diff options
Diffstat (limited to 'src/core/CL/cl_kernels/concatenate.cl')
-rw-r--r-- | src/core/CL/cl_kernels/concatenate.cl | 103 |
1 files changed, 50 insertions, 53 deletions
diff --git a/src/core/CL/cl_kernels/concatenate.cl b/src/core/CL/cl_kernels/concatenate.cl index 7c6c8d211a..19494b109f 100644 --- a/src/core/CL/cl_kernels/concatenate.cl +++ b/src/core/CL/cl_kernels/concatenate.cl @@ -53,7 +53,9 @@ inline VEC_QUANT requantize(VEC_QUANT input, float in_offset, float out_offset, #error "Element size not supported" #endif // ELEMENT_SIZE -#if VEC_SIZE == 2 +#if VEC_SIZE == 1 +#define SEQ ((int)(0)) +#elif VEC_SIZE == 2 #define SEQ ((int2)(0, 1)) #elif VEC_SIZE == 4 #define SEQ ((int4)(0, 1, 2, 3)) @@ -69,7 +71,7 @@ inline VEC_QUANT requantize(VEC_QUANT input, float in_offset, float out_offset, * * @note The data type has to be passed at compile time using -DDATA_TYPE. i.e. -DDATA_TYPE=float * @note Vector size has to be passed at compile time using -DVEC_SIZE. i.e. -DVEC_SIZE=16 - * @note The offset for the first spatial dimension has to be passed at compile time using -DWIDTH_OFFSET. i.e. -DWIDTH_OFFSET=128 + * @note Leftover vector size has to be passed at compile time using -DVEC_SIZE_LEFTOVER. e.g. -DVEC_SIZE_LEFTOVER=3. It is defined as the remainder between the input's first dimension and VEC_SIZE * @note Tensor depth should be given as a preprocessor argument using -DDEPTH=size. e.g. -DDEPTH=16 * @note First input tensor width should be given as a preprocessor argument using -DINPUT1_WIDTH=width. e.g. -DINPUT1_WIDTH=8 * @@ -103,34 +105,29 @@ inline VEC_QUANT requantize(VEC_QUANT input, float in_offset, float out_offset, * @param[in] dst_stride_w Stride of the destination tensor in Z dimension (in bytes) * @param[in] dst_step_w output_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] src1_pad_right Right paddings of the first input tensor in unit of elements - * @param[in] src1_pad_left Left paddings of the second input tensor in unit of elements */ __kernel void concatenate_width_x2( TENSOR4D_DECLARATION(src1), TENSOR4D_DECLARATION(src2), - TENSOR4D_DECLARATION(dst), - uint src1_pad_right, - uint src2_pad_left) + TENSOR4D_DECLARATION(dst)) { - Tensor4D dst = CONVERT_TO_TENSOR4D_STRUCT(dst, DEPTH); - // Calculate input indices - const int x = get_global_id(0) * (int)VEC_SIZE; + const int x = max((int)(get_global_id(0) * VEC_SIZE - (VEC_SIZE - VEC_SIZE_LEFTOVER) % VEC_SIZE), 0); const int y = get_global_id(1); const int z = get_global_id(2) % (int)DEPTH; const int w = get_global_id(2) / (int)DEPTH; - const int x1 = min(x, (int)INPUT1_WIDTH + (int)src1_pad_right - (int)VEC_SIZE); - const int x2 = max(x - (int)INPUT1_WIDTH, -(int)src2_pad_left); + const int x1 = min(x, (int)INPUT1_WIDTH - (int)VEC_SIZE); + const int x2 = max(x - (int)INPUT1_WIDTH, 0); // Calculate inputs and output addresses - const __global uchar *in1_ptr = src1_ptr + (int)src1_offset_first_element_in_bytes + x1 * (int)src1_stride_x + y * (int)src1_stride_y + z * (int)src1_stride_z + w * (int)src1_stride_w; - const __global uchar *in2_ptr = src2_ptr + (int)src2_offset_first_element_in_bytes + x2 * (int)src2_stride_x + y * (int)src2_stride_y + z * (int)src2_stride_z + w * (int)src2_stride_w; + const __global uchar *dst_addr = dst_ptr + (int)dst_offset_first_element_in_bytes + x * sizeof(DATA_TYPE) + y * (int)dst_stride_y + z * (int)dst_stride_z + w * (int)dst_stride_w; + const __global