28 vector
float d0, d1,
s,
zero = (vector float)vec_splat_u32(0);
29 for (i = 0; i < len - 7; i += 8) {
30 d0 = vec_ld( 0, src0 + i);
31 s = vec_ld( 0, src1 + i);
32 d1 = vec_ld(16, src0 + i);
33 d0 = vec_madd(d0, s, zero);
34 d1 = vec_madd(d1, vec_ld(16, src1 + i), zero);
35 vec_st(d0, 0, dst + i);
36 vec_st(d1, 16, dst + i);
41 const float *src1,
const float *win,
int len)
44 const vector
unsigned char reverse = vcprm(3, 2, 1, 0);
51 zero = (vector float)vec_splat_u32(0);
53 for (i = -len * 4, j = len * 4 - 16; i < 0; i += 16, j -= 16) {
59 s1 = vec_perm(s1, s1, reverse);
60 wj = vec_perm(wj, wj, reverse);
62 t0 = vec_madd(s0, wj, zero);
63 t0 = vec_nmsub(s1, wi, t0);
64 t1 = vec_madd(s0, wi, zero);
65 t1 = vec_madd(s1, wj, t1);
66 t1 = vec_perm(t1, t1, reverse);
74 const float *src1,
const float *src2,
79 vector
unsigned char align = vec_lvsr(0,dst),
80 mask = vec_lvsl(0, dst);
82 for (i = 0; i < len - 3; i += 4) {
83 t0 = vec_ld(0, dst + i);
84 t1 = vec_ld(15, dst + i);
85 s0 = vec_ld(0, src0 + i);
86 s1 = vec_ld(0, src1 + i);
87 s2 = vec_ld(0, src2 + i);
88 edges = vec_perm(t1, t0,
mask);
89 d = vec_madd(s0, s1, s2);
90 t1 = vec_perm(d, edges, align);
91 t0 = vec_perm(edges, d, align);
92 vec_st(t1, 15, dst + i);
93 vec_st(t0, 0, dst + i);
98 const float *src1,
int len)
101 vector
float d,
s0,
s1, h0, l0,
102 s2,
s3,
zero = (vector float) vec_splat_u32(0);
105 for(i = 0; i < len - 7; i += 8) {
106 s1 = vec_ld(0, src1 - i);
107 s0 = vec_ld(0, src0 + i);
108 l0 = vec_mergel(s1, s1);
109 s3 = vec_ld(-16, src1 - i);
110 h0 = vec_mergeh(s1, s1);
111 s2 = vec_ld(16, src0 + i);
112 s1 = vec_mergeh(vec_mergel(l0, h0),
115 l0 = vec_mergel(s3, s3);
116 d = vec_madd(s0, s1, zero);
117 h0 = vec_mergeh(s3, s3);
118 vec_st(d, 0, dst + i);
119 s3 = vec_mergeh(vec_mergel(l0, h0),
121 d = vec_madd(s2, s3, zero);
122 vec_st(d, 16, dst + i);