Aded CMSIS5 DSP and NN folder. Needs some work
Embed:
(wiki syntax)
Show/hide line numbers
arm_offset_q7.c
00001 /* ---------------------------------------------------------------------- 00002 * Project: CMSIS DSP Library 00003 * Title: arm_offset_q7.c 00004 * Description: Q7 vector offset 00005 * 00006 * $Date: 27. January 2017 00007 * $Revision: V.1.5.1 00008 * 00009 * Target Processor: Cortex-M cores 00010 * -------------------------------------------------------------------- */ 00011 /* 00012 * Copyright (C) 2010-2017 ARM Limited or its affiliates. All rights reserved. 00013 * 00014 * SPDX-License-Identifier: Apache-2.0 00015 * 00016 * Licensed under the Apache License, Version 2.0 (the License); you may 00017 * not use this file except in compliance with the License. 00018 * You may obtain a copy of the License at 00019 * 00020 * www.apache.org/licenses/LICENSE-2.0 00021 * 00022 * Unless required by applicable law or agreed to in writing, software 00023 * distributed under the License is distributed on an AS IS BASIS, WITHOUT 00024 * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 00025 * See the License for the specific language governing permissions and 00026 * limitations under the License. 00027 */ 00028 00029 #include "arm_math.h" 00030 00031 /** 00032 * @ingroup groupMath 00033 */ 00034 00035 /** 00036 * @addtogroup offset 00037 * @{ 00038 */ 00039 00040 /** 00041 * @brief Adds a constant offset to a Q7 vector. 00042 * @param[in] *pSrc points to the input vector 00043 * @param[in] offset is the offset to be added 00044 * @param[out] *pDst points to the output vector 00045 * @param[in] blockSize number of samples in the vector 00046 * @return none. 00047 * 00048 * <b>Scaling and Overflow Behavior:</b> 00049 * \par 00050 * The function uses saturating arithmetic. 00051 * Results outside of the allowable Q7 range [0x80 0x7F] are saturated. 00052 */ 00053 00054 void arm_offset_q7( 00055 q7_t * pSrc, 00056 q7_t offset, 00057 q7_t * pDst, 00058 uint32_t blockSize) 00059 { 00060 uint32_t blkCnt; /* loop counter */ 00061 00062 #if defined (ARM_MATH_DSP) 00063 00064 /* Run the below code for Cortex-M4 and Cortex-M3 */ 00065 q31_t offset_packed; /* Offset packed to 32 bit */ 00066 00067 00068 /*loop Unrolling */ 00069 blkCnt = blockSize >> 2U; 00070 00071 /* Offset is packed to 32 bit in order to use SIMD32 for addition */ 00072 offset_packed = __PACKq7(offset, offset, offset, offset); 00073 00074 /* First part of the processing with loop unrolling. Compute 4 outputs at a time. 00075 ** a second loop below computes the remaining 1 to 3 samples. */ 00076 while (blkCnt > 0U) 00077 { 00078 /* C = A + offset */ 00079 /* Add offset and then store the results in the destination bufferfor 4 samples at a time. */ 00080 *__SIMD32(pDst)++ = __QADD8(*__SIMD32(pSrc)++, offset_packed); 00081 00082 /* Decrement the loop counter */ 00083 blkCnt--; 00084 } 00085 00086 /* If the blockSize is not a multiple of 4, compute any remaining output samples here. 00087 ** No loop unrolling is used. */ 00088 blkCnt = blockSize % 0x4U; 00089 00090 while (blkCnt > 0U) 00091 { 00092 /* C = A + offset */ 00093 /* Add offset and then store the result in the destination buffer. */ 00094 *pDst++ = (q7_t) __SSAT(*pSrc++ + offset, 8); 00095 00096 /* Decrement the loop counter */ 00097 blkCnt--; 00098 } 00099 00100 #else 00101 00102 /* Run the below code for Cortex-M0 */ 00103 00104 /* Initialize blkCnt with number of samples */ 00105 blkCnt = blockSize; 00106 00107 while (blkCnt > 0U) 00108 { 00109 /* C = A + offset */ 00110 /* Add offset and then store the result in the destination buffer. */ 00111 *pDst++ = (q7_t) __SSAT((q15_t) * pSrc++ + offset, 8); 00112 00113 /* Decrement the loop counter */ 00114 blkCnt--; 00115 } 00116 00117 #endif /* #if defined (ARM_MATH_DSP) */ 00118 00119 } 00120 00121 /** 00122 * @} end of offset group 00123 */ 00124
Generated on Tue Jul 12 2022 16:47:27 by 1.7.2