uchar *src1_addr = src1_ptr + (int)src1_offset_first_element_in_bytes + x1 * sizeof(DATA_TYPE) + y * (int)src1_stride_y + z * (int)src1_stride_z + w * (int)src1_stride_w; + const __global uchar *src2_addr = src2_ptr + (int)src2_offset_first_element_in_bytes + x2 * sizeof(DATA_TYPE) + y * (int)src2_stride_y + z * (int)src2_stride_z + w * (int)src2_stride_w; VEC_DATA_TYPE(DATA_TYPE, VEC_SIZE) - src1_values = VLOAD(VEC_SIZE)(0, (__global DATA_TYPE *)in1_ptr); + src1_values = VLOAD(VEC_SIZE)(0, (__global DATA_TYPE *)src1_addr); VEC_DATA_TYPE(DATA_TYPE, VEC_SIZE) - src2_values = VLOAD(VEC_SIZE)(0, (__global DATA_TYPE *)in2_ptr); + src2_values = VLOAD(VEC_SIZE)(0, (__global DATA_TYPE *)src2_addr); #if defined(OFFSET_IN1) && defined(OFFSET_IN2) && defined(OFFSET_OUT) && defined(SCALE_IN1) && defined(SCALE_IN2) && defined(SCALE_OUT) src1_values = requantize(src1_values, OFFSET_IN1, OFFSET_OUT, SCALE_IN1, SCALE_OUT); @@ -138,10 +135,14 @@ __kernel void concatenate_width_x2( #endif /* defined(OFFSET_IN1) && defined(OFFSET_IN2) && defined(OFFSET_OUT) && defined(SCALE_IN1) && defined(SCALE_IN2) && defined(SCALE_OUT) */ const VEC_DATA_TYPE(int, VEC_SIZE) x_coords = SEQ + (VEC_DATA_TYPE(int, VEC_SIZE))(x); const VEC_DATA_TYPE(COND_DATA_TYPE, VEC_SIZE) cond = CONVERT(x_coords < (VEC_DATA_TYPE(int, VEC_SIZE))(INPUT1_WIDTH), VEC_DATA_TYPE(COND_DATA_TYPE, VEC_SIZE)); - const VEC_DATA_TYPE(DATA_TYPE, VEC_SIZE) values = select(src2_values, src1_values, cond); - VSTORE(VEC_SIZE) - (values, 0, (__global DATA_TYPE *)dst.ptr); + // Rotate src1/2_values, if values0 is a combination of src1_values and src2_values. + src1_values = (x < INPUT1_WIDTH && x1 == INPUT1_WIDTH - VEC_SIZE) ? ROTATE(src1_values, VEC_SIZE, INPUT1_ROTATE_N) : src1_values; + src2_values = (x < INPUT1_WIDTH && x1 == INPUT1_WIDTH - VEC_SIZE) ? ROTATE(src2_values, VEC_SIZE, INPUT1_ROTATE_N) : src2_values; + + const VEC_DATA_TYPE(DATA_TYPE, VEC_SIZE) values0 = select(src2_values, src1_values, cond); + + STORE_VECTOR_SELECT(values, DATA_TYPE, dst_addr, VEC_SIZE, VEC_SIZE_LEFTOVER, VEC_SIZE_LEFTOVER != 0 && get_global_id(0) == 0) } #if defined(INPUT2_WIDTH) && defined(INPUT3_WIDTH) @@ -149,7 +150,7 @@ __kernel void concatenate_width_x2( * * @note The data type has to be passed at compile time using -DDATA_TYPE. i.e. -DDATA_TYPE=float * @note Vector size has to be passed at compile time using -DVEC_SIZE. i.e. -DVEC_SIZE=16 - * @note The offset for the first spatial dimension has to be passed at compile time using -DWIDTH_OFFSET. i.e. -DWIDTH_OFFSET=128 + * @note Leftover vector size has to be passed at compile time using -DVEC_SIZE_LEFTOVER. e.g. -DVEC_SIZE_LEFTOVER=3. It is defined as the remainder between the input's first dimension and VEC_SIZE * @note Tensor depth should be given as a preprocessor argument using -DDEPTH=size. e.g. -DDEPTH=16 * @note First input tensor width should be given as a preprocessor argument using -DINPUT1_WIDTH=width. e.g. -DINPUT1_WIDTH=8 * @note Second input tensor width should be given as a preprocessor argument using -DINPUT2_WIDTH=width. e.g. -DINPUT2_WIDTH=8 @@ -205,53 +206,40 @@ __kernel void concatenate_width_x2( * @param[in] dst_stride_w Stride of the destination tensor in Z dimension (in bytes) * @param[in] dst_step_w output_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] src1_pad_right Right paddings of the first input tensor in unit of elements - * @param[in] src2_pad_left Left paddings of the second input tensor in unit of elements - * @param[in] src2_pad_right Right paddings of the second input tensor in unit of elements - * @param[in] src3_pad_left Left paddings of the third input tensor in unit of elements - * @param[in] src3_pad_right Right paddings of the third input tensor in unit of elements - * @param[in] src4_pad_left Left paddings of the fourth input tensor in unit of elements */ __kernel void concatenate_width_x4( TENSOR4D_DECLARATION(src1), TENSOR4D_DECLARATION(src2), TENSOR4D_DECLARATION(src3), TENSOR4D_DECLARATION(src4), - TENSOR4D_DECLARATION(dst), - uint src1_pad_right, - uint src2_pad_left, - uint src2_pad_right, - uint src3_pad_left, - uint src3_pad_right, - uint src4_pad_left) + TENSOR4D_DECLARATION(dst)) { - Tensor4D dst = CONVERT_TO_TENSOR4D_STRUCT(dst, DEPTH); - // Calculate input indices - const int x = get_global_id(0) * (int)VEC_SIZE; + const int x = max((int)(get_global_id(0) * VEC_SIZE - (VEC_SIZE - VEC_SIZE_LEFTOVER) % VEC_SIZE), 0); const int y = get_global_id(1); const int z = get_global_id(2) % (int)DEPTH; const int w = get_global_id(2) / (int)DEPTH; - const int x1 = min(x, (int)INPUT1_WIDTH + (int)src1_pad_right - (int)VEC_SIZE); - const int x2 = min(max(x - (int)INPUT1_WIDTH, -(int)src2_pad_left), (int)INPUT2_WIDTH + (int)src2_pad_right - (int)VEC_SIZE); - const int x3 = min(max(x - (int)INPUT1_WIDTH - (int)INPUT2_WIDTH, -(int)src3_pad_left), (int)INPUT3_WIDTH + (int)src3_pad_right - (int)VEC_SIZE); - const int x4 = max(x - (int)INPUT1_WIDTH - (int)INPUT2_WIDTH - (int)INPUT3_WIDTH, -(int)src4_pad_left); + const int x1 = min(x, (int)INPUT1_WIDTH - (int)VEC_SIZE); + const int x2 = min(max(x - (int)INPUT1_WIDTH, 0), (int)INPUT2_WIDTH - (int)VEC_SIZE); + const int x3 = min(max(x - (int)INPUT1_WIDTH - (int)INPUT2_WIDTH, 0), (int)INPUT3_WIDTH - (int)VEC_SIZE); + const int x4 = max(x - (int)INPUT1_WIDTH - (int)INPUT2_WIDTH - (int)INPUT3_WIDTH, 0); // Calculate inputs and output addresses - const __global uchar *in1_ptr = src1_ptr + (int)src1_offset_first_element_in_bytes + x1 * (int)src1_stride_x + y * (int)src1_stride_y + z * (int)src1_stride_z + w * (int)src1_stride_w; - const __global uchar *in2_ptr = src2_ptr + (int)src2_offset_first_element_in_bytes + x2 * (int)src2_stride_x + y * (int)src2_stride_y + z * (int)src2_stride_z + w * (int)src2_stride_w; - const __global uchar *in3_ptr = src3_ptr + (int)src3_offset_first_element_in_bytes + x3 * (int)src3_stride_x + y * (int)src3_stride_y + z * (int)src3_stride_z + w * (int)src3_stride_w; - const __global uchar *in4_ptr = src4_ptr + (int)src4_offset_first_element_in_bytes + x4 * (int)src4_stride_x + y * (int)src4_stride_y + z * (int)src4_stride_z + w * (int)src4_stride_w; + const __global uchar *dst_addr = dst_ptr + (int)dst_offset_first_element_in_bytes + x * sizeof(DATA_TYPE) + y * (int)dst_stride_y + z * (int)dst_stride_z + w * (int)dst_stride_w; + const __global uchar *src1_addr = src1_ptr + (int)src1_offset_first_element_in_bytes + x1 * sizeof(DATA_TYPE) + y * (int)src1_stride_y + z * (int)src1_stride_z + w * (int)src1_stride_w; + const __global uchar *src2_addr = src2_ptr + (int)src2_offset_first_element_in_bytes + x2 * sizeof(DATA_TYPE) + y * (int)src2_stride_y + z * (int)src2_stride_z + w * (int)src2_stride_w; + const __global uchar *src3_addr = src3_ptr + (int)src3_offset_first_element_in_bytes + x3 * sizeof(DATA_TYPE) + y * (int)src3_stride_y + z * (int)src3_stride_z + w * (int)src3_stride_w; + const __global uchar *src4_addr = src4_ptr + (int)src4_offset_first_element_in_bytes + x4 * sizeof(DATA_TYPE) + y * (int)src4_stride_y + z * (int)src4_stride_z + w * (int)src4_stride_w; VEC_DATA_TYPE(DATA_TYPE, VEC_SIZE) - src1_values = VLOAD(VEC_SIZE)(0, (__global DATA_TYPE *)in1_ptr); + src1_values = VLOAD(VEC_SIZE)(0, (__global DATA_TYPE *)src1_addr); VEC_DATA_TYPE(DATA_TYPE, VEC_SIZE) - src2_values = VLOAD(VEC_SIZE)(0, (__global DATA_TYPE *)in2_ptr); + src2_values = VLOAD(VEC_SIZE)(0, (__global DATA_TYPE *)src2_addr); VEC_DATA_TYPE(DATA_TYPE, VEC_SIZE) - src3_values = VLOAD(VEC_SIZE)(0, (__global DATA_TYPE *)in3_ptr); + src3_values = VLOAD(VEC_SIZE)(0, (__global DATA_TYPE *)src3_addr); VEC_DATA_TYPE(DATA_TYPE, VEC_SIZE) - src4_values = VLOAD(VEC_SIZE)(0, (__global DATA_TYPE *)in4_ptr); + src4_values = VLOAD(VEC_SIZE)(0, (__global DATA_TYPE *)src4_addr); #if defined(OFFSET_IN1) && defined(OFFSET_OUT) && defined(SCALE_IN1) && defined(SCALE_OUT) && defined(OFFSET_IN2) && defined(SCALE_IN2) && defined(OFFSET_IN3) && defined(SCALE_IN3) && defined(OFFSET_IN4) && defined(SCALE_IN4) src1_values = requantize(src1_values, OFFSET_IN1, OFFSET_OUT, SCALE_IN1, SCALE_OUT); @@ -266,13 +254,22 @@ __kernel void concatenate_width_x4( const VEC_DATA_TYPE(COND_DATA_TYPE, VEC_SIZE) cond_in3 = CONVERT(x_coords < (VEC_DATA_TYPE(int, VEC_SIZE))(INPUT1_WIDTH + INPUT2_WIDTH), VEC_DATA_TYPE(COND_DATA_TYPE, VEC_SIZE)); const VEC_DATA_TYPE(COND_DATA_TYPE, VEC_SIZE) cond_in4 = CONVERT(x_coords < (VEC_DATA_TYPE(int, VEC_SIZE))(INPUT1_WIDTH + INPUT2_WIDTH + INPUT3_WIDTH), VEC_DATA_TYPE(COND_DATA_TYPE, VEC_SIZE)); + // Rotate src1/2_values, if values0 is a combination of src1_values and src2_values. + src1_values = (x < INPUT1_WIDTH && x1 == INPUT1_WIDTH - VEC_SIZE) ? ROTATE(src1_values, VEC_SIZE, INPUT1_ROTATE_N) : src1_values; + src2_values = (x < INPUT1_WIDTH && x1 == INPUT1_WIDTH - VEC_SIZE) ? ROTATE(src2_values, VEC_SIZE, INPUT1_ROTATE_N) : src2_values; + // Rotate src2/3_values, if values0 is a combination of src2_values and src3_values. + src2_values = (x < INPUT1_WIDTH + INPUT2_WIDTH && x2 == INPUT2_WIDTH - VEC_SIZE) ? ROTATE(src2_values, VEC_SIZE, INPUT2_ROTATE_N) : src2_values; + src3_values = (x < INPUT1_WIDTH + INPUT2_WIDTH && x2 == INPUT2_WIDTH - VEC_SIZE) ? ROTATE(src3_values, VEC_SIZE, INPUT2_ROTATE_N) : src3_values; + // Rotate src3/4_values, if values0 is a combination of src3_values and src4_values. + src3_values = (x < INPUT1_WIDTH + INPUT2_WIDTH + INPUT3_WIDTH && x3 == INPUT3_WIDTH - VEC_SIZE) ? ROTATE(src3_values, VEC_SIZE, INPUT3_ROTATE_N) : src3_values; + src4_values = (x < INPUT1_WIDTH + INPUT2_WIDTH + INPUT3_WIDTH && x3 == INPUT3_WIDTH - VEC_SIZE) ? ROTATE(src4_values, VEC_SIZE, INPUT3_ROTATE_N) : src4_values; + VEC_DATA_TYPE(DATA_TYPE, VEC_SIZE) - values = select(src2_values, src1_values, cond_in2); - values = select(src3_values, values, cond_in3); - values = select(src4_values, values, cond_in4); + values0 = select(src2_values, src1_values, cond_in2); + values0 = select(src3_values, values0, cond_in3); + values0 = select(src4_values, values0, cond_in4); - VSTORE(VEC_SIZE) - (values, 0, (__global DATA_TYPE *)dst.ptr); + STORE_VECTOR_SELECT(values, DATA_TYPE, dst_addr, VEC_SIZE, VEC_SIZE_LEFTOVER, VEC_SIZE_LEFTOVER != 0 && get_global_id(0) == 0) } #endif /* defined(INPUT2_WIDTH) && defined(INPUT3_WIDTH) */ #endif /* defined(INPUT1_WIDTH) */ |