You cannot select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
933 lines
28 KiB
C
933 lines
28 KiB
C
/* ----------------------------------------------------------------------
|
|
* Project: CMSIS DSP Library
|
|
* Title: arm_mat_cmplx_mult_f16.c
|
|
* Description: Floating-point matrix multiplication
|
|
*
|
|
* $Date: 18. March 2020
|
|
* $Revision: V1.6.0
|
|
*
|
|
* Target Processor: Cortex-M cores
|
|
* -------------------------------------------------------------------- */
|
|
/*
|
|
* Copyright (C) 2010-2020 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/matrix_functions_f16.h"
|
|
|
|
#if defined(ARM_FLOAT16_SUPPORTED)
|
|
|
|
|
|
/**
|
|
@ingroup groupMatrix
|
|
*/
|
|
|
|
|
|
/**
|
|
@addtogroup CmplxMatrixMult
|
|
@{
|
|
*/
|
|
|
|
/**
|
|
@brief Floating-point Complex matrix multiplication.
|
|
@param[in] pSrcA points to first input complex matrix structure
|
|
@param[in] pSrcB points to second input complex matrix structure
|
|
@param[out] pDst points to output complex matrix structure
|
|
@return execution status
|
|
- \ref ARM_MATH_SUCCESS : Operation successful
|
|
- \ref ARM_MATH_SIZE_MISMATCH : Matrix size check failed
|
|
*/
|
|
|
|
#if defined(ARM_MATH_MVE_FLOAT16) && !defined(ARM_MATH_AUTOVECTORIZE) && defined(__CMSIS_GCC_H)
|
|
#pragma GCC warning "Scalar version of arm_mat_cmplx_mult_f16 built. Helium version has build issues with gcc."
|
|
#endif
|
|
|
|
#if defined(ARM_MATH_MVE_FLOAT16) && !defined(ARM_MATH_AUTOVECTORIZE) && !defined(__CMSIS_GCC_H)
|
|
|
|
#include "arm_helium_utils.h"
|
|
|
|
#define DONTCARE 0 /* inactive lane content */
|
|
|
|
|
|
__STATIC_FORCEINLINE arm_status arm_mat_cmplx_mult_f16_2x2_mve(
|
|
const arm_matrix_instance_f16 * pSrcA,
|
|
const arm_matrix_instance_f16 * pSrcB,
|
|
arm_matrix_instance_f16 * pDst)
|
|
{
|
|
const uint16_t MATRIX_DIM = 2;
|
|
float16_t const *pInB = pSrcB->pData; /* input data matrix pointer B */
|
|
float16_t *pInA = pSrcA->pData; /* input data matrix pointer A */
|
|
float16_t *pOut = pDst->pData; /* output data matrix pointer */
|
|
uint16x8_t vecColBOffs0,vecColAOffs0,vecColAOffs1;
|
|
float16_t *pInA0 = pInA;
|
|
f16x8_t acc0, acc1;
|
|
f16x8_t vecB, vecA0, vecA1;
|
|
f16x8_t vecTmp;
|
|
uint16_t tmp;
|
|
static const uint16_t offsetB0[8] = { 0, 1,
|
|
MATRIX_DIM * CMPLX_DIM, MATRIX_DIM * CMPLX_DIM + 1,
|
|
2, 3,
|
|
MATRIX_DIM * CMPLX_DIM + 2 , MATRIX_DIM * CMPLX_DIM + 3,
|
|
};
|
|
|
|
|
|
vecColBOffs0 = vldrhq_u16((uint16_t const *) offsetB0);
|
|
|
|
tmp = 0;
|
|
vecColAOffs0 = viwdupq_u16(tmp, 4, 1);
|
|
|
|
tmp = (CMPLX_DIM * MATRIX_DIM);
|
|
vecColAOffs1 = vecColAOffs0 + (uint16_t)(CMPLX_DIM * MATRIX_DIM);
|
|
|
|
|
|
pInB = (float16_t const *)pSrcB->pData;
|
|
|
|
vecA0 = vldrhq_gather_shifted_offset_f16(pInA0, vecColAOffs0);
|
|
vecA1 = vldrhq_gather_shifted_offset_f16(pInA0, vecColAOffs1);
|
|
|
|
|
|
vecB = vldrhq_gather_shifted_offset(pInB, vecColBOffs0);
|
|
|
|
acc0 = vcmulq(vecA0, vecB);
|
|
acc0 = vcmlaq_rot90(acc0, vecA0, vecB);
|
|
|
|
acc1 = vcmulq(vecA1, vecB);
|
|
acc1 = vcmlaq_rot90(acc1, vecA1, vecB);
|
|
|
|
|
|
/*
|
|
* Compute
|
|
* re0+re1 | im0+im1 | re0+re1 | im0+im1
|
|
* re2+re3 | im2+im3 | re2+re3 | im2+im3
|
|
*/
|
|
|
|
vecTmp = (f16x8_t) vrev64q_s32((int32x4_t) acc0);
|
|
