From 96d6da4e252b06dcfdc041e7df23e86161c33007 Mon Sep 17 00:00:00 2001 From: rihab kouki Date: Tue, 28 Jul 2020 11:24:49 +0100 Subject: Official ARM version: v5.6.0 --- DSP/Source/BasicMathFunctions/arm_dot_prod_f32.c | 130 +++++++++++++++-------- 1 file changed, 85 insertions(+), 45 deletions(-) (limited to 'DSP/Source/BasicMathFunctions/arm_dot_prod_f32.c') diff --git a/DSP/Source/BasicMathFunctions/arm_dot_prod_f32.c b/DSP/Source/BasicMathFunctions/arm_dot_prod_f32.c index 0cd0afc..3eee3b9 100644 --- a/DSP/Source/BasicMathFunctions/arm_dot_prod_f32.c +++ b/DSP/Source/BasicMathFunctions/arm_dot_prod_f32.c @@ -3,13 +3,13 @@ * Title: arm_dot_prod_f32.c * Description: Floating-point dot product * - * $Date: 27. January 2017 - * $Revision: V.1.5.1 + * $Date: 18. March 2019 + * $Revision: V1.6.0 * * Target Processor: Cortex-M cores * -------------------------------------------------------------------- */ /* - * Copyright (C) 2010-2017 ARM Limited or its affiliates. All rights reserved. + * Copyright (C) 2010-2019 ARM Limited or its affiliates. All rights reserved. * * SPDX-License-Identifier: Apache-2.0 * @@ -29,95 +29,135 @@ #include "arm_math.h" /** - * @ingroup groupMath + @ingroup groupMath */ /** - * @defgroup dot_prod Vector Dot Product - * - * Computes the dot product of two vectors. - * The vectors are multiplied element-by-element and then summed. - * - * 
- *     sum = pSrcA[0]*pSrcB[0] + pSrcA[1]*pSrcB[1] + ... + pSrcA[blockSize-1]*pSrcB[blockSize-1]
- *
- * - * There are separate functions for floating-point, Q7, Q15, and Q31 data types. + @defgroup BasicDotProd Vector Dot Product + + Computes the dot product of two vectors. + The vectors are multiplied element-by-element and then summed. + + 
+      sum = pSrcA[0]*pSrcB[0] + pSrcA[1]*pSrcB[1] + ... + pSrcA[blockSize-1]*pSrcB[blockSize-1]
+
+ + There are separate functions for floating-point, Q7, Q15, and Q31 data types. */ /** - * @addtogroup dot_prod - * @{ + @addtogroup BasicDotProd + @{ */ /** - * @brief Dot product of floating-point vectors. - * @param[in] *pSrcA points to the first input vector - * @param[in] *pSrcB points to the second input vector - * @param[in] blockSize number of samples in each vector - * @param[out] *result output result returned here - * @return none. + @brief Dot product of floating-point vectors. + @param[in] pSrcA points to the first input vector. + @param[in] pSrcB points to the second input vector. + @param[in] blockSize number of samples in each vector. + @param[out] result output result returned here. + @return none */ - void arm_dot_prod_f32( - float32_t * pSrcA, - float32_t * pSrcB, - uint32_t blockSize, - float32_t * result) + const float32_t * pSrcA, + const float32_t * pSrcB, + uint32_t blockSize, + float32_t * result) { - float32_t sum = 0.0f; /* Temporary result storage */ - uint32_t blkCnt; /* loop counter */ + uint32_t blkCnt; /* Loop counter */ + float32_t sum = 0.0f; /* Temporary return variable */ + +#if defined(ARM_MATH_NEON) + float32x4_t vec1; + float32x4_t vec2; + float32x4_t res; + float32x4_t accum = vdupq_n_f32(0); + + /* Compute 4 outputs at a time */ + blkCnt = blockSize >> 2U; + + vec1 = vld1q_f32(pSrcA); + vec2 = vld1q_f32(pSrcB); + + while (blkCnt > 0U) + { + /* C = A[0]*B[0] + A[1]*B[1] + A[2]*B[2] + ... + A[blockSize-1]*B[blockSize-1] */ + /* Calculate dot product and then store the result in a temporary buffer. */ + + accum = vmlaq_f32(accum, vec1, vec2); + + /* Increment pointers */ + pSrcA += 4; + pSrcB += 4; + + vec1 = vld1q_f32(pSrcA); + vec2 = vld1q_f32(pSrcB); + + /* Decrement the loop counter */ + blkCnt--; + } + +#if __aarch64__ + sum = vpadds_f32(vpadd_f32(vget_low_f32(accum), vget_high_f32(accum))); +#else + sum = (vpadd_f32(vget_low_f32(accum), vget_high_f32(accum)))[0] + (vpadd_f32(vget_low_f32(accum), vget_high_f32(accum)))[1]; +#endif + /* Tail */ + blkCnt = blockSize & 0x3; -#if defined (ARM_MATH_DSP) +#else +#if defined (ARM_MATH_LOOPUNROLL) -/* Run the below code for Cortex-M4 and Cortex-M3 */ - /*loop Unrolling */ + /* Loop unrolling: Compute 4 outputs at a time */ blkCnt = blockSize >> 2U; - /* First part of the processing with loop unrolling. Compute 4 outputs at a time. + /* First part of the processing with loop unrolling. Compute 4 outputs at a time. ** a second loop below computes the remaining 1 to 3 samples. */ while (blkCnt > 0U) { /* C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1] */ - /* Calculate dot product and then store the result in a temporary buffer */ + + /* Calculate dot product and store result in a temporary buffer. */ sum += (*pSrcA++) * (*pSrcB++); + sum += (*pSrcA++) * (*pSrcB++); + sum += (*pSrcA++) * (*pSrcB++); + sum += (*pSrcA++) * (*pSrcB++); - /* Decrement the loop counter */ + /* Decrement loop counter */ blkCnt--; } - /* If the blockSize is not a multiple of 4, compute any remaining output samples here. - ** No loop unrolling is used. */ + /* Loop unrolling: Compute remaining outputs */ blkCnt = blockSize % 0x4U; #else - /* Run the below code for Cortex-M0 */ - /* Initialize blkCnt with number of samples */ blkCnt = blockSize; -#endif /* #if defined (ARM_MATH_DSP) */ - +#endif /* #if defined (ARM_MATH_LOOPUNROLL) */ +#endif /* #if defined(ARM_MATH_NEON) */ while (blkCnt > 0U) { /* C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1] */ - /* Calculate dot product and then store the result in a temporary buffer. */ + + /* Calculate dot product and store result in a temporary buffer. */ sum += (*pSrcA++) * (*pSrcB++); - /* Decrement the loop counter */ + /* Decrement loop counter */ blkCnt--; } - /* Store the result back in the destination buffer */ + + /* Store result in destination buffer */ *result = sum; } /** - * @} end of dot_prod group + @} end of BasicDotProd group */ -- cgit