_values0.s0 + weight_offset); \ + acc += ((int8)(src0.s1357, src0.s9BDF) + input_offset) * ((int8)weights_values0.s1 + weight_offset); \ + acc += ((int8)(src0.s2468, src0.sACE, src1) + input_offset) * ((int8)weights_values0.s2 + weight_offset); \ + }) + +#elif KERNEL_SIZE == 1 + +#if STRIDE_X == 3 +#define INPUT_PIXEL extract_input_stride3 +#elif STRIDE_X == 2 +#define INPUT_PIXEL extract_input_stride2 +#elif STRIDE_X == 1 +#define INPUT_PIXEL extract_input_stride1 + +#else /* STRIDE_X not equals 1, 2 or 3 */ +#error "Only support strides 1, 2 and 3" +#endif /* STRIDE_X */ + +/** Extracts a 1D horizontal vector from the input tensor with stride as 1. + * + * @param[in] input_pixel Pointer to the first pixel. + * + * @return extracted input pixels. + */ +inline uchar8 extract_input_stride1(__global const uchar *input_pixel) +{ + return vload8(0, input_pixel); +} + +/** Extracts a 1D horizontal vector from the input tensor with stride as 2. + * + * @param[in] input_pixel Pointer to the first pixel. + * + * @return extracted input pixels. + */ +inline uchar8 extract_input_stride2(__global const uchar *input_pixel) +{ + uchar16 temp = vload16(0, input_pixel); + return temp.s02468ace; +} + +/** Extracts a 1D horizontal vector from the input tensor with stride as 3 and 8-bit data size. + * + * @param[in] input_pixel Pointer to the first pixel. + * + * @return extracted input pixels. + */ +inline uchar8 extract_input_stride3(__global const uchar *input_pixel) +{ + uchar16 temp1 = vload16(0, input_pixel); + uchar16 temp2 = vload16(0, input_pixel + 12); + return (uchar8)(temp1.s0369, temp2.s0369); +} + +#else /* KERNEL_SIZE not equals 1, 3 , 5, 9 */ +#error "Only kernel sizes 1, 3, 5 and 9 are supported" +#endif /* KERNEL_SIZE */ + +/** This kernel performs a direct convolution to convolve the low three dimensions. + * + * @note The convolution stride x must be passed at compile time using -DSTRIDE_X e.g. -DSTRIDE_X=1 + * @note The third dimensions of the weights tensors must be passed at compile time using -DWEIGHTS_DEPTH + * @note If biases are used then -DHAS_BIAS has to be passed at compile time + * + * @param[in] src_ptr Pointer to the source tensor. Supported data types: QASYMM8 + * @param[in] src_stride_x Stride of the source tensor in X dimension (in bytes) + * @param[in] src_step_x src_stride_x * number of elements along X processed per workitem(in bytes) + * @param[in] src_stride_y Stride of the source tensor in Y dimension (in bytes) + * @param[in] src_step_y src_stride_y * number of elements along Y processed per workitem(in bytes) + * @param[in] src_stride_z Stride of the source tensor in Z dimension (in bytes) + * @param[in] src_step_z src_stride_z * number of elements along Z processed per workitem(in bytes) + * @param[in] src_offset_first_element_in_bytes The offset of the first element in the source tensor + * @param[out] dst_ptr Pointer to the destination tensor. Supported data types: same as @p src_ptr + * @param[in] dst_stride_x Stride of the destination tensor in X dimension (in bytes) + * @param[in] dst_step_x dst_stride_x * number of elements along X processed per workitem(in bytes) + * @param[in] dst_stride_y Stride of the destination tensor in Y dimension (in bytes) + * @param[in] dst_step_y dst_stride_y * number of elements along Z processed per workitem(in bytes) + * @param[in] dst_stride_z Stride of the destination tensor in Z dimension (in bytes) + * @param[in] dst_step_z dst_stride_z * number of elements along Z processed per workitem(in bytes) + * @param[in] dst_offset_first_element_in_bytes The offset of the first element in the destination tensor + * @param[in] weights_ptr Pointer to the weights tensor. Supported data types: same as @p weights_ptr + * @param[in] weights_stride_x Stride of the weights tensor in X dimension (in bytes) + * @param[in] weights_step_x weights_stride_x * number of elements along X processed per workitem(in bytes) + * @param[in] weights_stride_y Stride of the weights tensor in Y dimension (in bytes) + * @param[in] weights_step_y weights_stride_y * number of elements along y processed per workitem(in bytes) + * @param[in] weights_stride_z Stride of the weights tensor in Z dimension (in bytes) + * @param[in] weights_step_z weights_stride_z * number of elements along Z processed per workitem(in bytes) + * @param[in] weights_offset_first_element_in_bytes The offset of the first element in the weights tensor + * @param[in] biases_ptr Pointer to the biases tensor. Supported data types: S32 + * @param[in] biases_stride_x Stride of the biases tensor in X dimension (in bytes) + * @param[in] biases_step_x biases_stride_x * number