275 lines
6.1 KiB
C
275 lines
6.1 KiB
C
/* ----------------------------------------------------------------------
|
|
* Project: CMSIS DSP Library
|
|
* Title: arm_barycenter_f16.c
|
|
* Description: Barycenter
|
|
*
|
|
* $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/support_functions_f16.h"
|
|
|
|
#if defined(ARM_FLOAT16_SUPPORTED)
|
|
|
|
#include <limits.h>
|
|
#include <math.h>
|
|
|
|
/**
|
|
@ingroup groupSupport
|
|
*/
|
|
|
|
/**
|
|
@defgroup barycenter Barycenter
|
|
|
|
Barycenter of weighted vectors
|
|
*/
|
|
|
|
/**
|
|
@addtogroup barycenter
|
|
@{
|
|
*/
|
|
|
|
|
|
/**
|
|
* @brief Barycenter
|
|
*
|
|
*
|
|
* @param[in] *in List of vectors
|
|
* @param[in] *weights Weights of the vectors
|
|
* @param[out] *out Barycenter
|
|
* @param[in] nbVectors Number of vectors
|
|
* @param[in] vecDim Dimension of space (vector dimension)
|
|
* @return None
|
|
*
|
|
*/
|
|
|
|
#if defined(ARM_MATH_MVE_FLOAT16) && !defined(ARM_MATH_AUTOVECTORIZE)
|
|
|
|
void arm_barycenter_f16(const float16_t *in,
|
|
const float16_t *weights,
|
|
float16_t *out,
|
|
uint32_t nbVectors,
|
|
uint32_t vecDim)
|
|
{
|
|
const float16_t *pIn, *pW;
|
|
const float16_t *pIn1, *pIn2, *pIn3, *pIn4;
|
|
float16_t *pOut;
|
|
uint32_t blkCntVector, blkCntSample;
|
|
float16_t accum, w;
|
|
|
|
blkCntVector = nbVectors;
|
|
blkCntSample = vecDim;
|
|
|
|
accum = 0.0f;
|
|
|
|
pW = weights;
|
|
pIn = in;
|
|
|
|
|
|
arm_fill_f16(0.0f, out, vecDim);
|
|
|
|
|
|
/* Sum */
|
|
pIn1 = pIn;
|
|
pIn2 = pIn1 + vecDim;
|
|
pIn3 = pIn2 + vecDim;
|
|
pIn4 = pIn3 + vecDim;
|
|
|
|
blkCntVector = nbVectors >> 2;
|
|
while (blkCntVector > 0)
|
|
{
|
|
f16x8_t outV, inV1, inV2, inV3, inV4;
|
|
float16_t w1, w2, w3, w4;
|
|
|
|
pOut = out;
|
|
w1 = *pW++;
|
|
w2 = *pW++;
|
|
w3 = *pW++;
|
|
w4 = *pW++;
|
|
accum += (_Float16)w1 + (_Float16)w2 + (_Float16)w3 + (_Float16)w4;
|
|
|
|
blkCntSample = vecDim >> 3;
|
|
while (blkCntSample > 0) {
|
|
outV = vld1q((const float16_t *) pOut);
|
|
inV1 = vld1q(pIn1);
|
|
inV2 = vld1q(pIn2);
|
|
inV3 = vld1q(pIn3);
|
|
inV4 = vld1q(pIn4);
|
|
outV = vfmaq(outV, inV1, w1);
|
|
outV = vfmaq(outV, inV2, w2);
|
|
outV = vfmaq(outV, inV3, w3);
|
|
outV = vfmaq(outV, inV4, w4);
|
|
vst1q(pOut, outV);
|
|
|
|
pOut += 8;
|
|
pIn1 += 8;
|
|
pIn2 += 8;
|
|
pIn3 += 8;
|
|
pIn4 += 8;
|
|
|
|
blkCntSample--;
|
|
}
|
|
|
|
blkCntSample = vecDim & 7;
|
|
while (blkCntSample > 0) {
|
|
*pOut = (_Float16)*pOut + (_Float16)*pIn1++ * (_Float16)w1;
|
|
*pOut = (_Float16)*pOut + (_Float16)*pIn2++ * (_Float16)w2;
|
|
