mth/ebiten
1
0
Fork 0
mirror of https://github.com/hajimehoshi/ebiten.git synced 2024-11-10 04:57:26 +01:00
Code Issues Packages Projects Releases Wiki Activity

examples/video: performance optimization by encouraging BCE

Browse Source
This commit is contained in:
Hajime Hoshi 2024-04-12 21:17:50 +09:00
parent c105224705
commit 2bd9472d2c
1 changed files with 12 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
examples/video/mpegplayer.go
View File

@ -159,13 +159,18 @@ func (p *mpegPlayer) updateFrame() error {
w, h := p.mpg.Width(), p.mpg.Height() w, h := p.mpg.Width(), p.mpg.Height()
for j := 0; j < h; j++ { for j := 0; j < h; j++ {
yi := j * img.YStride yi := j * img.YStride
ci := j / 2 * img.CStride ci := (j / 2) * img.CStride
// Create temporary slices to encourage BCE (boundary-checking elimination).
ys := img.Y[yi : yi+w]
cbs := img.Cb[ci : ci+w/2]
crs := img.Cr[ci : ci+w/2]
for i := 0; i < w; i++ { for i := 0; i < w; i++ {
idx := 4 * (i + j*w) idx := 4 * (j*w + i)
p.yCbCrBytes[idx] = img.Y[yi+i] buf := p.yCbCrBytes[idx : idx+3]
p.yCbCrBytes[idx+1] = img.Cb[ci+i/2] buf[0] = ys[i]
p.yCbCrBytes[idx+2] = img.Cr[ci+i/2] buf[1] = cbs[i/2]
p.yCbCrBytes[idx+3] = 0xff buf[2] = crs[i/2]
// p.yCbCrBytes[3] = 0xff is not needed as the shader ignores this part.
} }
} }
@ -174,6 +179,7 @@ func (p *mpegPlayer) updateFrame() error {
// Converting YCbCr to RGB on CPU is slow. Use a shader instead. // Converting YCbCr to RGB on CPU is slow. Use a shader instead.
op := &ebiten.DrawRectShaderOptions{} op := &ebiten.DrawRectShaderOptions{}
op.Images[0] = p.yCbCrImage op.Images[0] = p.yCbCrImage
op.Blend = ebiten.BlendCopy
p.frameImage.DrawRectShader(w, h, p.yCbCrShader, op) p.frameImage.DrawRectShader(w, h, p.yCbCrShader, op)
return nil return nil