Go to most recent revision | Details | Last modification | View Log | RSS feed
| Rev | Author | Line No. | Line |
|---|---|---|---|
| 56 | mjames | 1 | /* ---------------------------------------------------------------------- |
| 2 | * Project: CMSIS DSP Library |
||
| 3 | * Title: arm_mat_add_q31.c |
||
| 4 | * Description: Q31 matrix addition |
||
| 5 | * |
||
| 6 | * $Date: 27. January 2017 |
||
| 7 | * $Revision: V.1.5.1 |
||
| 8 | * |
||
| 9 | * Target Processor: Cortex-M cores |
||
| 10 | * -------------------------------------------------------------------- */ |
||
| 11 | /* |
||
| 12 | * Copyright (C) 2010-2017 ARM Limited or its affiliates. All rights reserved. |
||
| 13 | * |
||
| 14 | * SPDX-License-Identifier: Apache-2.0 |
||
| 15 | * |
||
| 16 | * Licensed under the Apache License, Version 2.0 (the License); you may |
||
| 17 | * not use this file except in compliance with the License. |
||
| 18 | * You may obtain a copy of the License at |
||
| 19 | * |
||
| 20 | * www.apache.org/licenses/LICENSE-2.0 |
||
| 21 | * |
||
| 22 | * Unless required by applicable law or agreed to in writing, software |
||
| 23 | * distributed under the License is distributed on an AS IS BASIS, WITHOUT |
||
| 24 | * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
||
| 25 | * See the License for the specific language governing permissions and |
||
| 26 | * limitations under the License. |
||
| 27 | */ |
||
| 28 | |||
| 29 | #include "arm_math.h" |
||
| 30 | |||
| 31 | /** |
||
| 32 | * @ingroup groupMatrix |
||
| 33 | */ |
||
| 34 | |||
| 35 | /** |
||
| 36 | * @addtogroup MatrixAdd |
||
| 37 | * @{ |
||
| 38 | */ |
||
| 39 | |||
| 40 | /** |
||
| 41 | * @brief Q31 matrix addition. |
||
| 42 | * @param[in] *pSrcA points to the first input matrix structure |
||
| 43 | * @param[in] *pSrcB points to the second input matrix structure |
||
| 44 | * @param[out] *pDst points to output matrix structure |
||
| 45 | * @return The function returns either |
||
| 46 | * <code>ARM_MATH_SIZE_MISMATCH</code> or <code>ARM_MATH_SUCCESS</code> based on the outcome of size checking. |
||
| 47 | * |
||
| 48 | * <b>Scaling and Overflow Behavior:</b> |
||
| 49 | * \par |
||
| 50 | * The function uses saturating arithmetic. |
||
| 51 | * Results outside of the allowable Q31 range [0x80000000 0x7FFFFFFF] will be saturated. |
||
| 52 | */ |
||
| 53 | |||
| 54 | arm_status arm_mat_add_q31( |
||
| 55 | const arm_matrix_instance_q31 * pSrcA, |
||
| 56 | const arm_matrix_instance_q31 * pSrcB, |
||
| 57 | arm_matrix_instance_q31 * pDst) |
||
| 58 | { |
||
| 59 | q31_t *pIn1 = pSrcA->pData; /* input data matrix pointer A */ |
||
| 60 | q31_t *pIn2 = pSrcB->pData; /* input data matrix pointer B */ |
||
| 61 | q31_t *pOut = pDst->pData; /* output data matrix pointer */ |
||
| 62 | q31_t inA1, inB1; /* temporary variables */ |
||
| 63 | |||
| 64 | #if defined (ARM_MATH_DSP) |
||
| 65 | |||
| 66 | q31_t inA2, inB2; /* temporary variables */ |
||
| 67 | q31_t out1, out2; /* temporary variables */ |
||
| 68 | |||
| 69 | #endif // #if defined (ARM_MATH_DSP) |
||
| 70 | |||
| 71 | uint32_t numSamples; /* total number of elements in the matrix */ |
||
| 72 | uint32_t blkCnt; /* loop counters */ |
||
| 73 | arm_status status; /* status of matrix addition */ |
||
| 74 | |||
| 75 | #ifdef ARM_MATH_MATRIX_CHECK |
||
| 76 | /* Check for matrix mismatch condition */ |
||
| 77 | if ((pSrcA->numRows != pSrcB->numRows) || |
||
| 78 | (pSrcA->numCols != pSrcB->numCols) || |
||
| 79 | (pSrcA->numRows != pDst->numRows) || (pSrcA->numCols != pDst->numCols)) |
||
