aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorSebastian Vater <cdgs.basty@googlemail.com>2010-04-26 22:38:41 +0000
committerRonald S. Bultje <rsbultje@gmail.com>2010-04-26 22:38:41 +0000
commitfe51b5ce504d118757407c0855e957e97ca90f78 (patch)
tree7e545df3ccd5b973a721dcb34adff5153cd2c869
parent473147bed01c0c6c82d85fd79d3e1c1d65542663 (diff)
downloadffmpeg-fe51b5ce504d118757407c0855e957e97ca90f78.tar.gz
Move some branches outside looped code. Should improve the generated asm (and
thus performance) slightly. Patch by Sebastian Vater <cdgs.basty googlemail com>. Originally committed as revision 22975 to svn://svn.ffmpeg.org/ffmpeg/trunk
-rw-r--r--libavcodec/iff.c43
1 files changed, 37 insertions, 6 deletions
diff --git a/libavcodec/iff.c b/libavcodec/iff.c
index b5ee2e6e47..e628c5e111 100644
--- a/libavcodec/iff.c
+++ b/libavcodec/iff.c
@@ -140,16 +140,23 @@ static int decode_frame_ilbm(AVCodecContext *avctx,
return -1;
}
+ if (avctx->pix_fmt == PIX_FMT_PAL8) {
for(y = 0; y < avctx->height; y++ ) {
uint8_t *row = &s->frame.data[0][ y*s->frame.linesize[0] ];
- memset(row, 0, avctx->pix_fmt == PIX_FMT_PAL8 ? avctx->width : (avctx->width * 4));
+ memset(row, 0, avctx->width);
for (plane = 0; plane < avctx->bits_per_coded_sample && buf < buf_end; plane++) {
- if (avctx->pix_fmt == PIX_FMT_PAL8) {
decodeplane8(row, buf, FFMIN(s->planesize, buf_end - buf), avctx->bits_per_coded_sample, plane);
+ buf += s->planesize;
+ }
+ }
} else { // PIX_FMT_BGR32
+ for(y = 0; y < avctx->height; y++ ) {
+ uint8_t *row = &s->frame.data[0][y*s->frame.linesize[0]];
+ memset(row, 0, avctx->width << 2);
+ for (plane = 0; plane < avctx->bits_per_coded_sample && buf < buf_end; plane++) {
decodeplane32((uint32_t *) row, buf, FFMIN(s->planesize, buf_end - buf), avctx->bits_per_coded_sample, plane);
- }
buf += s->planesize;
+ }
}
}
@@ -173,10 +180,11 @@ static int decode_frame_byterun1(AVCodecContext *avctx,
return -1;
}
+ if (avctx->codec_tag == MKTAG('I','L','B','M')) { //interleaved
+ if (avctx->pix_fmt == PIX_FMT_PAL8) {
for(y = 0; y < avctx->height ; y++ ) {
uint8_t *row = &s->frame.data[0][ y*s->frame.linesize[0] ];
- if (avctx->codec_tag == MKTAG('I','L','B','M')) { //interleaved
- memset(row, 0, avctx->pix_fmt == PIX_FMT_PAL8 ? avctx->width : (avctx->width * 4));
+ memset(row, 0, avctx->width);
for (plane = 0; plane < avctx->bits_per_coded_sample; plane++) {
for(x = 0; x < s->planesize && buf < buf_end; ) {
int8_t value = *buf++;
@@ -193,13 +201,36 @@ static int decode_frame_byterun1(AVCodecContext *avctx,
}
x += length;
}
- if (avctx->pix_fmt == PIX_FMT_PAL8) {
decodeplane8(row, s->planebuf, s->planesize, avctx->bits_per_coded_sample, plane);
+ }
+ }
} else { //PIX_FMT_BGR32
+ for(y = 0; y < avctx->height ; y++ ) {
+ uint8_t *row = &s->frame.data[0][y*s->frame.linesize[0]];
+ memset(row, 0, avctx->width << 2);
+ for (plane = 0; plane < avctx->bits_per_coded_sample; plane++) {
+ for(x = 0; x < s->planesize && buf < buf_end; ) {
+ int8_t value = *buf++;
+ unsigned length;
+ if (value >= 0) {
+ length = value + 1;
+ memcpy(s->planebuf + x, buf, FFMIN3(length, s->planesize - x, buf_end - buf));
+ buf += length;
+ } else if (value > -128) {
+ length = -value + 1;
+ memset(s->planebuf + x, *buf++, FFMIN(length, s->planesize - x));
+ } else { // noop
+ continue;
+ }
+ x += length;
+ }
decodeplane32((uint32_t *) row, s->planebuf, s->planesize, avctx->bits_per_coded_sample, plane);
}
}
+ }
} else {
+ for(y = 0; y < avctx->height ; y++ ) {
+ uint8_t *row = &s->frame.data[0][y*s->frame.linesize[0]];
for(x = 0; x < avctx->width && buf < buf_end; ) {
int8_t value = *buf++;
unsigned length;