vecTmp = vaddq(vecTmp, acc0);
|
|
|
|
|
|
*(float32_t *)(&pOut[0 * CMPLX_DIM * MATRIX_DIM]) = ((f32x4_t)vecTmp)[0];
|
|
*(float32_t *)(&pOut[0 * CMPLX_DIM * MATRIX_DIM + CMPLX_DIM]) = ((f32x4_t)vecTmp)[2];
|
|
|
|
vecTmp = (f16x8_t) vrev64q_s32((int32x4_t) acc1);
|
|
vecTmp = vaddq(vecTmp, acc1);
|
|
|
|
*(float32_t *)(&pOut[1 * CMPLX_DIM * MATRIX_DIM]) = ((f32x4_t)vecTmp)[0];
|
|
*(float32_t *)(&pOut[1 * CMPLX_DIM * MATRIX_DIM + CMPLX_DIM]) = ((f32x4_t)vecTmp)[2];
|
|
|
|
/*
|
|
* Return to application
|
|
*/
|
|
return (ARM_MATH_SUCCESS);
|
|
}
|
|
|
|
|
|
|
|
__STATIC_FORCEINLINE arm_status arm_mat_cmplx_mult_f16_3x3_mve(
|
|
const arm_matrix_instance_f16 * pSrcA,
|
|
const arm_matrix_instance_f16 * pSrcB,
|
|
arm_matrix_instance_f16 * pDst)
|
|
{
|
|
const uint16_t MATRIX_DIM = 3;
|
|
float16_t const *pInB = pSrcB->pData; /* input data matrix pointer B */
|
|
float16_t *pInA = pSrcA->pData; /* input data matrix pointer A */
|
|
float16_t *pOut = pDst->pData; /* output data matrix pointer */
|
|
uint16x8_t vecColBOffs0;
|
|
float16_t *pInA0 = pInA;
|
|
float16_t *pInA1 = pInA0 + CMPLX_DIM * MATRIX_DIM;
|
|
float16_t *pInA2 = pInA1 + CMPLX_DIM * MATRIX_DIM;
|
|
f16x8_t acc0, acc1, acc2;
|
|
f16x8_t vecB, vecA0, vecA1, vecA2;
|
|
static const uint16_t offsetB0[8] = { 0, 1,
|
|
MATRIX_DIM * CMPLX_DIM, MATRIX_DIM * CMPLX_DIM + 1,
|
|
2 * MATRIX_DIM * CMPLX_DIM, 2 * MATRIX_DIM * CMPLX_DIM + 1,
|
|
DONTCARE, DONTCARE
|
|
};
|
|
|
|
|
|
/* enable predication to disable upper half complex vector element */
|
|
mve_pred16_t p0 = vctp16q(MATRIX_DIM * CMPLX_DIM);
|
|
|
|
vecColBOffs0 = vldrhq_u16((uint16_t const *) offsetB0);
|
|
|
|
pInB = (float16_t const *)pSrcB->pData;
|
|
|
|
vecA0 = vldrhq_f16(pInA0);
|
|
vecA1 = vldrhq_f16(pInA1);
|
|
vecA2 = vldrhq_f16(pInA2);
|
|
|
|
vecB = vldrhq_gather_shifted_offset_z(pInB, vecColBOffs0, p0);
|
|
|
|
acc0 = vcmulq(vecA0, vecB);
|
|
acc0 = vcmlaq_rot90(acc0, vecA0, vecB);
|
|
|
|
acc1 = vcmulq(vecA1, vecB);
|
|
acc1 = vcmlaq_rot90(acc1, vecA1, vecB);
|
|
|
|
acc2 = vcmulq(vecA2, vecB);
|
|
acc2 = vcmlaq_rot90(acc2, vecA2, vecB);
|
|
|
|
mve_cmplx_sum_intra_vec_f16(acc0, &pOut[0 * CMPLX_DIM * MATRIX_DIM]);
|
|
mve_cmplx_sum_intra_vec_f16(acc1, &pOut[1 * CMPLX_DIM * MATRIX_DIM]);
|
|
mve_cmplx_sum_intra_vec_f16(acc2, &pOut[2 * CMPLX_DIM * MATRIX_DIM]);
|
|
pOut += CMPLX_DIM;
|
|
/*
|
|
* move to next B column
|
|
*/
|
|
pInB = pInB + CMPLX_DIM;
|
|
|
|
vecB = vldrhq_gather_shifted_offset_z(pInB, vecColBOffs0, p0);
|
|
|
|
acc0 = vcmulq(vecA0, vecB);
|
|
acc0 = vcmlaq_rot90(acc0, vecA0, vecB);
|
|
|
|
acc1 = vcmulq(vecA1, vecB);
|
|
acc1 = vcmlaq_rot90(acc1, vecA1, vecB);
|
|
|
|
acc2 = vcmulq(vecA2, vecB);
|
|
acc2 = vcmlaq_rot90(acc2, vecA2, vecB);
|
|
|
|
mve_cmplx_sum_intra_vec_f16(acc0, &pOut[0 * CMPLX_DIM * MATRIX_DIM]);
|
|
mve_cmplx_sum_intra_vec_f16(acc1, &pOut[1 * CMPLX_DIM * MATRIX_DIM]);
|
|
mve_cmplx_sum_intra_vec_f16(acc2, &pOut[2 * CMPLX_DIM * MATRIX_DIM]);
|
|
pOut += CMPLX_DIM;
|
|
/*
|
|
* move to next B column
|
|
*/
|
|
pInB = pInB + CMPLX_DIM;
|
|
|
|
vecB = vldrhq_gather_shifted_offset_z(pInB, vecColBOffs0, p0);
|
|
|
|
acc0 = vcmulq(vecA0, vecB);
|
|
acc0 = vcmlaq_rot90(acc0, vecA0, vecB);
|
|
|
|
acc1 = vcmulq(vecA1, vecB);
|
|
acc1 = vcmlaq_rot90(acc1, vecA1, vecB);
