Coverage Report

Created: 2026-01-16 20:29

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
/root/doris/contrib/openblas/kernel/generic/dot.c
Line
Count
Source
1
/***************************************************************************
2
Copyright (c) 2014, The OpenBLAS Project
3
All rights reserved.
4
Redistribution and use in source and binary forms, with or without
5
modification, are permitted provided that the following conditions are
6
met:
7
1. Redistributions of source code must retain the above copyright
8
notice, this list of conditions and the following disclaimer.
9
2. Redistributions in binary form must reproduce the above copyright
10
notice, this list of conditions and the following disclaimer in
11
the documentation and/or other materials provided with the
12
distribution.
13
3. Neither the name of the OpenBLAS project nor the names of
14
its contributors may be used to endorse or promote products
15
derived from this software without specific prior written permission.
16
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
17
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19
ARE DISCLAIMED. IN NO EVENT SHALL THE OPENBLAS PROJECT OR CONTRIBUTORS BE
20
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21
DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
22
SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
23
CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
24
OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE
25
USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
26
*****************************************************************************/
27
28
29
#include "common.h"
30
#include "../simd/intrin.h"
31
#if defined(DSDOT)
32
double CNAME(BLASLONG n, FLOAT *x, BLASLONG inc_x, FLOAT *y, BLASLONG inc_y)
33
#else
34
FLOAT CNAME(BLASLONG n, FLOAT *x, BLASLONG inc_x, FLOAT *y, BLASLONG inc_y)
35
#endif
36
0
{
37
0
  BLASLONG i=0;
38
0
  BLASLONG ix=0,iy=0;
39
40
#if defined(DSDOT)
41
  double dot = 0.0 ;
42
#else
43
0
  FLOAT  dot = 0.0 ;
44
#endif
45
46
0
  if ( n < 1 )  return(dot);
47
48
0
  if ( (inc_x == 1) && (inc_y == 1) )
49
0
  {
50
#if V_SIMD && !defined(DSDOT)
51
0
        const int vstep = v_nlanes_f32;
52
        const int unrollx4 = n & (-vstep * 4);
53
        const int unrollx  = n &  -vstep;
54
0
    v_f32 vsum0 = v_zero_f32();
55
0
        v_f32 vsum1 = v_zero_f32();
56
0
        v_f32 vsum2 = v_zero_f32();
57
0
        v_f32 vsum3 = v_zero_f32();
58
0
    while(i < unrollx4)
59
0
        {
60
0
            vsum0 = v_muladd_f32(
61
0
                v_loadu_f32(x + i),           v_loadu_f32(y + i),           vsum0
62
0
            );
63
0
            vsum1 = v_muladd_f32(
64
0
                v_loadu_f32(x + i + vstep),   v_loadu_f32(y + i + vstep),   vsum1
65
0
            );
66
0
            vsum2 = v_muladd_f32(
67
0
                v_loadu_f32(x + i + vstep*2), v_loadu_f32(y + i + vstep*2), vsum2
68
0
            );
69
0
            vsum3 = v_muladd_f32(
70
0
                v_loadu_f32(x + i + vstep*3), v_loadu_f32(y + i + vstep*3), vsum3
71
0
            );
72
0
            i += vstep*4;
73
0
        }
74
0
        vsum0 = v_add_f32(
75
0
            v_add_f32(vsum0, vsum1), v_add_f32(vsum2 , vsum3)
76
        );
77
0
    while(i < unrollx)
78
0
        {
79
0
            vsum0 = v_muladd_f32(
80
0
                v_loadu_f32(x + i), v_loadu_f32(y + i), vsum0
81
0
            );
82
0
            i += vstep;
83
0
        }
84
        dot = v_sum_f32(vsum0);
85
#elif defined(DSDOT)
86
        int n1 = n & -4;
87
0
    for (; i < n1; i += 4)
88
0
    {
89
0
      dot += (double) y[i] * (double) x[i]
90
0
          + (double) y[i+1] * (double) x[i+1]
91
0
          + (double) y[i+2] * (double) x[i+2]
92
0
          + (double) y[i+3] * (double) x[i+3] ;
93
0
    }
94
#else
95
        int n1 = n & -4;
96
    for (; i < n1; i += 4)
97
    {
98
      dot += y[i] * x[i]
99
          + y[i+1] * x[i+1]
100
          + y[i+2] * x[i+2]
101
          + y[i+3] * x[i+3] ;
102
    }
103
#endif
104
0
    while(i < n)
105
0
    {
106
107
#if defined(DSDOT)
108
      dot += (double) y[i] * (double) x[i] ;
109
#else
110
      dot += y[i] * x[i] ;
111
#endif
112
0
      i++ ;
113
114
0
    }
115
0
    return(dot);
116
117
118
0
  }
119
120
0
  while(i < n)
121
0
  {
122
123
#if defined(DSDOT)
124
    dot += (double) y[iy] * (double) x[ix] ;
125
#else
126
    dot += y[iy] * x[ix] ;
127
#endif
128
0
    ix  += inc_x ;
129
0
    iy  += inc_y ;
130
0
    i++ ;
131
132
0
  }
133
0
  return(dot);
134
135
0
}
Unexecuted instantiation: sdot_k_PRESCOTT
Unexecuted instantiation: dsdot_k_PRESCOTT
Unexecuted instantiation: sdot_k_CORE2
Unexecuted instantiation: dsdot_k_CORE2
Unexecuted instantiation: sdot_k_BARCELONA
Unexecuted instantiation: dsdot_k_BARCELONA
136
137