diff options
author | Clyne Sullivan <clyne@bitgloo.com> | 2025-01-29 21:34:25 -0500 |
---|---|---|
committer | Clyne Sullivan <clyne@bitgloo.com> | 2025-01-29 21:34:25 -0500 |
commit | 5b81bc8ccbd342b8566d88fc9f17a73aec03b5b6 (patch) | |
tree | cc57486912cfa74c6440d8b97c28f451ec787d78 /Drivers/CMSIS/DSP/Source/DistanceFunctions/arm_euclidean_distance_f32.c |
initial commit
Diffstat (limited to 'Drivers/CMSIS/DSP/Source/DistanceFunctions/arm_euclidean_distance_f32.c')
-rw-r--r-- | Drivers/CMSIS/DSP/Source/DistanceFunctions/arm_euclidean_distance_f32.c | 152 |
1 files changed, 152 insertions, 0 deletions
diff --git a/Drivers/CMSIS/DSP/Source/DistanceFunctions/arm_euclidean_distance_f32.c b/Drivers/CMSIS/DSP/Source/DistanceFunctions/arm_euclidean_distance_f32.c new file mode 100644 index 0000000..e226617 --- /dev/null +++ b/Drivers/CMSIS/DSP/Source/DistanceFunctions/arm_euclidean_distance_f32.c @@ -0,0 +1,152 @@ + +/* ---------------------------------------------------------------------- + * Project: CMSIS DSP Library + * Title: arm_euclidean_distance_f32.c + * Description: Euclidean distance between two vectors + * + * $Date: 23 April 2021 + * $Revision: V1.9.0 + * + * Target Processor: Cortex-M and Cortex-A cores + * -------------------------------------------------------------------- */ +/* + * Copyright (C) 2010-2021 ARM Limited or its affiliates. All rights reserved. + * + * SPDX-License-Identifier: Apache-2.0 + * + * Licensed under the Apache License, Version 2.0 (the License); you may + * not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an AS IS BASIS, WITHOUT + * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include "dsp/distance_functions.h" +#include <limits.h> +#include <math.h> + + + +/** + @addtogroup Euclidean + @{ + */ + + +/** + * @brief Euclidean distance between two vectors + * @param[in] pA First vector + * @param[in] pB Second vector + * @param[in] blockSize vector length + * @return distance + * + */ +#if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) + +#include "arm_helium_utils.h" +#include "arm_vec_math.h" +float32_t arm_euclidean_distance_f32(const float32_t *pA,const float32_t *pB, uint32_t blockSize) +{ + uint32_t blkCnt; + float32_t tmp; + f32x4_t a, b, accumV, tempV; + + accumV = vdupq_n_f32(0.0f); + + blkCnt = blockSize >> 2; + while (blkCnt > 0U) { + a = vld1q(pA); + b = vld1q(pB); + + tempV = vsubq(a, b); + accumV = vfmaq(accumV, tempV, tempV); + + pA += 4; + pB += 4; + blkCnt--; + } + + /* + * tail + * (will be merged thru tail predication) + */ + blkCnt = blockSize & 3; + if (blkCnt > 0U) { + mve_pred16_t p0 = vctp32q(blkCnt); + + a = vldrwq_z_f32(pA, p0); + b = vldrwq_z_f32(pB, p0); + + tempV = vsubq(a, b); + accumV = vfmaq_m(accumV, tempV, tempV, p0); + } + + arm_sqrt_f32(vecAddAcrossF32Mve(accumV), &tmp); + return (tmp); +} +#else +#if defined(ARM_MATH_NEON) + +#include "NEMath.h" + +float32_t arm_euclidean_distance_f32(const float32_t *pA,const float32_t *pB, uint32_t blockSize) +{ + float32_t accum=0.0f,tmp; + uint32_t blkCnt; + float32x4_t a,b,accumV; + float32x2_t accumV2; + + accumV = vdupq_n_f32(0.0f); + blkCnt = blockSize >> 2; + while(blkCnt > 0) + { + a = vld1q_f32(pA); + b = vld1q_f32(pB); + + a = vsubq_f32(a,b); + accumV = vmlaq_f32(accumV,a,a); + pA += 4; + pB += 4; + blkCnt --; + } + accumV2 = vpadd_f32(vget_low_f32(accumV),vget_high_f32(accumV)); + accum = vget_lane_f32(accumV2, 0) + vget_lane_f32(accumV2, 1); + + blkCnt = blockSize & 3; + while(blkCnt > 0) + { + tmp = *pA++ - *pB++; + accum += SQ(tmp); + blkCnt --; + } + arm_sqrt_f32(accum,&tmp); + return(tmp); +} + +#else +float32_t arm_euclidean_distance_f32(const float32_t *pA,const float32_t *pB, uint32_t blockSize) +{ + float32_t accum=0.0f,tmp; + + while(blockSize > 0) + { + tmp = *pA++ - *pB++; + accum += SQ(tmp); + blockSize --; + } + arm_sqrt_f32(accum,&tmp); + return(tmp); +} +#endif +#endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */ + + +/** + * @} end of Euclidean group + */ |