|
|
|
|
acc2 = vcmulq(vecA2, vecB);
|
|
acc2 = vcmlaq_rot90(acc2, vecA2, vecB);
|
|
|
|
mve_cmplx_sum_intra_vec_f16(acc0, &pOut[0 * CMPLX_DIM * MATRIX_DIM]);
|
|
mve_cmplx_sum_intra_vec_f16(acc1, &pOut[1 * CMPLX_DIM * MATRIX_DIM]);
|
|
mve_cmplx_sum_intra_vec_f16(acc2, &pOut[2 * CMPLX_DIM * MATRIX_DIM]);
|
|
/*
|
|
* Return to application
|
|
*/
|
|
return (ARM_MATH_SUCCESS);
|
|
}
|
|
|
|
|
|
|
|
|
|
__STATIC_FORCEINLINE arm_status arm_mat_cmplx_mult_f16_4x4_mve(
|
|
const arm_matrix_instance_f16 * pSrcA,
|
|
const arm_matrix_instance_f16 * pSrcB,
|
|
arm_matrix_instance_f16 * pDst)
|
|
{
|
|
const uint16_t MATRIX_DIM = 4;
|
|
float16_t const *pInB = pSrcB->pData; /* input data matrix pointer B */
|
|
float16_t *pInA = pSrcA->pData; /* input data matrix pointer A */
|
|
float16_t *pOut = pDst->pData; /* output data matrix pointer */
|
|
uint16x8_t vecColBOffs0;
|
|
float16_t *pInA0 = pInA;
|
|
float16_t *pInA1 = pInA0 + CMPLX_DIM * MATRIX_DIM;
|
|
float16_t *pInA2 = pInA1 + CMPLX_DIM * MATRIX_DIM;
|
|
float16_t *pInA3 = pInA2 + CMPLX_DIM * MATRIX_DIM;
|
|
f16x8_t acc0, acc1, acc2, acc3;
|
|
f16x8_t vecB, vecA;
|
|
static const uint16_t offsetB0[8] = { 0, 1,
|
|
MATRIX_DIM * CMPLX_DIM, MATRIX_DIM * CMPLX_DIM + 1,
|
|
2 * MATRIX_DIM * CMPLX_DIM, 2 * MATRIX_DIM * CMPLX_DIM + 1,
|
|
3 * MATRIX_DIM * CMPLX_DIM, 3 * MATRIX_DIM * CMPLX_DIM + 1
|
|
};
|
|
|
|
vecColBOffs0 = vldrhq_u16((uint16_t const *) offsetB0);
|
|
|
|
pInB = (float16_t const *)pSrcB->pData;
|
|
|
|
vecB = vldrhq_gather_shifted_offset(pInB, vecColBOffs0);
|
|
|
|
vecA = vldrhq_f16(pInA0);
|
|
acc0 = vcmulq(vecA, vecB);
|
|
acc0 = vcmlaq_rot90(acc0, vecA, vecB);
|
|
|
|
vecA = vldrhq_f16(pInA1);
|
|
acc1 = vcmulq(vecA, vecB);
|
|
acc1 = vcmlaq_rot90(acc1, vecA, vecB);
|
|
|
|
vecA = vldrhq_f16(pInA2);
|
|
acc2 = vcmulq(vecA, vecB);
|
|
acc2 = vcmlaq_rot90(acc2, vecA, vecB);
|
|
|
|
vecA = vldrhq_f16(pInA3);
|
|
acc3 = vcmulq(vecA, vecB);
|
|
acc3 = vcmlaq_rot90(acc3, vecA, vecB);
|
|
|
|
|
|
mve_cmplx_sum_intra_vec_f16(acc0, &pOut[0 * CMPLX_DIM * MATRIX_DIM]);
|
|
mve_cmplx_sum_intra_vec_f16(acc1, &pOut[1 * CMPLX_DIM * MATRIX_DIM]);
|
|
mve_cmplx_sum_intra_vec_f16(acc2, &pOut[2 * CMPLX_DIM * MATRIX_DIM]);
|
|
mve_cmplx_sum_intra_vec_f16(acc3, &pOut[3 * CMPLX_DIM * MATRIX_DIM]);
|
|
pOut += CMPLX_DIM;
|
|
/*
|
|
* move to next B column
|
|
*/
|
|
pInB = pInB + CMPLX_DIM;
|
|
|
|
vecB = vldrhq_gather_shifted_offset(pInB, vecColBOffs0);
|
|
|
|
vecA = vldrhq_f16(pInA0);
|
|
acc0 = vcmulq(vecA, vecB);
|
|
acc0 = vcmlaq_rot90(acc0, vecA, vecB);
|
|
|
|
vecA = vldrhq_f16(pInA1);
|
|
acc1 = vcmulq(vecA, vecB);
|
|
acc1 = vcmlaq_rot90(acc1, vecA, vecB);
|
|
|
|
vecA = vldrhq_f16(pInA2);
|
|
acc2 = vcmulq(vecA, vecB);
|
|
acc2 = vcmlaq_rot90(acc2, vecA, vecB);
|
|
|
|
vecA = vldrhq_f16(pInA3);
|
|
acc3 = vcmulq(vecA, vecB);
|
|
acc3 = vcmlaq_rot90(acc3, vecA, vecB);
|
|
|
|
|
|
mve_cmplx_sum_intra_vec_f16(acc0, &pOut[0 * CMPLX_DIM * MATRIX_DIM]);
|
|
mve_cmplx_sum_intra_vec_f16(acc1, &pOut[1 * CMPLX_DIM * MATRIX_DIM]);
|
|
mve_cmplx_sum_intra_vec_f16(acc2, &pOut[2 * CMPLX_DIM * MATRIX_DIM]);
|
|
mve_cmplx_sum_intra_vec_f16(acc3, &pOut[3 * CMPLX_DIM * MATRIX_DIM]);
|
|
pOut += CMPLX_DIM;
|
|
/*
|
|
* move to next B column
|
|
*/
|
|
