mirror/tesseract
1
0
Fork 0
mirror of https://github.com/tesseract-ocr/tesseract.git synced 2024-12-04 09:49:16 +08:00
Code Issues Actions 11 Packages Projects Releases Wiki Activity

avx2: Remove unused local variables

Browse Source 
Signed-off-by: Stefan Weil <sw@weil.de>
This commit is contained in:
Stefan Weil 2020-12-30 11:19:10 +01:00
parent 1305ac8185
commit 7866677a0c
1 changed files with 4 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
src/arch/intsimdmatrixavx2.cpp
View File

@ -2,7 +2,6 @@
// File: intsimdmatrixavx2.cpp // File: intsimdmatrixavx2.cpp
// Description: matrix-vector product for 8-bit data on avx2. // Description: matrix-vector product for 8-bit data on avx2.
// Author: Ray Smith // Author: Ray Smith
// Created: Fri Aug 04 13:26:20 PST 2017
// //
// (C) Copyright 2017, Google Inc. // (C) Copyright 2017, Google Inc.
// Licensed under the Apache License, Version 2.0 (the "License"); // Licensed under the Apache License, Version 2.0 (the "License");
@ -98,12 +97,11 @@ static inline void ExtractResults8(__m256i result,
__m256i bias_scale = _mm256_set_epi32(127,127,127,127,127,127,127,127); __m256i bias_scale = _mm256_set_epi32(127,127,127,127,127,127,127,127);
__m256d scale0123 = _mm256_loadu_pd(scales); __m256d scale0123 = _mm256_loadu_pd(scales);
__m256d scale4567 = _mm256_loadu_pd(scales+4); __m256d scale4567 = _mm256_loadu_pd(scales+4);
__m256d bias0123, bias4567, res0123, res4567;
w256 = _mm256_mullo_epi32(w256, bias_scale); // 8x32 <bias * 127> w256 = _mm256_mullo_epi32(w256, bias_scale); // 8x32 <bias * 127>
result = _mm256_add_epi32(result, w256); // result += bias * 127 result = _mm256_add_epi32(result, w256); // result += bias * 127
res0123 = _mm256_cvtepi32_pd(_mm256_castsi256_si128(result)); __m256d res0123 = _mm256_cvtepi32_pd(_mm256_castsi256_si128(result));
result = _mm256_permute4x64_epi64(result, 2+(3<<2)); result = _mm256_permute4x64_epi64(result, 2+(3<<2));
res4567 = _mm256_cvtepi32_pd(_mm256_castsi256_si128(result)); __m256d res4567 = _mm256_cvtepi32_pd(_mm256_castsi256_si128(result));
res0123 = _mm256_mul_pd(res0123, scale0123); res0123 = _mm256_mul_pd(res0123, scale0123);
res4567 = _mm256_mul_pd(res4567, scale4567); res4567 = _mm256_mul_pd(res4567, scale4567);
_mm256_storeu_pd(v, res0123); _mm256_storeu_pd(v, res0123);
@ -121,12 +119,11 @@ static inline void ExtractResults16(__m256i result0,
__m256i w256 = _mm256_cvtepi8_epi32(w8); // 8x32bit vals in 256bit reg __m256i w256 = _mm256_cvtepi8_epi32(w8); // 8x32bit vals in 256bit reg
__m256d scale0123 = _mm256_loadu_pd(scales); __m256d scale0123 = _mm256_loadu_pd(scales);
__m256d scale4567 = _mm256_loadu_pd(scales+4); __m256d scale4567 = _mm256_loadu_pd(scales+4);
__m256d bias0123, bias4567, res0123, res4567;
w256 = _mm256_mullo_epi32(w256, bias_scale); // 8x32 <bias * 127> w256 = _mm256_mullo_epi32(w256, bias_scale); // 8x32 <bias * 127>
result0 = _mm256_add_epi32(result0, w256); // result += bias * 127 result0 = _mm256_add_epi32(result0, w256); // result += bias * 127
res0123 = _mm256_cvtepi32_pd(_mm256_castsi256_si128(result0)); __m256d res0123 = _mm256_cvtepi32_pd(_mm256_castsi256_si128(result0));
result0 = _mm256_permute4x64_epi64(result0, 2+(3<<2)); result0 = _mm256_permute4x64_epi64(result0, 2+(3<<2));
res4567 = _mm256_cvtepi32_pd(_mm256_castsi256_si128(result0)); __m256d res4567 = _mm256_cvtepi32_pd(_mm256_castsi256_si128(result0));
res0123 = _mm256_mul_pd(res0123, scale0123); res0123 = _mm256_mul_pd(res0123, scale0123);
res4567 = _mm256_mul_pd(res4567, scale4567); res4567 = _mm256_mul_pd(res4567, scale4567);
_mm256_storeu_pd(v, res0123); _mm256_storeu_pd(v, res0123);
@ -286,8 +283,6 @@ static inline void PartialMatrixDotVector8(const int8_t *wi,
const int8_t *u, const int8_t *u,
int num_in, int num_in,
double *v) { double *v) {
double *ov = v;
double temp[8];
// Register containing 16-bit ones for horizontal add with 16->32 bit // Register containing 16-bit ones for horizontal add with 16->32 bit
// conversion. // conversion.
__m256i ones = __m256i ones =