*pOut = (_Float16)*pOut + (_Float16)*pIn3++ * (_Float16)w3;
|
|
*pOut = (_Float16)*pOut + (_Float16)*pIn4++ * (_Float16)w4;
|
|
pOut++;
|
|
blkCntSample--;
|
|
}
|
|
|
|
pIn1 += 3 * vecDim;
|
|
pIn2 += 3 * vecDim;
|
|
pIn3 += 3 * vecDim;
|
|
pIn4 += 3 * vecDim;
|
|
|
|
blkCntVector--;
|
|
}
|
|
|
|
pIn = pIn1;
|
|
|
|
blkCntVector = nbVectors & 3;
|
|
while (blkCntVector > 0)
|
|
{
|
|
f16x8_t inV, outV;
|
|
|
|
pOut = out;
|
|
w = *pW++;
|
|
accum += (_Float16)w;
|
|
|
|
blkCntSample = vecDim >> 3;
|
|
while (blkCntSample > 0)
|
|
{
|
|
outV = vld1q_f16(pOut);
|
|
inV = vld1q_f16(pIn);
|
|
outV = vfmaq(outV, inV, w);
|
|
vst1q_f16(pOut, outV);
|
|
pOut += 8;
|
|
pIn += 8;
|
|
|
|
blkCntSample--;
|
|
}
|
|
|
|
blkCntSample = vecDim & 7;
|
|
while (blkCntSample > 0)
|
|
{
|
|
*pOut = (_Float16)*pOut + (_Float16)*pIn++ * (_Float16)w;
|
|
pOut++;
|
|
blkCntSample--;
|
|
}
|
|
|
|
blkCntVector--;
|
|
}
|
|
|
|
/* Normalize */
|
|
pOut = out;
|
|
accum = 1.0f16 / (_Float16)accum;
|
|
|
|
blkCntSample = vecDim >> 3;
|
|
while (blkCntSample > 0)
|
|
{
|
|
f16x8_t tmp;
|
|
|
|
tmp = vld1q((const float16_t *) pOut);
|
|
tmp = vmulq(tmp, accum);
|
|
vst1q(pOut, tmp);
|
|
pOut += 8;
|
|
blkCntSample--;
|
|
}
|
|
|
|
blkCntSample = vecDim & 7;
|
|
while (blkCntSample > 0)
|
|
{
|
|
*pOut = (_Float16)*pOut * (_Float16)accum;
|
|
pOut++;
|
|
blkCntSample--;
|
|
}
|
|
}
|
|
#else
|
|
void arm_barycenter_f16(const float16_t *in, const float16_t *weights, float16_t *out, uint32_t nbVectors,uint32_t vecDim)
|
|
{
|
|
|
|
const float16_t *pIn,*pW;
|
|
float16_t *pOut;
|
|
uint32_t blkCntVector,blkCntSample;
|
|
float16_t accum, w;
|
|
|
|
blkCntVector = nbVectors;
|
|
blkCntSample = vecDim;
|
|
|
|
accum = 0.0f16;
|
|
|
|
pW = weights;
|
|
pIn = in;
|
|
|
|
/* Set counters to 0 */
|
|
blkCntSample = vecDim;
|
|
pOut = out;
|
|
|
|
while(blkCntSample > 0)
|
|
{
|
|
*pOut = 0.0f16;
|
|
pOut++;
|
|
blkCntSample--;
|
|
}
|
|
|
|
/* Sum */
|
|
while(blkCntVector > 0)
|
|
{
|
|
pOut = out;
|
|
w = *pW++;
|
|
accum += (_Float16)w;
|
|
|
|
blkCntSample = vecDim;
|
|
while(blkCntSample > 0)
|
|
{
|
|
*pOut = (_Float16)*pOut + (_Float16)*pIn++ * (_Float16)w;
|
|
pOut++;
|
|
blkCntSample--;
|
|
}
|
|
|
|
blkCntVector--;
|
|
}
|
|
|
|
/* Normalize */
|
|
blkCntSample = vecDim;
|
|
pOut = out;
|
|
|
|
while(blkCntSample > 0)
|
|
{
|
|
*pOut = (_Float16)*pOut / (_Float16)accum;
|
|
pOut++;
|
|
blkCntSample--;
|
|
}
|
|
|
|
}
|
|
#endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
|
|
|
|
/**
|
|
* @} end of barycenter group
|
|
*/
|
|
|
|
#endif /* #if defined(ARM_FLOAT16_SUPPORTED) */
|
|
|