pInB = pInB + CMPLX_DIM;
|
|
|
|
vecB = vldrhq_gather_shifted_offset(pInB, vecColBOffs0);
|
|
|
|
vecA = vldrhq_f16(pInA0);
|
|
acc0 = vcmulq(vecA, vecB);
|
|
acc0 = vcmlaq_rot90(acc0, vecA, vecB);
|
|
|
|
vecA = vldrhq_f16(pInA1);
|
|
acc1 = vcmulq(vecA, vecB);
|
|
acc1 = vcmlaq_rot90(acc1, vecA, vecB);
|
|
|
|
vecA = vldrhq_f16(pInA2);
|
|
acc2 = vcmulq(vecA, vecB);
|
|
acc2 = vcmlaq_rot90(acc2, vecA, vecB);
|
|
|
|
vecA = vldrhq_f16(pInA3);
|
|
acc3 = vcmulq(vecA, vecB);
|
|
acc3 = vcmlaq_rot90(acc3, vecA, vecB);
|
|
|
|
|
|
mve_cmplx_sum_intra_vec_f16(acc0, &pOut[0 * CMPLX_DIM * MATRIX_DIM]);
|
|
mve_cmplx_sum_intra_vec_f16(acc1, &pOut[1 * CMPLX_DIM * MATRIX_DIM]);
|
|
mve_cmplx_sum_intra_vec_f16(acc2, &pOut[2 * CMPLX_DIM * MATRIX_DIM]);
|
|
mve_cmplx_sum_intra_vec_f16(acc3, &pOut[3 * CMPLX_DIM * MATRIX_DIM]);
|
|
pOut += CMPLX_DIM;
|
|
/*
|
|
* move to next B column
|
|
*/
|
|
pInB = pInB + CMPLX_DIM;
|
|
|
|
vecB = vldrhq_gather_shifted_offset(pInB, vecColBOffs0);
|
|
|
|
vecA = vldrhq_f16(pInA0);
|
|
acc0 = vcmulq(vecA, vecB);
|
|
acc0 = vcmlaq_rot90(acc0, vecA, vecB);
|
|
|
|
vecA = vldrhq_f16(pInA1);
|
|
acc1 = vcmulq(vecA, vecB);
|
|
acc1 = vcmlaq_rot90(acc1, vecA, vecB);
|
|
|
|
vecA = vldrhq_f16(pInA2);
|
|
acc2 = vcmulq(vecA, vecB);
|
|
acc2 = vcmlaq_rot90(acc2, vecA, vecB);
|
|
|
|
vecA = vldrhq_f16(pInA3);
|
|
acc3 = vcmulq(vecA, vecB);
|
|
acc3 = vcmlaq_rot90(acc3, vecA, vecB);
|
|
|
|
|
|
mve_cmplx_sum_intra_vec_f16(acc0, &pOut[0 * CMPLX_DIM * MATRIX_DIM]);
|
|
mve_cmplx_sum_intra_vec_f16(acc1, &pOut[1 * CMPLX_DIM * MATRIX_DIM]);
|
|
mve_cmplx_sum_intra_vec_f16(acc2, &pOut[2 * CMPLX_DIM * MATRIX_DIM]);
|
|
mve_cmplx_sum_intra_vec_f16(acc3, &pOut[3 * CMPLX_DIM * MATRIX_DIM]);
|
|
/*
|
|
* Return to application
|
|
*/
|
|
return (ARM_MATH_SUCCESS);
|
|
}
|
|
|
|
|
|
|
|
arm_status arm_mat_cmplx_mult_f16(
|
|
const arm_matrix_instance_f16 * pSrcA,
|
|
const arm_matrix_instance_f16 * pSrcB,
|
|
arm_matrix_instance_f16 * pDst)
|
|
{
|
|
float16_t const *pInB = (float16_t const *) pSrcB->pData; /* input data matrix pointer B */
|
|
float16_t const *pInA = (float16_t const *) pSrcA->pData; /* input data matrix pointer A */
|
|
float16_t *pOut = pDst->pData; /* output data matrix pointer */
|
|
float16_t *px; /* Temporary output data matrix pointer */
|
|
uint16_t numRowsA = pSrcA->numRows; /* number of rows of input matrix A */
|
|
uint16_t numColsB = pSrcB->numCols; /* number of columns of input matrix B */
|
|
uint16_t numColsA = pSrcA->numCols; /* number of columns of input matrix A */
|
|
uint16_t col, i = 0U, row = numRowsA; /* loop counters */
|
|
arm_status status; /* status of matrix multiplication */
|
|
uint16x8_t vecOffs, vecColBOffs;
|
|
uint32_t blkCnt,rowCnt; /* loop counters */
|
|
|
|
#ifdef ARM_MATH_MATRIX_CHECK
|
|
|
|
/* Check for matrix mismatch condition */
|
|
if ((pSrcA->numCols != pSrcB->numRows) ||
|
|
(pSrcA->numRows != pDst->numRows) ||
|
|
(pSrcB->numCols != pDst->numCols) )
|
|
{
|
|
/* Set status as ARM_MATH_SIZE_MISMATCH */
|
|
status = ARM_MATH_SIZE_MISMATCH;
|
|
}
|
|
else
|
|
|
|
#endif /* #ifdef ARM_MATH_MATRIX_CHECK */
|
|
|
|
{
|
|
|
|
/*
|
|
* small squared matrix specialized routines
|
|
*/
|
|
if (numRowsA == numColsB && numColsB == numColsA)
|
|
{
|
|
