mirror of
https://github.com/tesseract-ocr/tesseract.git
synced 2025-01-06 09:17:49 +08:00
58 lines
1.9 KiB
C++
58 lines
1.9 KiB
C++
///////////////////////////////////////////////////////////////////////
|
|
// File: dotproductfma.cpp
|
|
// Description: Architecture-specific dot-product function.
|
|
// Author: Stefan Weil
|
|
//
|
|
// (C) Copyright 2015, Google Inc.
|
|
// 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
|
|
// http://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.
|
|
///////////////////////////////////////////////////////////////////////
|
|
|
|
#if !defined(__FMA__)
|
|
#error Implementation only for FMA capable architectures
|
|
#endif
|
|
|
|
#include <immintrin.h>
|
|
#include <cstdint>
|
|
#include "dotproduct.h"
|
|
|
|
namespace tesseract {
|
|
|
|
// Computes and returns the dot product of the n-vectors u and v.
|
|
// Uses Intel FMA intrinsics to access the SIMD instruction set.
|
|
double DotProductFMA(const double* u, const double* v, int n) {
|
|
const unsigned quot = n / 8;
|
|
const unsigned rem = n % 8;
|
|
__m256d t0 = _mm256_setzero_pd();
|
|
__m256d t1 = _mm256_setzero_pd();
|
|
for (unsigned k = 0; k < quot; k++) {
|
|
__m256d f0 = _mm256_loadu_pd(u);
|
|
__m256d f1 = _mm256_loadu_pd(v);
|
|
t0 = _mm256_fmadd_pd(f0, f1, t0);
|
|
u += 4;
|
|
v += 4;
|
|
__m256d f2 = _mm256_loadu_pd(u);
|
|
__m256d f3 = _mm256_loadu_pd(v);
|
|
t1 = _mm256_fmadd_pd(f2, f3, t1);
|
|
u += 4;
|
|
v += 4;
|
|
}
|
|
t0 = _mm256_hadd_pd(t0, t1);
|
|
alignas(32) double tmp[4];
|
|
_mm256_store_pd(tmp, t0);
|
|
double result = tmp[0] + tmp[1] + tmp[2] + tmp[3];
|
|
for (unsigned k = 0; k < rem; k++) {
|
|
result += *u++ * *v++;
|
|
}
|
|
return result;
|
|
}
|
|
|
|
} // namespace tesseract.
|