6 void __attribute__((noinline
)) matmul(const int coreid
, const int ncores
, const int lda
, const data_t A
[], const data_t B
[], data_t C
[] )
9 // ***************************** //
10 // **** ADD YOUR CODE HERE ***** //
11 // ***************************** //
13 // feel free to make a separate function for MI and MSI versions.
15 data_t temp0
, temp1
, temp2
, temp3
, temp4
, temp5
, temp6
, temp7
;
16 data_t temp8
, temp9
, temp10
, temp11
, temp12
, temp13
, temp14
, temp15
;
18 for(j
= 0; j
< 32; j
++) {
19 temp0
= 0; //C[j*lda];
20 temp1
= 0; //C[1 + j*lda];
21 temp2
= 0; //C[2 + j*lda];
22 temp3
= 0; //C[3 + j*lda];
23 temp4
= 0; //C[4 + j*lda];
24 temp5
= 0; //C[5 + j*lda];
25 temp6
= 0; //C[6 + j*lda];
26 temp7
= 0; //C[7 + j*lda];
27 temp8
= 0; //C[8 + j*lda];
28 temp9
= 0; //C[9 + j*lda];
29 temp10
= 0; //C[10 + j*lda];
30 temp11
= 0; //C[11 + j*lda];
31 temp12
= 0; //C[12 + j*lda];
32 temp13
= 0; //C[13 + j*lda];
33 temp14
= 0; //C[14 + j*lda];
34 temp15
= 0; //C[15 + j*lda];
35 for(k
= 0; k
< 32; k
++) {
36 temp0
+= A
[j
*lda
+ k
] * B
[k
*lda
];
37 temp1
+= A
[j
*lda
+ k
] * B
[1+k
*lda
];
38 temp2
+= A
[j
*lda
+ k
] * B
[2+k
*lda
];
39 temp3
+= A
[j
*lda
+ k
] * B
[3+k
*lda
];
40 temp4
+= A
[j
*lda
+ k
] * B
[4+k
*lda
];
41 temp5
+= A
[j
*lda
+ k
] * B
[5+k
*lda
];
42 temp6
+= A
[j
*lda
+ k
] * B
[6+k
*lda
];
43 temp7
+= A
[j
*lda
+ k
] * B
[7+k
*lda
];
44 temp8
+= A
[j
*lda
+ k
] * B
[8+k
*lda
];
45 temp9
+= A
[j
*lda
+ k
] * B
[9+k
*lda
];
46 temp10
+= A
[j
*lda
+ k
] * B
[10+k
*lda
];
47 temp11
+= A
[j
*lda
+ k
] * B
[11+k
*lda
];
48 temp12
+= A
[j
*lda
+ k
] * B
[12+k
*lda
];
49 temp13
+= A
[j
*lda
+ k
] * B
[13+k
*lda
];
50 temp14
+= A
[j
*lda
+ k
] * B
[14+k
*lda
];
51 temp15
+= A
[j
*lda
+ k
] * B
[15+k
*lda
];
63 C
[10 + j
*lda
] = temp10
;
64 C
[11 + j
*lda
] = temp11
;
65 C
[12 + j
*lda
] = temp12
;
66 C
[13 + j
*lda
] = temp13
;
67 C
[14 + j
*lda
] = temp14
;
68 C
[15 + j
*lda
] = temp15
;
72 if(coreid
== 1 || ncores
== 1) {
73 for(j
= 0; j
< 32; j
++) {
74 temp0
= 0; //C[16+j*lda];
75 temp1
= 0; //C[17+j*lda];
76 temp2
= 0; //C[18+j*lda];
77 temp3
= 0; //C[19+j*lda];
78 temp4
= 0; //C[20+j*lda];
79 temp5
= 0; //C[21+j*lda];
80 temp6
= 0; //C[22+j*lda];
81 temp7
= 0; //C[23+j*lda];
82 temp8
= 0; //C[24+j*lda];
83 temp9
= 0; //C[25+j*lda];
84 temp10
= 0; //C[26+j*lda];
85 temp11
= 0; //C[27+j*lda];
86 temp12
= 0; //C[28+j*lda];
87 temp13
= 0; //C[29+j*lda];
88 temp14
= 0; //C[30+j*lda];
89 temp15
= 0; //C[31+j*lda];
90 for(k
= 0; k
< 32; k
++) {
91 temp0
+= A
[j
*lda
+ k
] * B
[16+k
*lda
];
92 temp1
+= A
[j
*lda
+ k
] * B
[17+k
*lda
];
93 temp2
+= A
[j
*lda
+ k
] * B
[18+k
*lda
];
94 temp3
+= A
[j
*lda
+ k
] * B
[19+k
*lda
];
95 temp4
+= A
[j
*lda
+ k
] * B
[20+k
*lda
];
96 temp5
+= A
[j
*lda
+ k
] * B
[21+k
*lda
];
97 temp6
+= A
[j
*lda
+ k
] * B
[22+k
*lda
];
98 temp7
+= A
[j
*lda
+ k
] * B
[23+k
*lda
];
99 temp8
+= A
[j
*lda
+ k
] * B
[24+k
*lda
];
100 temp9
+= A
[j
*lda
+ k
] * B
[25+k
*lda
];
101 temp10
+= A
[j
*lda
+ k
] * B
[26+k
*lda
];
102 temp11
+= A
[j
*lda
+ k
] * B
[27+k
*lda
];
103 temp12
+= A
[j
*lda
+ k
] * B
[28+k
*lda
];
104 temp13
+= A
[j
*lda
+ k
] * B
[29+k
*lda
];
105 temp14
+= A
[j
*lda
+ k
] * B
[30+k
*lda
];
106 temp15
+= A
[j
*lda
+ k
] * B
[31+k
*lda
];
108 C
[16 + j
*lda
] = temp0
;
109 C
[17 + j
*lda
] = temp1
;
110 C
[18 + j
*lda
] = temp2
;
111 C
[19 + j
*lda
] = temp3
;
112 C
[20 + j
*lda
] = temp4
;
113 C
[21 + j
*lda
] = temp5
;
114 C
[22 + j
*lda
] = temp6
;
115 C
[23 + j
*lda
] = temp7
;
116 C
[24 + j
*lda
] = temp8
;
117 C
[25 + j
*lda
] = temp9
;
118 C
[26 + j
*lda
] = temp10
;
119 C
[27 + j
*lda
] = temp11
;
120 C
[28 + j
*lda
] = temp12
;
121 C
[29 + j
*lda
] = temp13
;
122 C
[30 + j
*lda
] = temp14
;
123 C
[31 + j
*lda
] = temp15
;