if (numRowsA == 1)
|
|
{
|
|
pOut[0] = pInA[0] * pInB[0] - pInA[1] * pInB[1];
|
|
pOut[1] = pInA[0] * pInB[1] + pInA[1] * pInB[0];
|
|
return (ARM_MATH_SUCCESS);
|
|
}
|
|
else if (numRowsA == 2)
|
|
return arm_mat_cmplx_mult_f16_2x2_mve(pSrcA, pSrcB, pDst);
|
|
else if (numRowsA == 3)
|
|
return arm_mat_cmplx_mult_f16_3x3_mve(pSrcA, pSrcB, pDst);
|
|
else if (numRowsA == 4)
|
|
return arm_mat_cmplx_mult_f16_4x4_mve(pSrcA, pSrcB, pDst);
|
|
}
|
|
|
|
vecColBOffs[0] = 0;
|
|
vecColBOffs[1] = 1;
|
|
vecColBOffs[2] = numColsB * CMPLX_DIM;
|
|
vecColBOffs[3] = (numColsB * CMPLX_DIM) + 1;
|
|
vecColBOffs[4] = 2*numColsB * CMPLX_DIM;
|
|
vecColBOffs[5] = 2*(numColsB * CMPLX_DIM) + 1;
|
|
vecColBOffs[6] = 3*numColsB * CMPLX_DIM;
|
|
vecColBOffs[7] = 3*(numColsB * CMPLX_DIM) + 1;
|
|
|
|
/*
|
|
* The following loop performs the dot-product of each row in pSrcA with each column in pSrcB
|
|
*/
|
|
|
|
/*
|
|
* row loop
|
|
*/
|
|
rowCnt = row >> 2;
|
|
while (rowCnt > 0u)
|
|
{
|
|
/*
|
|
* Output pointer is set to starting address of the row being processed
|
|
*/
|
|
px = pOut + i * CMPLX_DIM;
|
|
i = i + 4 * numColsB;
|
|
/*
|
|
* For every row wise process, the column loop counter is to be initiated
|
|
*/
|
|
col = numColsB;
|
|
/*
|
|
* For every row wise process, the pInB pointer is set
|
|
* to the starting address of the pSrcB data
|
|
*/
|
|
pInB = (float16_t const *) pSrcB->pData;
|
|
/*
|
|
* column loop
|
|
*/
|
|
while (col > 0u)
|
|
{
|
|
/*
|
|
* generate 4 columns elements
|
|
*/
|
|
/*
|
|
* Matrix A columns number of MAC operations are to be performed
|
|
*/
|
|
|
|
float16_t const *pSrcA0Vec, *pSrcA1Vec, *pSrcA2Vec, *pSrcA3Vec;
|
|
float16_t const *pInA0 = pInA;
|
|
float16_t const *pInA1 = pInA0 + numColsA * CMPLX_DIM;
|
|
float16_t const *pInA2 = pInA1 + numColsA * CMPLX_DIM;
|
|
float16_t const *pInA3 = pInA2 + numColsA * CMPLX_DIM;
|
|
f16x8_t acc0, acc1, acc2, acc3;
|
|
|
|
acc0 = vdupq_n_f16(0.0f16);
|
|
acc1 = vdupq_n_f16(0.0f16);
|
|
acc2 = vdupq_n_f16(0.0f16);
|
|
acc3 = vdupq_n_f16(0.0f16);
|
|
|
|
pSrcA0Vec = (float16_t const *) pInA0;
|
|
pSrcA1Vec = (float16_t const *) pInA1;
|
|
pSrcA2Vec = (float16_t const *) pInA2;
|
|
pSrcA3Vec = (float16_t const *) pInA3;
|
|
|
|
vecOffs = vecColBOffs;
|
|
|
|
/*
|
|
* process 1 x 4 block output
|
|
*/
|
|
blkCnt = (numColsA * CMPLX_DIM) >> 3;
|
|
while (blkCnt > 0U)
|
|
{
|
|
f16x8_t vecB, vecA;
|
|
|
|
vecB = vldrhq_gather_shifted_offset_f16(pInB, vecOffs);
|
|
/*
|
|
* move Matrix B read offsets, 4 rows down
|
|
*/
|
|
vecOffs = vaddq_n_u16(vecOffs , (uint16_t) (numColsB * 4 * CMPLX_DIM));
|
|
|
|
vecA = vld1q(pSrcA0Vec); pSrcA0Vec += 8;
|
|
acc0 = vcmlaq(acc0, vecA, vecB);
|
|
acc0 = vcmlaq_rot90(acc0, vecA, vecB);
|
|
|
|
vecA = vld1q(pSrcA1Vec); pSrcA1Vec += 8;
|
|
acc1 = vcmlaq(acc1, vecA, vecB);
|
|
acc1 = vcmlaq_rot90(acc1, vecA, vecB);
|
|
|
|
vecA = vld1q(pSrcA2Vec); pSrcA2Vec += 8;
|
|
acc2 = vcmlaq(acc2, vecA, vecB);
|
|
acc2 = vcmlaq_rot90(acc2, vecA, vecB);
|
|
|
|
vecA = vld1q(pSrcA3Vec); pSrcA3Vec += 8;
|
|
acc3 = vcmlaq(acc3, vecA, vecB);
|
|
acc3 = vcmlaq_rot90(acc3, vecA, vecB);
|
|
|
|
blkCnt--;
|
|
}
|
|
/*
|
|
* Unsupported addressing mode compiler crash
|
|
*/
|
|
/*
|
|
* tail
|
|
* (will be merged thru tail predication)
|
|
