blob: ae225d42c5cee79f553078b1ea14f37480dc57ee (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
|
#include "stdlib.h"
#include "util.h"
#include "dataset.h"
void __attribute__((noinline)) matmul(const int coreid, const int ncores, const int lda, const data_t A[], const data_t B[], data_t C[] )
{
// ***************************** //
// **** ADD YOUR CODE HERE ***** //
// ***************************** //
//
// feel free to make a separate function for MI and MSI versions.
int i, j, k;
int space=lda/ncores;
int max= space*coreid+space;
data_t temp=0;
data_t temp1=0;
data_t temp2=0;
data_t temp3=0;
data_t temp4=0;
data_t temp_1=0;
data_t temp1_1=0;
data_t temp2_1=0;
data_t temp3_1=0;
data_t temp4_1=0;
data_t temp_2=0;
data_t temp1_2=0;
data_t temp2_2=0;
data_t temp3_2=0;
data_t temp4_2=0;
data_t temp_3=0;
data_t temp1_3=0;
data_t temp2_3=0;
data_t temp3_3=0;
data_t temp4_3=0;
if (coreid!=ncores-1){
//main loop
for (i=space*coreid;i<max/4*4;i+=4)
{
for(j=0;j<lda;j+=4)
{
temp1=C[j+i*lda];
temp2=C[j+1+i*lda];
temp3=C[j+2+i*lda];
temp4=C[j+3+i*lda];
temp1_1=C[j+(i+1)*lda];
temp2_1=C[j+1+(i+1)*lda];
temp3_1=C[j+2+(i+1)*lda];
temp4_1=C[j+3+(i+1)*lda];
temp1_2=C[j+(i+2)*lda];
temp2_2=C[j+1+(i+2)*lda];
temp3_2=C[j+2+(i+2)*lda];
temp4_2=C[j+3+(i+2)*lda];
temp1_3=C[j+(i+3)*lda];
temp2_3=C[j+1+(i+3)*lda];
temp3_3=C[j+2+(i+3)*lda];
temp4_3=C[j+3+(i+3)*lda];
for (k=0;k<lda;k++)
{
temp=A[k+i*lda];
temp1+=temp*B[j+k*lda];
temp2+=temp*B[j+1+k*lda];
temp3+=temp*B[j+2+k*lda];
temp4+=temp*B[j+3+k*lda];
temp_1=A[k+(i+1)*lda];
temp1_1+=temp_1*B[j+k*lda];
temp2_1+=temp_1*B[j+1+k*lda];
temp3_1+=temp_1*B[j+2+k*lda];
temp4_1+=temp_1*B[j+3+k*lda];
temp_2=A[k+(i+2)*lda];
temp1_2+=temp_2*B[j+k*lda];
temp2_2+=temp_2*B[j+1+k*lda];
temp3_2+=temp_2*B[j+2+k*lda];
temp4_2+=temp_2*B[j+3+k*lda];
temp_3=A[k+(i+3)*lda];
temp1_3+=temp_3*B[j+k*lda];
temp2_3+=temp_3*B[j+1+k*lda];
temp3_3+=temp_3*B[j+2+k*lda];
temp4_3+=temp_3*B[j+3+k*lda];
}
C[j+i*lda]=temp1;
C[j+1+i*lda]=temp2;
C[j+2+i*lda]=temp3;
C[j+3+i*lda]=temp4;
C[j+(i+1)*lda]=temp1_1;
C[j+1+(i+1)*lda]=temp2_1;
C[j+2+(i+1)*lda]=temp3_1;
C[j+3+(i+1)*lda]=temp4_1;
C[j+(i+2)*lda]=temp1_2;
C[j+1+(i+2)*lda]=temp2_2;
C[j+2+(i+2)*lda]=temp3_2;
C[j+3+(i+2)*lda]=temp4_2;
C[j+(i+3)*lda]=temp1_3;
C[j+1+(i+3)*lda]=temp2_3;
C[j+2+(i+3)*lda]=temp3_3;
C[j+3+(i+3)*lda]=temp4_3;
}
}
}
//second core
else{
for (i=space*coreid;i<lda/4*4;i+=4)
{
for(j=0;j<lda;j+=4)
{
temp1=C[j+i*lda];
temp2=C[j+1+i*lda];
temp3=C[j+2+i*lda];
temp4=C[j+3+i*lda];
temp1_1=C[j+(i+1)*lda];
temp2_1=C[j+1+(i+1)*lda];
temp3_1=C[j+2+(i+1)*lda];
temp4_1=C[j+3+(i+1)*lda];
temp1_2=C[j+(i+2)*lda];
temp2_2=C[j+1+(i+2)*lda];
temp3_2=C[j+2+(i+2)*lda];
temp4_2=C[j+3+(i+2)*lda];
temp1_3=C[j+(i+3)*lda];
temp2_3=C[j+1+(i+3)*lda];
temp3_3=C[j+2+(i+3)*lda];
temp4_3=C[j+3+(i+3)*lda];
for (k=0;k<lda;k++)
{
temp=A[k+i*lda];
temp1+=temp*B[j+k*lda];
temp2+=temp*B[j+1+k*lda];
temp3+=temp*B[j+2+k*lda];
temp4+=temp*B[j+3+k*lda];
temp_1=A[k+(i+1)*lda];
temp1_1+=temp_1*B[j+k*lda];
temp2_1+=temp_1*B[j+1+k*lda];
temp3_1+=temp_1*B[j+2+k*lda];
temp4_1+=temp_1*B[j+3+k*lda];
temp_2=A[k+(i+2)*lda];
temp1_2+=temp_2*B[j+k*lda];
temp2_2+=temp_2*B[j+1+k*lda];
temp3_2+=temp_2*B[j+2+k*lda];
temp4_2+=temp_2*B[j+3+k*lda];
temp_3=A[k+(i+3)*lda];
temp1_3+=temp_3*B[j+k*lda];
temp2_3+=temp_3*B[j+1+k*lda];
temp3_3+=temp_3*B[j+2+k*lda];
temp4_3+=temp_3*B[j+3+k*lda];
}
C[j+i*lda]=temp1;
C[j+1+i*lda]=temp2;
C[j+2+i*lda]=temp3;
C[j+3+i*lda]=temp4;
C[j+(i+1)*lda]=temp1_1;
C[j+1+(i+1)*lda]=temp2_1;
C[j+2+(i+1)*lda]=temp3_1;
C[j+3+(i+1)*lda]=temp4_1;
C[j+(i+2)*lda]=temp1_2;
C[j+1+(i+2)*lda]=temp2_2;
C[j+2+(i+2)*lda]=temp3_2;
C[j+3+(i+2)*lda]=temp4_2;
C[j+(i+3)*lda]=temp1_3;
C[j+1+(i+3)*lda]=temp2_3;
C[j+2+(i+3)*lda]=temp3_3;
C[j+3+(i+3)*lda]=temp4_3;
}
}
}
}
|
