Vidhya Sudhan Loganathan | 5e96be7 | 2018-12-18 14:17:00 +0000 | [diff] [blame] | 1 | /* |
| 2 | * Copyright (c) 2018 ARM Limited. |
| 3 | * |
| 4 | * SPDX-License-Identifier: MIT |
| 5 | * |
| 6 | * Permission is hereby granted, free of charge, to any person obtaining a copy |
| 7 | * of this software and associated documentation files (the "Software"), to |
| 8 | * deal in the Software without restriction, including without limitation the |
| 9 | * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or |
| 10 | * sell copies of the Software, and to permit persons to whom the Software is |
| 11 | * furnished to do so, subject to the following conditions: |
| 12 | * |
| 13 | * The above copyright notice and this permission notice shall be included in all |
| 14 | * copies or substantial portions of the Software. |
| 15 | * |
| 16 | * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR |
| 17 | * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, |
| 18 | * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE |
| 19 | * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER |
| 20 | * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, |
| 21 | * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE |
| 22 | * SOFTWARE. |
| 23 | */ |
| 24 | #include "helpers.h" |
| 25 | |
| 26 | #if defined(VECTOR_SIZE) && defined(START) && defined(STEP) && defined(DATA_TYPE) |
| 27 | /** Generates a sequence of numbers starting from START and extends by increments of 'STEP' up to but not including 'END'. |
| 28 | * |
| 29 | * @note starting value of the sequence must be given as a preprocessor argument using -DSTART=value. e.g. -DSTART=0 |
| 30 | * @note difference between consequtive elements of the sequence must be given as a preprocessor argument using -DSTEP=value. e.g. -DSTEP=1 |
| 31 | * @note Datatype must be given as a preprocessor argument using -DDATA_TYPE=type. e.g. -DDATA_TYPE=short |
| 32 | * @note vector size supported by the device must be given as a preprocessor argument using -DVECTOR_SIZE=value. e.g. -DDATA_TYPE=4 |
| 33 | * |
| 34 | * @param[out] out_ptr Pointer to the destination tensor. Supported data types: U8/S8/U16/S16/U32/S32/F16/F32. |
| 35 | * @param[in] out_stride_x Stride of the destination tensor in X dimension (in bytes) |
| 36 | * @param[in] out_step_x dst_stride_x * number of elements along X processed per workitem(in bytes) |
| 37 | * @param[in] out_offset_first_element_in_bytes The offset of the first element in the destination tensor |
| 38 | */ |
| 39 | __kernel void range( |
| 40 | VECTOR_DECLARATION(out)) |
| 41 | { |
| 42 | uint id = get_global_id(0) * VECTOR_SIZE; |
| 43 | __global void *dst_ptr = out_ptr + out_offset_first_element_in_bytes + id * sizeof(DATA_TYPE); |
| 44 | #if VECTOR_SIZE == 1 |
| 45 | DATA_TYPE seq; |
| 46 | seq = (DATA_TYPE)START + (DATA_TYPE)id * (DATA_TYPE)STEP; |
| 47 | |
| 48 | *((__global DATA_TYPE *)dst_ptr) = seq; |
| 49 | #else // VECTOR_SIZE == 1 |
| 50 | VEC_DATA_TYPE(DATA_TYPE, VECTOR_SIZE) |
| 51 | seq; |
| 52 | |
| 53 | seq.s0 = ((DATA_TYPE)START + (DATA_TYPE)id * (DATA_TYPE)STEP); |
| 54 | #if VECTOR_SIZE > 1 |
| 55 | seq.s1 = seq.s0 + (DATA_TYPE)STEP; |
| 56 | #if VECTOR_SIZE > 2 |
| 57 | seq.s2 = seq.s1 + (DATA_TYPE)STEP; |
| 58 | #if VECTOR_SIZE > 3 |
| 59 | seq.s3 = seq.s2 + (DATA_TYPE)STEP; |
| 60 | #if VECTOR_SIZE > 4 |
| 61 | seq.s4 = seq.s3 + (DATA_TYPE)STEP; |
| 62 | #if VECTOR_SIZE > 5 |
| 63 | seq.s5 = seq.s4 + (DATA_TYPE)STEP; |
| 64 | #if VECTOR_SIZE > 6 |
| 65 | seq.s6 = seq.s5 + (DATA_TYPE)STEP; |
| 66 | #if VECTOR_SIZE > 7 |
| 67 | seq.s7 = seq.s6 + (DATA_TYPE)STEP; |
| 68 | #endif // VECTOR_SIZE > 7 |
| 69 | #endif // VECTOR_SIZE > 6 |
| 70 | #endif // VECTOR_SIZE > 5 |
| 71 | #endif // VECTOR_SIZE > 4 |
| 72 | #endif // VECTOR_SIZE > 3 |
| 73 | #endif // VECTOR_SIZE > 2 |
| 74 | #endif // VECTOR_SIZE > 1 |