*/
|
|
blkCnt = (numColsA * CMPLX_DIM) & 7;
|
|
if (blkCnt > 0U)
|
|
{
|
|
mve_pred16_t p0 = vctp16q(blkCnt);
|
|
f16x8_t vecB, vecA;
|
|
|
|
vecB = vldrhq_gather_shifted_offset_z_f16(pInB, vecOffs, p0);
|
|
/*
|
|
* move Matrix B read offsets, 4 rows down
|
|
*/
|
|
vecOffs = vaddq_n_u16(vecOffs, (uint16_t) (numColsB * 4 * CMPLX_DIM));
|
|
|
|
vecA = vld1q(pSrcA0Vec);
|
|
acc0 = vcmlaq(acc0, vecA, vecB);
|
|
acc0 = vcmlaq_rot90(acc0, vecA, vecB);
|
|
|
|
vecA = vld1q(pSrcA1Vec);
|
|
acc1 = vcmlaq(acc1, vecA, vecB);
|
|
acc1 = vcmlaq_rot90(acc1, vecA, vecB);
|
|
|
|
vecA = vld1q(pSrcA2Vec);
|
|
acc2 = vcmlaq(acc2, vecA, vecB);
|
|
acc2 = vcmlaq_rot90(acc2, vecA, vecB);
|
|
|
|
vecA = vld1q(pSrcA3Vec);
|
|
acc3 = vcmlaq(acc3, vecA, vecB);
|
|
acc3 = vcmlaq_rot90(acc3, vecA, vecB);
|
|
|
|
}
|
|
|
|
|
|
mve_cmplx_sum_intra_vec_f16(acc0, &px[0 * CMPLX_DIM * numColsB + 0]);
|
|
mve_cmplx_sum_intra_vec_f16(acc1, &px[1 * CMPLX_DIM * numColsB + 0]);
|
|
mve_cmplx_sum_intra_vec_f16(acc2, &px[2 * CMPLX_DIM * numColsB + 0]);
|
|
mve_cmplx_sum_intra_vec_f16(acc3, &px[3 * CMPLX_DIM * numColsB + 0]);
|
|
|
|
px += CMPLX_DIM;
|
|
/*
|
|
* Decrement the column loop counter
|
|
*/
|
|
col--;
|
|
/*
|
|
* Update the pointer pInB to point to the starting address of the next column
|
|
*/
|
|
pInB = (float16_t const *) pSrcB->pData + (numColsB - col) * CMPLX_DIM;
|
|
}
|
|
|
|
/*
|
|
* Update the pointer pInA to point to the starting address of the next row
|
|
*/
|
|
pInA += (numColsA * 4) * CMPLX_DIM;
|
|
/*
|
|
* Decrement the row loop counter
|
|
*/
|
|
rowCnt --;
|
|
|
|
}
|
|
|
|
rowCnt = row & 3;
|
|
while (rowCnt > 0u)
|
|
{
|
|
/*
|
|
* Output pointer is set to starting address of the row being processed
|
|
*/
|
|
px = pOut + i * CMPLX_DIM;
|
|
i = i + numColsB;
|
|
/*
|
|
* For every row wise process, the column loop counter is to be initiated
|
|
*/
|
|
col = numColsB;
|
|
/*
|
|
* For every row wise process, the pInB pointer is set
|
|
* to the starting address of the pSrcB data
|
|
*/
|
|
pInB = (float16_t const *) pSrcB->pData;
|
|
/*
|
|
* column loop
|
|
*/
|
|
while (col > 0u)
|
|
{
|
|
/*
|
|
* generate 4 columns elements
|
|
*/
|
|
/*
|
|
* Matrix A columns number of MAC operations are to be performed
|
|
*/
|
|
|
|
float16_t const *pSrcA0Vec;
|
|
float16_t const *pInA0 = pInA;
|
|
f16x8_t acc0;
|
|
|
|
acc0 = vdupq_n_f16(0.0f16);
|
|
|
|
pSrcA0Vec = (float16_t const *) pInA0;
|
|
|
|
vecOffs = vecColBOffs;
|
|
|
|
/*
|
|
* process 1 x 4 block output
|
|
*/
|
|
blkCnt = (numColsA * CMPLX_DIM) >> 3;
|
|
while (blkCnt > 0U)
|
|
{
|
|
f16x8_t vecB, vecA;
|
|
|
|
vecB = vldrhq_gather_shifted_offset(pInB, vecOffs);
|
|
/*
|
|
* move Matrix B read offsets, 4 rows down
|
|
*/
|
|
vecOffs = vaddq_n_u16(vecOffs, (uint16_t) (4*numColsB * CMPLX_DIM));
|
|
|
|
vecA = vld1q(pSrcA0Vec);
|
|
pSrcA0Vec += 8;
|
|
acc0 = vcmlaq(acc0, vecA, vecB);
|
|
acc0 = vcmlaq_rot90(acc0, vecA, vecB);
|
|
|
|
|
|
blkCnt--;
|
|
}
|
|
|
|
|
|
/*
|
|
* tail
|
|
*/
|
|
blkCnt = (numColsA * CMPLX_DIM) & 7;
|
|
if (blkCnt > 0U)
|
|
{
|
|
mve_pred16_t p0 = vctp16q(blkCnt);
|
|
f16x8_t vecB, vecA;
|
|
|
|
vecB = vldrhq_gather_shifted_offset_z(pInB, vecOffs, p0);
|
|
|
|
vecA = vld1q(pSrcA0Vec);
|
|
acc0 = vcmlaq(acc0, vecA, vecB);
|
|
