@@ -132,29 +132,42 @@ static av_cold int xavs2_init(AVCodecContext *avctx)
static void xavs2_copy_frame_with_shift(xavs2_picture_t *pic, const AVFrame *frame, const int shift_in)
{
- int j, k;
- for (k = 0; k < 3; k++) {
- int i_stride = pic->img.i_stride[k];
- for (j = 0; j < pic->img.i_lines[k]; j++) {
- uint16_t *p_plane = (uint16_t *)&pic->img.img_planes[k][j * i_stride];
- int i;
- uint8_t *p_buffer = frame->data[k] + frame->linesize[k] * j;
- memset(p_plane, 0, i_stride);
- for (i = 0; i < pic->img.i_width[k]; i++) {
- p_plane[i] = p_buffer[i] << shift_in;
+ uint16_t *p_plane;
+ uint8_t *p_buffer;
+ int wIdx;
+ int hIdx;
+ int plane;
+
+ for (plane = 0; plane < 3; ++plane) {
+ p_plane = (uint16_t *)pic->img.img_planes[plane];
+ p_buffer = frame->data[plane];
+ for (hIdx = 0; hIdx < pic->img.i_lines[plane]; ++hIdx) {
+ memset(p_plane, 0, pic->img.i_stride[plane]);
+ for (wIdx = 0; wIdx < pic->img.i_width[plane]; ++wIdx) {
+ p_plane[wIdx] = p_buffer[wIdx] << shift_in;
}
+ p_plane += pic->img.i_stride[plane];
+ p_buffer += frame->linesize[plane];
}
}
}
static void xavs2_copy_frame(xavs2_picture_t *pic, const AVFrame *frame)
{
- int j, k;
- for (k = 0; k < 3; k++) {
- for (j = 0; j < pic->img.i_lines[k]; j++) {
- memcpy( pic->img.img_planes[k] + pic->img.i_stride[k] * j,
- frame->data[k]+frame->linesize[k] * j,
- pic->img.i_width[k] * pic->img.in_sample_size);
+ uint8_t *p_plane;
+ uint8_t *p_buffer;
+ int hIdx;
+ int plane;
+ int stride;
+
+ for (plane = 0; plane < 3; ++plane) {
+ p_plane = pic->img.img_planes[plane];
+ p_buffer = frame->data[plane];
+ stride = pic->img.i_width[plane] * pic->img.in_sample_size;
+ for (hIdx = 0; hIdx < pic->img.i_lines[plane]; ++hIdx) {
+ memcpy(p_plane, p_buffer, stride);
+ p_plane += pic->img.i_stride[plane];
+ p_buffer += frame->linesize[plane];
}
}
}
Optimize data access from multiplication to iteration. Signed-off-by: hwren <hwrenx@126.com> --- libavcodec/libxavs2.c | 45 +++++++++++++++++++++++++++++---------------- 1 file changed, 29 insertions(+), 16 deletions(-)