generic-library/openjpeg
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Significant optimizations of MCT, DWT, MQ and T1 modules by Peter Wimmer (thanks Peter)

Browse Source
This commit is contained in:
Francois-Olivier Devaux
2010-04-08 17:22:58 +00:00
parent d03779ee2a
commit 627f484bce
9 changed files with 612 additions and 86 deletions
Download Patch File Download Diff File Expand all files Collapse all files

79
libopenjpeg/dwt.c
View File

@@ -570,6 +570,20 @@ static void v4dwt_interleave_h(v4dwt_t* restrict w, float* restrict a, int x, in
int count = w->sn;
int i, k;
for(k = 0; k < 2; ++k){
if (count + 3 * x < size && ((int) a & 0x0f) == 0 && ((int) bi & 0x0f) == 0 && (x & 0x0f) == 0) {
/* Fast code path */
for(i = 0; i < count; ++i){
int j = i;
bi[i*8 ] = a[j];
j += x;
bi[i*8 + 1] = a[j];
j += x;
bi[i*8 + 2] = a[j];
j += x;
bi[i*8 + 3] = a[j];
}
} else {
/* Slow code path */
for(i = 0; i < count; ++i){
int j = i;
bi[i*8 ] = a[j];
@@ -583,6 +597,7 @@ static void v4dwt_interleave_h(v4dwt_t* restrict w, float* restrict a, int x, in
if(j > size) continue;
bi[i*8 + 3] = a[j];
}
}
bi = (float*) (w->wavelet + 1 - w->cas);
a += w->sn;
size -= w->sn;
@@ -608,9 +623,21 @@ static void v4dwt_interleave_v(v4dwt_t* restrict v , float* restrict a , int x){
static void v4dwt_decode_step1_sse(v4* w, int count, const __m128 c){
__m128* restrict vw = (__m128*) w;
int i;
/* 4x unrolled loop */
for(i = 0; i < count >> 2; ++i){
*vw = _mm_mul_ps(*vw, c);
vw += 2;
*vw = _mm_mul_ps(*vw, c);
vw += 2;
*vw = _mm_mul_ps(*vw, c);
vw += 2;
*vw = _mm_mul_ps(*vw, c);
vw += 2;
}
count &= 3;
for(i = 0; i < count; ++i){
__m128 tmp = vw[i*2];
vw[i*2] = tmp * c;
*vw = _mm_mul_ps(*vw, c);
vw += 2;
}
}
@@ -618,22 +645,24 @@ static void v4dwt_decode_step2_sse(v4* l, v4* w, int k, int m, __m128 c){
__m128* restrict vl = (__m128*) l;
__m128* restrict vw = (__m128*) w;
int i;
__m128 tmp1, tmp2, tmp3;
tmp1 = vl[0];
for(i = 0; i < m; ++i){
__m128 tmp1 = vl[ 0];
__m128 tmp2 = vw[-1];
__m128 tmp3 = vw[ 0];
vw[-1] = tmp2 + ((tmp1 + tmp3) * c);
vl = vw;
tmp2 = vw[-1];
tmp3 = vw[ 0];
vw[-1] = _mm_add_ps(tmp2, _mm_mul_ps(_mm_add_ps(tmp1, tmp3), c));
tmp1 = tmp3;
vw += 2;
}
vl = vw - 2;
if(m >= k){
return;
}
c += c;
c *= vl[0];
c = _mm_add_ps(c, c);
c = _mm_mul_ps(c, vl[0]);
for(; m < k; ++m){
__m128 tmp = vw[-1];
vw[-1] = tmp + c;
vw[-1] = _mm_add_ps(tmp, c);
vw += 2;
}
}
@@ -773,19 +802,24 @@ void dwt_decode_real(opj_tcd_tilecomp_t* restrict tilec, int numres){
h.dn = rw - h.sn;
h.cas = res->x0 % 2;
for(j = rh; j > 0; j -= 4){
for(j = rh; j > 3; j -= 4){
int k;
v4dwt_interleave_h(&h, aj, w, bufsize);
v4dwt_decode(&h);
if(j >= 4){
int k;
for(k = rw; --k >= 0;){
aj[k ] = h.wavelet[k].f[0];
aj[k+w ] = h.wavelet[k].f[1];
aj[k+w*2] = h.wavelet[k].f[2];
aj[k+w*3] = h.wavelet[k].f[3];
}
}else{
aj += w*4;
bufsize -= w*4;
}
if (rh & 0x03) {
int k;
j = rh & 0x03;
v4dwt_interleave_h(&h, aj, w, bufsize);
v4dwt_decode(&h);
for(k = rw; --k >= 0;){
switch(j) {
case 3: aj[k+w*2] = h.wavelet[k].f[2];
@@ -794,30 +828,29 @@ void dwt_decode_real(opj_tcd_tilecomp_t* restrict tilec, int numres){
}
}
}
aj += w*4;
bufsize -= w*4;
}
v.dn = rh - v.sn;
v.cas = res->y0 % 2;
aj = (float*) tilec->data;
for(j = rw; j > 0; j -= 4){
for(j = rw; j > 3; j -= 4){
int k;
v4dwt_interleave_v(&v, aj, w);
v4dwt_decode(&v);
if(j >= 4){
int k;
for(k = 0; k < rh; ++k){
memcpy(&aj[k*w], &v.wavelet[k], 4 * sizeof(float));
}
}else{
aj += 4;
}
if (rw & 0x03){
int k;
j = rw & 0x03;
v4dwt_interleave_v(&v, aj, w);
v4dwt_decode(&v);
for(k = 0; k < rh; ++k){
memcpy(&aj[k*w], &v.wavelet[k], j * sizeof(float));
}
}
aj += 4;
}
}
opj_aligned_free(h.wavelet);