acc0 = vcmlaq_rot90(acc0, vecA, vecB);
|
|
|
|
}
|
|
|
|
mve_cmplx_sum_intra_vec_f16(acc0, &px[0]);
|
|
|
|
|
|
px += CMPLX_DIM;
|
|
/*
|
|
* Decrement the column loop counter
|
|
*/
|
|
col--;
|
|
/*
|
|
* Update the pointer pInB to point to the starting address of the next column
|
|
*/
|
|
pInB = (float16_t const *) pSrcB->pData + (numColsB - col) * CMPLX_DIM;
|
|
}
|
|
|
|
/*
|
|
* Update the pointer pInA to point to the starting address of the next row
|
|
*/
|
|
pInA += numColsA * CMPLX_DIM;
|
|
rowCnt--;
|
|
}
|
|
|
|
/*
|
|
* set status as ARM_MATH_SUCCESS
|
|
*/
|
|
status = ARM_MATH_SUCCESS;
|
|
}
|
|
/*
|
|
* Return to application
|
|
*/
|
|
return (status);
|
|
}
|
|
#else
|
|
|
|
arm_status arm_mat_cmplx_mult_f16(
|
|
const arm_matrix_instance_f16 * pSrcA,
|
|
const arm_matrix_instance_f16 * pSrcB,
|
|
arm_matrix_instance_f16 * pDst)
|
|
{
|
|
float16_t *pIn1 = pSrcA->pData; /* Input data matrix pointer A */
|
|
float16_t *pIn2 = pSrcB->pData; /* Input data matrix pointer B */
|
|
float16_t *pInA = pSrcA->pData; /* Input data matrix pointer A */
|
|
float16_t *pOut = pDst->pData; /* Output data matrix pointer */
|
|
float16_t *px; /* Temporary output data matrix pointer */
|
|
uint16_t numRowsA = pSrcA->numRows; /* Number of rows of input matrix A */
|
|
uint16_t numColsB = pSrcB->numCols; /* Number of columns of input matrix B */
|
|
uint16_t numColsA = pSrcA->numCols; /* Number of columns of input matrix A */
|
|
_Float16 sumReal, sumImag; /* Accumulator */
|
|
_Float16 a1, b1, c1, d1;
|
|
uint32_t col, i = 0U, j, row = numRowsA, colCnt; /* loop counters */
|
|
arm_status status; /* status of matrix multiplication */
|
|
|
|
#if defined (ARM_MATH_LOOPUNROLL)
|
|
_Float16 a0, b0, c0, d0;
|
|
#endif
|
|
|
|
#ifdef ARM_MATH_MATRIX_CHECK
|
|
|
|
/* Check for matrix mismatch condition */
|
|
if ((pSrcA->numCols != pSrcB->numRows) ||
|
|
(pSrcA->numRows != pDst->numRows) ||
|
|
(pSrcB->numCols != pDst->numCols) )
|
|
{
|
|
/* Set status as ARM_MATH_SIZE_MISMATCH */
|
|
status = ARM_MATH_SIZE_MISMATCH;
|
|
}
|
|
else
|
|
|
|
#endif /* #ifdef ARM_MATH_MATRIX_CHECK */
|
|
|
|
{
|
|
/* The following loop performs the dot-product of each row in pSrcA with each column in pSrcB */
|
|
/* row loop */
|
|
do
|
|
{
|
|
/* Output pointer is set to starting address of the row being processed */
|
|
px = pOut + 2 * i;
|
|
|
|
/* For every row wise process, the column loop counter is to be initiated */
|
|
col = numColsB;
|
|
|
|
/* For every row wise process, the pIn2 pointer is set
|
|
** to the starting address of the pSrcB data */
|
|
pIn2 = pSrcB->pData;
|
|
|
|
j = 0U;
|
|
|
|
/* column loop */
|
|
do
|
|
{
|
|
/* Set the variable sum, that acts as accumulator, to zero */
|
|
sumReal = 0.0f16;
|
|
sumImag = 0.0f16;
|
|
|
|
/* Initiate pointer pIn1 to point to starting address of column being processed */
|
|
pIn1 = pInA;
|
|
|
|
#if defined (ARM_MATH_LOOPUNROLL)
|
|
|
|
/* Apply loop unrolling and compute 4 MACs simultaneously. */
|
|
colCnt = numColsA >> 2U;
|
|
|
|
/* matrix multiplication */
|
|
while (colCnt > 0U)
|
|
{
|
|
|
|
/* Reading real part of complex matrix A */
|
|
a0 = *pIn1;
|
|
|
|
/* Reading real part of complex matrix B */
|
|
c0 = *pIn2;
|
|
|
|
/* Reading imaginary part of complex matrix A */
