Commit 95dff4ac authored by Måns Rullgård's avatar Måns Rullgård

AAC: Unswitch loop over window groups in decode_spectrum_and_dequant()

GCC should be able to do this, but doesn't.  Slightly faster code.

Originally committed as revision 21190 to svn://svn.ffmpeg.org/ffmpeg/trunk
parent 05f9d8fc
...@@ -982,13 +982,12 @@ static int decode_spectrum_and_dequant(AACContext *ac, float coef[1024], ...@@ -982,13 +982,12 @@ static int decode_spectrum_and_dequant(AACContext *ac, float coef[1024],
VLC_TYPE (*vlc_tab)[2] = vlc_spectral[cbt_m1].table; VLC_TYPE (*vlc_tab)[2] = vlc_spectral[cbt_m1].table;
const int cb_size = ff_aac_spectral_sizes[cbt_m1]; const int cb_size = ff_aac_spectral_sizes[cbt_m1];
for (group = 0; group < g_len; group++, cfo+=128) { switch (cbt_m1 >> 1) {
float *cf = cfo; case 0:
uint32_t *icf = (uint32_t *) cf; for (group = 0; group < g_len; group++, cfo+=128) {
int len = off_len; float *cf = cfo;
int len = off_len;
switch (cbt_m1 >> 1) {
case 0:
do { do {
const int index = get_vlc2(gb, vlc_tab, 6, 3); const int index = get_vlc2(gb, vlc_tab, 6, 3);
unsigned cb_idx; unsigned cb_idx;
...@@ -1001,8 +1000,14 @@ static int decode_spectrum_and_dequant(AACContext *ac, float coef[1024], ...@@ -1001,8 +1000,14 @@ static int decode_spectrum_and_dequant(AACContext *ac, float coef[1024],
cb_idx = cb_vector_idx[index]; cb_idx = cb_vector_idx[index];
cf = VMUL4(cf, vq, cb_idx, sf + idx); cf = VMUL4(cf, vq, cb_idx, sf + idx);
} while (len -= 4); } while (len -= 4);
break; }
case 1: break;
case 1:
for (group = 0; group < g_len; group++, cfo+=128) {
float *cf = cfo;
int len = off_len;
do { do {
const int index = get_vlc2(gb, vlc_tab, 6, 3); const int index = get_vlc2(gb, vlc_tab, 6, 3);
unsigned nnz; unsigned nnz;
...@@ -1019,8 +1024,14 @@ static int decode_spectrum_and_dequant(AACContext *ac, float coef[1024], ...@@ -1019,8 +1024,14 @@ static int decode_spectrum_and_dequant(AACContext *ac, float coef[1024],
bits = get_bits(gb, nnz) << (32-nnz); bits = get_bits(gb, nnz) << (32-nnz);
cf = VMUL4S(cf, vq, cb_idx, bits, sf + idx); cf = VMUL4S(cf, vq, cb_idx, bits, sf + idx);
} while (len -= 4); } while (len -= 4);
break; }
case 2: break;
case 2:
for (group = 0; group < g_len; group++, cfo+=128) {
float *cf = cfo;
int len = off_len;
do { do {
const int index = get_vlc2(gb, vlc_tab, 6, 3); const int index = get_vlc2(gb, vlc_tab, 6, 3);
unsigned cb_idx; unsigned cb_idx;
...@@ -1033,9 +1044,15 @@ static int decode_spectrum_and_dequant(AACContext *ac, float coef[1024], ...@@ -1033,9 +1044,15 @@ static int decode_spectrum_and_dequant(AACContext *ac, float coef[1024],
cb_idx = cb_vector_idx[index]; cb_idx = cb_vector_idx[index];
cf = VMUL2(cf, vq, cb_idx, sf + idx); cf = VMUL2(cf, vq, cb_idx, sf + idx);
} while (len -= 2); } while (len -= 2);
break; }
case 3: break;
case 4:
case 3:
case 4:
for (group = 0; group < g_len; group++, cfo+=128) {
float *cf = cfo;
int len = off_len;
do { do {
const int index = get_vlc2(gb, vlc_tab, 6, 3); const int index = get_vlc2(gb, vlc_tab, 6, 3);
unsigned nnz; unsigned nnz;
...@@ -1052,8 +1069,15 @@ static int decode_spectrum_and_dequant(AACContext *ac, float coef[1024], ...@@ -1052,8 +1069,15 @@ static int decode_spectrum_and_dequant(AACContext *ac, float coef[1024],
sign = get_bits(gb, nnz) << (cb_idx >> 12); sign = get_bits(gb, nnz) << (cb_idx >> 12);
cf = VMUL2S(cf, vq, cb_idx, sign, sf + idx); cf = VMUL2S(cf, vq, cb_idx, sign, sf + idx);
} while (len -= 2); } while (len -= 2);
break; }
default: break;
default:
for (group = 0; group < g_len; group++, cfo+=128) {
float *cf = cfo;
uint32_t *icf = (uint32_t *) cf;
int len = off_len;
do { do {
const int index = get_vlc2(gb, vlc_tab, 6, 3); const int index = get_vlc2(gb, vlc_tab, 6, 3);
unsigned nzt, nnz; unsigned nzt, nnz;
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment