forked from kostrzewa/tmLQCD
-
Notifications
You must be signed in to change notification settings - Fork 0
/
bgq.h
96 lines (78 loc) · 3 KB
/
bgq.h
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
#ifndef _BGQ_H
#define _BGQ_H
#include "bgq2.h"
#define _vec_load_spinor(r0, r1, r2, r3, r4, r5, phi) \
r0 = vec_ld(0L, (double*) &(phi).c0); \
r1 = vec_ld(32L, (double*) &(phi).c0); \
r2 = vec_ld(64L, (double*) &(phi).c0); \
r3 = vec_ld(96L, (double*) &(phi).c0); \
r4 = vec_ld(128L, (double*) &(phi).c0); \
r5 = vec_ld(160L, (double*) &(phi).c0);
#define _vec_load_halfspinor(r0, r1, r2, phi) \
r0 = vec_ld(0L, (double*) &(phi).c0); \
r1 = vec_ld(32L, (double*) &(phi).c0); \
r2 = vec_ld(64L, (double*) &(phi).c0);
#define _vec_store_spinor(phi, r0, r1, r2, r3, r4, r5) \
vec_st(r0, 0L, (double*) &(phi).c0); \
vec_st(r1, 32L, (double*) &(phi).c0); \
vec_st(r2, 64L, (double*) &(phi).c0); \
vec_st(r3, 96L, (double*) &(phi).c0); \
vec_st(r4, 128L, (double*) &(phi).c0); \
vec_st(r5, 160L, (double*) &(phi).c0);
#define _vec_add_ul_spinor(rs0, rs1, rs2, r0, r1, r2, r3, r4, r5) \
rs0 = vec_add(r0, r3); \
rs1 = vec_add(r1, r4); \
rs2 = vec_add(r2, r5);
#define _vec_sub_ul_spinor(rs0, rs1, rs2, r0, r1, r2, r3, r4, r5) \
rs0 = vec_sub(r0, r3); \
rs1 = vec_sub(r1, r4); \
rs2 = vec_sub(r2, r5);
// requires 32 byte alignment of phi
#define _vec_load(r0, r1, phi) \
r0 = vec_ld(0L, (double*) &(phi).c0); \
r1 = vec_ld2(0L, (double*) &(phi).c2);
// works also with 16 byte alignement of phi
#define _vec_load16(r0, r1, phi, tmp) \
r0 = vec_ld2(0L, (double*) &(phi).c0); \
r1 = vec_ld(0L, (double*) &(phi).c1); \
tmp = vec_gpci(00145); \
r0 = vec_perm(r0, r1, tmp); \
tmp = vec_gpci(02301); \
r1 = vec_perm(r1, r0, tmp);
// alternative
#define _vec_load16c(r0, r1, phi, tmp) \
r0 = vec_ld2(0L, (double*) &(phi).c0); \
r1 = vec_ld(0L, (double*) &(phi).c1); \
tmp = vec_gpci(00145); \
r0 = vec_perm(r0, r1, tmp); \
r1 = vec_ld2(0L, (double*) &(phi).c2);
// requires 32 byte alignment of phi
#define _vec_store(phi, r0, r1) \
vec_st((r0), 0L, (double*) &(phi).c0); \
vec_st2((r1), 0L, (double*) &(phi).c2);
// requires 16 (and must not be 32) byte alignment of phi
#define _vec_store16(phi, r0, r1, tmp) \
vec_st2((r0), 0L, (double*) &(phi).c0); \
tmp = vec_gpci(02345); \
r0 = vec_perm(r0, r1, tmp); \
vec_st((r0), 0L, (double *) &(phi).c1);
// requires 32 byte alignment of phi
#define _vec_store_halfspinor(phi, r0, r1, r2) \
vec_st((r0), 0L, (double*) &(phi).c0); \
vec_st((r1), 32L, (double*) &(phi).c0); \
vec_st((r2), 64L, (double*) &(phi).c0);
#define _vec_add(rs0, rs1, r0, r1, s0, s1) \
rs0 = vec_add(r0, s0); \
rs1 = vec_add(r1, s1);
#define _vec_sub(rs0, rs1, r0, r1, s0, s1) \
rs0 = vec_sub(r0, s0); \
rs1 = vec_sub(r1, s1);
#define _vec_i_mul_add(rs0, rs1, r0, r1, s0, s1, tmp) \
tmp = vec_splats(1.); \
rs0 = vec_xxnpmadd(s0, tmp, r0); \
rs1 = vec_xxnpmadd(s1, tmp, r1);
#define _vec_i_mul_sub(rs0, rs1, r0, r1, s0, s1, tmp) \
tmp = vec_splats(-1.); \
rs0 = vec_xxnpmadd(s0, tmp, r0); \
rs1 = vec_xxnpmadd(s1, tmp, r1);
#endif