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
| | This patch fixes two separate issues with ustream sources:
* Normalize import paths in jsrc/cip.c
Upstream claims to have some build requirements that force them to use strange
import paths. However, these paths do not exist inside our build chroot.
* Fix unititialized variable warning
Clang 9 issues some warnings which cause the build to fail since upstream
compiles with -Werror.
diff --git a/jsrc/cip.c b/jsrc/cip.c
index 61da4088..fb3c03b6 100644
--- a/jsrc/cip.c
+++ b/jsrc/cip.c
@@ -3,9 +3,9 @@
/* */
/* Conjunctions: Inner Product */
-#include "../../jsource/jsrc/j.h"
-#include "../../jsource/jsrc/vasm.h"
-#include "../../jsource/jsrc/gemm.h"
+#include "j.h"
+#include "vasm.h"
+#include "gemm.h"
#define MAXAROWS 384 // max rows of a that we can process to stay in L2 cache a strip is m*CACHEHEIGHT, z strip is m*CACHEWIDTH this is wired to 128*3 - check if you chage
@@ -1057,15 +1057,15 @@ static A jtipbx(J jt,A a,A w,C c,C d){A g=0,x0,x1,z;B*av,*av0,b,*v0,*v1,*zv;C c0
switch(c){
case CPLUSDOT:
#define F |=
-#include "../../jsource/jsrc/cip_t.h"
+#include "cip_t.h"
break;
case CSTARDOT:
#define F &=
-#include "../../jsource/jsrc/cip_t.h"
+#include "cip_t.h"
break;
case CNE:
#define F ^=
-#include "../../jsource/jsrc/cip_t.h"
+#include "cip_t.h"
break;
}
R z;
diff --git a/jsrc/gemm.c b/jsrc/gemm.c
index 51fe306e..b105dfc1 100644
--- a/jsrc/gemm.c
+++ b/jsrc/gemm.c
@@ -318,7 +318,7 @@ dgemm_nn (I m,
_B);
// loop 3
- I i;
+ I i=0;
#pragma omp parallel for default(none),private(i),shared(j,l,A,C,mb,nc,kc,alpha,_beta,_mc,_B,rs_a,cs_a,rs_c,cs_c)
for (i=0; i<mb; ++i) {
I mc;
@@ -501,7 +501,7 @@ igemm_nn (I m,
_B);
// loop 3
- I i;
+ I i=0;
#pragma omp parallel for default(none),private(i),shared(j,l,A,C,mb,nc,kc,alpha,_beta,_mc,_B,rs_a,cs_a,rs_c,cs_c)
for (i=0; i<mb; ++i) {
I mc;
@@ -831,7 +831,7 @@ zgemm_nn (I m,
_B);
// loop 3
- I i;
+ I i=0;
#pragma omp parallel for default(none),private(i),shared(j,l,A,C,mb,nc,kc,alpha,_beta,_mc,_B,rs_a,cs_a,rs_c,cs_c)
for (i=0; i<mb; ++i) {
I mc;
|