| 80 | { |
||
| 81 | /* Set status as ARM_MATH_SIZE_MISMATCH */ |
||
| 82 | status = ARM_MATH_SIZE_MISMATCH; |
||
| 83 | } |
||
| 84 | else |
||
| 85 | #endif |
||
| 86 | { |
||
| 87 | /* Total number of samples in the input matrix */ |
||
| 88 | numSamples = (uint32_t) pSrcA->numRows * pSrcA->numCols; |
||
| 89 | |||
| 90 | #if defined (ARM_MATH_DSP) |
||
| 91 | |||
| 92 | /* Run the below code for Cortex-M4 and Cortex-M3 */ |
||
| 93 | |||
| 94 | /* Loop Unrolling */ |
||
| 95 | blkCnt = numSamples >> 2U; |
||
| 96 | |||
| 97 | |||
| 98 | /* First part of the processing with loop unrolling. Compute 4 outputs at a time. |
||
| 99 | ** a second loop below computes the remaining 1 to 3 samples. */ |
||
| 100 | while (blkCnt > 0U) |
||
| 101 | { |
||
| 102 | /* C(m,n) = A(m,n) + B(m,n) */ |
||
| 103 | /* Add, saturate and then store the results in the destination buffer. */ |
||
| 104 | /* Read values from source A */ |
||
| 105 | inA1 = pIn1[0]; |
||
| 106 | |||
| 107 | /* Read values from source B */ |
||
| 108 | inB1 = pIn2[0]; |
||
| 109 | |||
| 110 | /* Read values from source A */ |
||
| 111 | inA2 = pIn1[1]; |
||
| 112 | |||
| 113 | /* Add and saturate */ |
||
| 114 | out1 = __QADD(inA1, inB1); |
||
| 115 | |||
| 116 | /* Read values from source B */ |
||
| 117 | inB2 = pIn2[1]; |
||
| 118 | |||
| 119 | /* Read values from source A */ |
||
| 120 | inA1 = pIn1[2]; |
||
| 121 | |||
| 122 | /* Add and saturate */ |
||
| 123 | out2 = __QADD(inA2, inB2); |
||
| 124 | |||
| 125 | /* Read values from source B */ |
||
| 126 | inB1 = pIn2[2]; |
||
| 127 | |||
| 128 | /* Store result in destination */ |
||
| 129 | pOut[0] = out1; |
||
| 130 | pOut[1] = out2; |
||
| 131 | |||
| 132 | /* Read values from source A */ |
||
| 133 | inA2 = pIn1[3]; |
||
| 134 | |||
| 135 | /* Read values from source B */ |
||
| 136 | inB2 = pIn2[3]; |
||
| 137 | |||
| 138 | /* Add and saturate */ |
||
| 139 | out1 = __QADD(inA1, inB1); |
||
| 140 | out2 = __QADD(inA2, inB2); |
||
| 141 | |||
| 142 | /* Store result in destination */ |
||
| 143 | pOut[2] = out1; |
||
| 144 | pOut[3] = out2; |
||
| 145 | |||
| 146 | /* update pointers to process next sampels */ |
||
| 147 | pIn1 += 4U; |
||
| 148 | pIn2 += 4U; |
||
| 149 | pOut += 4U; |
||
| 150 | |||
| 151 | /* Decrement the loop counter */ |
||
| 152 | blkCnt--; |
||
| 153 | } |
||
| 154 | |||
| 155 | /* If the numSamples is not a multiple of 4, compute any remaining output samples here. |
||
| 156 | ** No loop unrolling is used. */ |
||
| 157 | blkCnt = numSamples % 0x4U; |
||
| 158 | |||
| 159 | #else |
||
| 160 | |||
| 161 | /* Run the below code for Cortex-M0 */ |
||
| 162 | |||
| 163 | /* Initialize blkCnt with number of samples */ |
||
| 164 | blkCnt = numSamples; |
||
| 165 | |||
| 166 | |||
| 167 | #endif /* #if defined (ARM_MATH_DSP) */ |
||
| 168 | |||
| 169 | while (blkCnt > 0U) |
||
| 170 | { |
||
| 171 | /* C(m,n) = A(m,n) + B(m,n) */ |
||
| 172 | /* Add, saturate and then store the results in the destination buffer. */ |
||
| 173 | inA1 = *pIn1++; |
||
| 174 | inB1 = *pIn2++; |
||
| 175 | |||
| 176 | inA1 = __QADD(inA1, inB1); |
||
| 177 | |||
| 178 | /* Decrement the loop counter */ |
||
| 179 | blkCnt--; |
||
| 180 | |||
| 181 | *pOut++ = inA1; |
||
| 182 | |||
| 183 | } |
||
| 184 | |||
| 185 | /* set status as ARM_MATH_SUCCESS */ |
||
| 186 | status = ARM_MATH_SUCCESS; |
||
| 187 | } |
||
| 188 | |||
| 189 | /* Return to application */ |
||
| 190 | return (status); |
||
| 191 | } |
||
| 192 | |||
| 193 | /** |
||
| 194 | * @} end of MatrixAdd group |
||
| 195 | */ |