|
|
b0 = *(pIn1 + 1U);
|
|
|
|
/* Reading imaginary part of complex matrix B */
|
|
d0 = *(pIn2 + 1U);
|
|
|
|
/* Multiply and Accumlates */
|
|
sumReal += a0 * c0;
|
|
sumImag += b0 * c0;
|
|
|
|
/* update pointers */
|
|
pIn1 += 2U;
|
|
pIn2 += 2 * numColsB;
|
|
|
|
/* Multiply and Accumlates */
|
|
sumReal -= b0 * d0;
|
|
sumImag += a0 * d0;
|
|
|
|
/* c(m,n) = a(1,1) * b(1,1) + a(1,2) * b(2,1) + .... + a(m,p) * b(p,n) */
|
|
|
|
/* read real and imag values from pSrcA and pSrcB buffer */
|
|
a1 = *(pIn1 );
|
|
c1 = *(pIn2 );
|
|
b1 = *(pIn1 + 1U);
|
|
d1 = *(pIn2 + 1U);
|
|
|
|
/* Multiply and Accumlates */
|
|
sumReal += a1 * c1;
|
|
sumImag += b1 * c1;
|
|
|
|
/* update pointers */
|
|
pIn1 += 2U;
|
|
pIn2 += 2 * numColsB;
|
|
|
|
/* Multiply and Accumlates */
|
|
sumReal -= b1 * d1;
|
|
sumImag += a1 * d1;
|
|
|
|
a0 = *(pIn1 );
|
|
c0 = *(pIn2 );
|
|
b0 = *(pIn1 + 1U);
|
|
d0 = *(pIn2 + 1U);
|
|
|
|
/* Multiply and Accumlates */
|
|
sumReal += a0 * c0;
|
|
sumImag += b0 * c0;
|
|
|
|
/* update pointers */
|
|
pIn1 += 2U;
|
|
pIn2 += 2 * numColsB;
|
|
|
|
/* Multiply and Accumlates */
|
|
sumReal -= b0 * d0;
|
|
sumImag += a0 * d0;
|
|
|
|
/* c(m,n) = a(1,1) * b(1,1) + a(1,2) * b(2,1) + .... + a(m,p) * b(p,n) */
|
|
|
|
a1 = *(pIn1 );
|
|
c1 = *(pIn2 );
|
|
b1 = *(pIn1 + 1U);
|
|
d1 = *(pIn2 + 1U);
|
|
|
|
/* Multiply and Accumlates */
|
|
sumReal += a1 * c1;
|
|
sumImag += b1 * c1;
|
|
|
|
/* update pointers */
|
|
pIn1 += 2U;
|
|
pIn2 += 2 * numColsB;
|
|
|
|
/* Multiply and Accumlates */
|
|
sumReal -= b1 * d1;
|
|
sumImag += a1 * d1;
|
|
|
|
/* Decrement loop count */
|
|
colCnt--;
|
|
}
|
|
|
|
/* If the columns of pSrcA is not a multiple of 4, compute any remaining MACs here.
|
|
** No loop unrolling is used. */
|
|
colCnt = numColsA % 0x4U;
|
|
|
|
#else
|
|
|
|
/* Initialize blkCnt with number of samples */
|
|
colCnt = numColsA;
|
|
|
|
#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
|
|
|
|
while (colCnt > 0U)
|
|
{
|
|
/* c(m,n) = a(1,1) * b(1,1) + a(1,2) * b(2,1) + .... + a(m,p) * b(p,n) */
|
|
a1 = *(pIn1 );
|
|
c1 = *(pIn2 );
|
|
b1 = *(pIn1 + 1U);
|
|
d1 = *(pIn2 + 1U);
|
|
|
|
/* Multiply and Accumlates */
|
|
sumReal += a1 * c1;
|
|
sumImag += b1 * c1;
|
|
|
|
/* update pointers */
|
|
pIn1 += 2U;
|
|
pIn2 += 2 * numColsB;
|
|
|
|
/* Multiply and Accumlates */
|
|
sumReal -= b1 * d1;
|
|
sumImag += a1 * d1;
|
|
|
|
/* Decrement loop counter */
|
|
colCnt--;
|
|
}
|
|
|
|
/* Store result in destination buffer */
|
|
*px++ = sumReal;
|
|
*px++ = sumImag;
|
|
|
|
/* Update pointer pIn2 to point to starting address of next column */
|
|
j++;
|
|
pIn2 = pSrcB->pData + 2U * j;
|
|
|
|
/* Decrement column loop counter */
|
|
col--;
|
|
|
|
} while (col > 0U);
|
|
|
|
/* Update pointer pInA to point to starting address of next row */
|
|
i = i + numColsB;
|
|
pInA = pInA + 2 * numColsA;
|
|
|
|
/* Decrement row loop counter */
|
|
row--;
|
|
|
|
} while (row > 0U);
|
|
|
|
/* Set status as ARM_MATH_SUCCESS */
|
|
status = ARM_MATH_SUCCESS;
|
|
}
|
|
|
|
/* Return to application */
|
|
return (status);
|
|
}
|
|
|
|
#endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
|
|
|
|
/**
|
|
@} end of MatrixMult group
|
|
*/
|
|
|
|
#endif /* #if defined(ARM_FLOAT16_SUPPORTED) */
|
|
|