of elements along X processed per workitem(in bytes) + * @param[in] biases_offset_first_element_in_bytes The offset of the first element in the biases tensor + * @param[in] weights_stride_w Stride of the weights tensor in the 4th dimension + * @param[in] input_offset Input offset quantization parameter + * @param[in] weight_offset Weights offset quantization parameter + * @param[in] output_offset Output offset quantization parameter + * @param[in] output_multiplier Output integer multiplier quantization parameter + * @param[in] output_shift Output integer shift quantization parameter + */ +__kernel void direct_convolution_quantized( + TENSOR3D_DECLARATION(src), + TENSOR3D_DECLARATION(dst), + TENSOR3D_DECLARATION(weights), +#ifdef HAS_BIAS + VECTOR_DECLARATION(biases), +#endif /* defined(HAS_BIAS) */ + unsigned int weights_stride_w, + int input_offset, + int weight_offset, + int output_offset, + int output_multiplier, + int output_shift) +{ + Image src = CONVERT_TO_IMAGE_STRUCT(src); + Tensor3D weights = CONVERT_TO_TENSOR3D_STRUCT_NO_STEP(weights); + Tensor3D dst = CONVERT_TO_TENSOR3D_STRUCT(dst); + + int8 pixels0 = 0; + + __global uchar *weights_addr = (__global uchar *)tensor3D_offset(&weights, 0, 0, 0); + __global uchar *src_addr = (__global uchar *)offset(&src, 0, 0); + + const int kernel_index = get_global_id(2); + weights_addr += kernel_index * weights_stride_w; + + for(volatile int d = 0; d < WEIGHTS_DEPTH; ++d) + { +#if KERNEL_SIZE == 9 + CONVOLUTION1x9(pixels0, (__global uchar *)(src_addr + 0 * src_stride_y), (__global uchar *)(weights_addr + 0 * weights_stride_y)); + CONVOLUTION1x9(pixels0, (__global uchar *)(src_addr + 1 * src_stride_y), (__global uchar *)(weights_addr + 1 * weights_stride_y)); + CONVOLUTION1x9(pixels0, (__global uchar *)(src_addr + 2 * src_stride_y), (__global uchar *)(weights_addr + 2 * weights_stride_y)); + CONVOLUTION1x9(pixels0, (__global uchar *)(src_addr + 3 * src_stride_y), (__global uchar *)(weights_addr + 3 * weights_stride_y)); + CONVOLUTION1x9(pixels0, (__global uchar *)(src_addr + 4 * src_stride_y), (__global uchar *)(weights_addr + 4 * weights_stride_y)); + CONVOLUTION1x9(pixels0, (__global uchar *)(src_addr + 5 * src_stride_y), (__global uchar *)(weights_addr + 5 * weights_stride_y)); + CONVOLUTION1x9(pixels0, (__global uchar *)(src_addr + 6 * src_stride_y), (__global uchar *)(weights_addr + 6 * weights_stride_y)); + CONVOLUTION1x9(pixels0, (__global uchar *)(src_addr + 7 * src_stride_y), (__global uchar *)(weights_addr + 7 * weights_stride_y)); + CONVOLUTION1x9(pixels0, (__global uchar *)(src_addr + 8 * src_stride_y), (__global uchar *)(weights_addr + 8 * weights_stride_y)); +#elif KERNEL_SIZE == 5 + CONVOLUTION1x5(pixels0, (__global uchar *)src_addr, (__global uchar *)weights_addr); + CONVOLUTION1x5(pixels0, (__global uchar *)(src_addr + 1 * src_stride_y), (__global uchar *)(weights_addr + 1 * weights_stride_y)); + CONVOLUTION1x5(pixels0, (__global uchar *)(src_addr + 2 * src_stride_y), (__global uchar *)(weights_addr + 2 * weights_stride_y)); + CONVOLUTION1x5(pixels0, (__global uchar *)(src_addr + 3 * src_stride_y), (__global uchar *)(weights_addr + 3 * weights_stride_y)); + CONVOLUTION1x5(pixels0, (__global uchar *)(src_addr + 4 * src_stride_y), (__global uchar *)(weights_addr + 4 * weights_stride_y)); +#elif KERNEL_SIZE == 3 + CONVOLUTION1x3(pixels0, (__global uchar *)(src_addr + 0 * src_stride_y), (__global uchar *)(weights_addr + 0 * weights_stride_y)); + CONVOLUTION1x3(pixels0, (__global uchar *)(src_addr + 1 * src_stride_y), (__global uchar *)(weights_addr + 1 * weights_stride_y)); + CONVOLUTION1x3(pixels0, (__global uchar *)(src_addr + 2 * src_stride_y), (__global uchar *)(weights_addr + 2 * weights_stride_y)); +#elif KERNEL_SIZE == 1 + int weight = convert_int(*(__global uchar *)weights_addr); + int8 input_pixel = convert_int8(INPUT_PIXEL((__global uchar *)src_addr)); + pixels0 += (input_pixel + input_offset) * ((int8)weight + weight_offset); +#endif /* (KERNEL_SIZE == 1) || (KERNEL_SIZE == 3) || (KERNEL_SIZE == 5) */ + + src_addr += src_stride_z; + weights_addr += weights_stride_z; + } + +#ifdef HAS_BIAS + Vector biases = CONVERT_TO_VECTOR_STRUCT_NO_STEP(biases); + __global int *bias_addr = ((__global int *)(vector_offset(&biases, kernel_index))); + pixels0 += (int8)(*bias_addr); +#endif /* defined(HAS_BIAS) */ + + pixels0 = ASYMM_MULT_BY_QUANT_MULTIPLIER_LESS_THAN_ONE(pixels0, output_multiplier, output_shift, 8); + pixels0 = pixels0 + output_offset; + + vstore8(convert_uchar8_sat(pixels0), 0, (__global uchar *)dst.ptr); +} +#endif // defined(DATA_TYPE) && defined(STRIDE_X) && defined(WEIGHTS_DEPTH) -- cgit v1.2.1