| 75 | VSTORE(VECTOR_SIZE) |
| 76 | (seq, 0, ((__global DATA_TYPE *)dst_ptr)); |
| 77 | #endif //VECTOR_SIZE == 1 |
| 78 | } |
| 79 | |
| 80 | #if defined(OFFSET_OUT) && defined(SCALE_OUT) |
| 81 | |
| 82 | #define CONVERT_RTE(x, type) (convert_##type##_rte((x))) |
| 83 | #define CONVERT_DOWN(x, type) CONVERT_RTE(x, type) |
| 84 | |
| 85 | /** Generates a sequence of numbers starting from START and extends by increments of 'STEP' up to but not including 'END'. |
| 86 | * |
| 87 | * @note starting value of the sequence must be given as a preprocessor argument using -DSTART=value. e.g. -DSTART=0 |
| 88 | * @note difference between consequtive elements of the sequence must be given as a preprocessor argument using -DSTEP=value. e.g. -DSTEP=1 |
| 89 | * @note Datatype must be given as a preprocessor argument using -DDATA_TYPE=type. e.g. -DDATA_TYPE=short |
| 90 | * @note vector size supported by the device must be given as a preprocessor argument using -DVECTOR_SIZE=vector_size. e.g. -DDATA_TYPE=4 |
| 91 | * @note The quantization offset of the output must be passed at compile time using -DOFFSET_OUT, i.e. -DOFFSET_OUT=10 |
| 92 | * @note The quantization scale of the output must be passed at compile time using -DSCALE_OUT, i.e. -DSCALE_OUT=10 |
| 93 | * |
| 94 | * @param[out] out_ptr Pointer to the destination tensor. Supported data types: QASYMM8. |
| 95 | * @param[in] out_stride_x Stride of the destination tensor in X dimension (in bytes) |
| 96 | * @param[in] out_step_x dst_stride_x * number of elements along X processed per workitem(in bytes) |
| 97 | * @param[in] out_offset_first_element_in_bytes The offset of the first element in the destination tensor |
| 98 | */ |
| 99 | __kernel void range_quantized( |
| 100 | VECTOR_DECLARATION(out)) |
| 101 | { |
| 102 | size_t id = get_global_id(0) * VECTOR_SIZE; |
| 103 | __global void *dst_ptr = out_ptr + out_offset_first_element_in_bytes + id * sizeof(DATA_TYPE); |
| 104 | #if VECTOR_SIZE == 1 |
| 105 | float seq; |
| 106 | seq = (float)START + (float)id * (float)STEP; |
| 107 | seq = (DATA_TYPE)(int)(seq / ((float)SCALE_OUT) + (float)OFFSET_OUT); |
| 108 | seq = max(0.0f, min(seq, 255.0f)); |
| 109 | *((__global uchar *)dst_ptr) = CONVERT_SAT(CONVERT_DOWN(seq, int), uchar); |
| 110 | #else // VECTOR_SIZE == 1 |
| 111 | VEC_DATA_TYPE(float, VECTOR_SIZE) |
| 112 | seq; |
| 113 | seq.s0 = (float)START + id * (float)STEP; |
| 114 | #if VECTOR_SIZE > 1 |
| 115 | seq.s1 = seq.s0 + (float)STEP; |
| 116 | #if VECTOR_SIZE > 2 |
| 117 | seq.s2 = seq.s1 + (float)STEP; |
| 118 | #if VECTOR_SIZE > 3 |
| 119 | seq.s3 = seq.s2 + (float)STEP; |
| 120 | #if VECTOR_SIZE > 4 |
| 121 | seq.s4 = seq.s3 + (float)STEP; |
| 122 | #if VECTOR_SIZE > 5 |
| 123 | seq.s5 = seq.s4 + (float)STEP; |
| 124 | #if VECTOR_SIZE > 6 |
| 125 | seq.s6 = seq.s5 + (float)STEP; |
| 126 | #if VECTOR_SIZE > 7 |
| 127 | seq.s7 = seq.s6 + (float)STEP; |
| 128 | #endif // VECTOR_SIZE > 7 |
| 129 | #endif // VECTOR_SIZE > 6 |
| 130 | #endif // VECTOR_SIZE > 5 |
| 131 | #endif // VECTOR_SIZE > 4 |
| 132 | #endif // VECTOR_SIZE > 3 |
| 133 | #endif // VECTOR_SIZE > 2 |
| 134 | #endif // VECTOR_SIZE > 1 |
| 135 | seq = seq / ((VEC_DATA_TYPE(float, VECTOR_SIZE))((float)SCALE_OUT)) + ((VEC_DATA_TYPE(float, VECTOR_SIZE))((float)OFFSET_OUT)); |
| 136 | seq = max((VEC_DATA_TYPE(float, VECTOR_SIZE))(0.0f), min(seq, (VEC_DATA_TYPE(float, VECTOR_SIZE))(255.0f))); |
| 137 | VEC_DATA_TYPE(uchar, VECTOR_SIZE) |
| 138 | res = CONVERT_SAT(CONVERT_DOWN(seq, VEC_DATA_TYPE(int, VECTOR_SIZE)), VEC_DATA_TYPE(uchar, VECTOR_SIZE)); |
| 139 | VSTORE(VECTOR_SIZE) |
| 140 | (res, 0, ((__global DATA_TYPE *)dst_ptr)); |
| 141 | #endif // VECTOR_SIZE == 1 |
| 142 | } |
| 143 | #endif // defined(OFFSET_OUT) && defined(SCALE_OUT) |
| 144 | |
| 145 | #endif // defined(VECTOR_SIZE) && defined(START) && defined(STEP) && defined(DATA_TYPE) |