uterm: video: greatly enhance performance by special casing

If we special-case the "src == 0" and "src == 0xff" cases, then we can
avoid all the heavy calculations most of the time. This reduces rendering
time by like 50%.

Signed-off-by: David Herrmann <dh.herrmann@googlemail.com>
This commit is contained in:
David Herrmann committed 2012-09-04 17:56:03 +02:00
1 parent 816dcfb148
commit 720d8d8dfd
2 files changed
+162 -76

No files matched your search

+38 -18
View File
@@ -463,15 +463,25 @@ static int display_blend(struct uterm_display *disp,
* speed by like 20% on slower machines.
* Downside is, full white is 254/254/254
* instead of 255/255/255. */
r = fr * src[i] +
br * (255 - src[i]);
r /= 256;
g = fg * src[i] +
bg * (255 - src[i]);
g /= 256;
b = fb * src[i] +
bb * (255 - src[i]);
b /= 256;
if (src[i] == 0) {
r = br;
g = bg;
b = bb;
} else if (src[i] == 255) {
r = fr;
g = fg;
b = fb;
} else {
r = fr * src[i] +
br * (255 - src[i]);
r /= 256;
g = fg * src[i] +
bg * (255 - src[i]);
g /= 256;
b = fb * src[i] +
bb * (255 - src[i]);
b /= 256;
}
((uint32_t*)dst)[i] = (r << 16) | (g << 8) | b;
}
dst += rb->stride;
@@ -533,15 +543,25 @@ static int display_blendv(struct uterm_display *disp,
* speed by like 20% on slower machines.
* Downside is, full white is 254/254/254
* instead of 255/255/255. */
r = req->fr * src[i] +
req->br * (255 - src[i]);
r /= 256;
g = req->fg * src[i] +
req->bg * (255 - src[i]);
g /= 256;
b = req->fb * src[i] +
req->bb * (255 - src[i]);
b /= 256;
if (src[i] == 0) {
r = req->br;
g = req->bg;
b = req->bb;
} else if (src[i] == 255) {
r = req->fr;
g = req->fg;
b = req->fb;
} else {
r = req->fr * src[i] +
req->br * (255 - src[i]);
r /= 256;
g = req->fg * src[i] +
req->bg * (255 - src[i]);
g /= 256;
b = req->fb * src[i] +
req->bb * (255 - src[i]);
b /= 256;
}
((uint32_t*)dst)[i] = (r << 16) | (g << 8) | b;
}
dst += rb->stride;
+124 -58
View File
@@ -542,15 +542,25 @@ static int display_blend(struct uterm_display *disp,
if (disp->fbdev.xrgb32) {
while (height--) {
for (i = 0; i < width; ++i) {
r = fr * src[i] +
br * (255 - src[i]);
r /= 256;
g = fg * src[i] +
bg * (255 - src[i]);
g /= 256;
b = fb * src[i] +
bb * (255 - src[i]);
b /= 256;
if (src[i] == 0) {
r = br;
g = bg;
b = bb;
} else if (src[i] == 255) {
r = fr;
g = fg;
b = fb;
} else {
r = fr * src[i] +
br * (255 - src[i]);
r /= 256;
g = fg * src[i] +
bg * (255 - src[i]);
g /= 256;
b = fb * src[i] +
bb * (255 - src[i]);
b /= 256;
}
val = (r << 16) | (g << 8) | b;
((uint32_t*)dst)[i] = val;
}
@@ -560,16 +570,28 @@ static int display_blend(struct uterm_display *disp,
} else if (disp->fbdev.Bpp == 2) {
while (height--) {
for (i = 0; i < width; ++i) {
r = (fr & 0xff) * src[i] / 255 +
(br & 0xff) * (255 - src[i]) / 255;
g = (fg & 0xff) * src[i] / 255 +
(bg & 0xff) * (255 - src[i]) / 255;
b = (fb & 0xff) * src[i] / 255 +
(bb & 0xff) * (255 - src[i]) / 255;
val = (r & 0xff) << 16;
val |= (g & 0xff) << 8;
val |= (b & 0xff) << 0;
((uint16_t*)dst)[i] = xrgb32_to_device(disp, val);
if (src[i] == 0) {
r = br;
g = bg;
b = bb;
} else if (src[i] == 255) {
r = fr;
g = fg;
b = fb;
} else {
r = fr * src[i] +
br * (255 - src[i]);
r /= 256;
g = fg * src[i] +
bg * (255 - src[i]);
g /= 256;
b = fb * src[i] +
bb * (255 - src[i]);
b /= 256;
}
val = (r << 16) | (g << 8) | b;
((uint16_t*)dst)[i] = xrgb32_to_device(disp,
val);
}
dst += disp->fbdev.stride;
src += buf->stride;
@@ -577,16 +599,28 @@ static int display_blend(struct uterm_display *disp,
} else if (disp->fbdev.Bpp == 4) {
while (height--) {
for (i = 0; i < width; ++i) {
r = (fr & 0xff) * src[i] / 255 +
(br & 0xff) * (255 - src[i]) / 255;
g = (fg & 0xff) * src[i] / 255 +
(bg & 0xff) * (255 - src[i]) / 255;
b = (fb & 0xff) * src[i] / 255 +
(bb & 0xff) * (255 - src[i]) / 255;
val = (r & 0xff) << 16;
val |= (g & 0xff) << 8;
val |= (b & 0xff) << 0;
((uint32_t*)dst)[i] = xrgb32_to_device(disp, val);
if (src[i] == 0) {
r = br;
g = bg;
b = bb;
} else if (src[i] == 255) {
r = fr;
g = fg;
b = fb;
} else {
r = fr * src[i] +
br * (255 - src[i]);
r /= 256;
g = fg * src[i] +
bg * (255 - src[i]);
g /= 256;
b = fb * src[i] +
bb * (255 - src[i]);
b /= 256;
}
val = (r << 16) | (g << 8) | b;
((uint32_t*)dst)[i] = xrgb32_to_device(disp,
val);
}
dst += disp->fbdev.stride;
src += buf->stride;
@@ -646,15 +680,25 @@ static int display_blendv(struct uterm_display *disp,
if (disp->fbdev.xrgb32) {
while (height--) {
for (i = 0; i < width; ++i) {
r = req->fr * src[i] +
req->br * (255 - src[i]);
r /= 256;
g = req->fg * src[i] +
req->bg * (255 - src[i]);
g /= 256;
b = req->fb * src[i] +
req->bb * (255 - src[i]);
b /= 256;
if (src[i] == 0) {
r = req->br;
g = req->bg;
b = req->bb;
} else if (src[i] == 255) {
r = req->fr;
g = req->fg;
b = req->fb;
} else {
r = req->fr * src[i] +
req->br * (255 - src[i]);
r /= 256;
g = req->fg * src[i] +
req->bg * (255 - src[i]);
g /= 256;
b = req->fb * src[i] +
req->bb * (255 - src[i]);
b /= 256;
}
val = (r << 16) | (g << 8) | b;
((uint32_t*)dst)[i] = val;
}
@@ -664,17 +708,28 @@ static int display_blendv(struct uterm_display *disp,
} else if (disp->fbdev.Bpp == 2) {
while (height--) {
for (i = 0; i < width; ++i) {
r = req->fr * src[i] +
req->br * (255 - src[i]);
r /= 256;
g = req->fg * src[i] +
req->bg * (255 - src[i]);
g /= 256;
b = req->fb * src[i] +
req->bb * (255 - src[i]);
b /= 256;
if (src[i] == 0) {
r = req->br;
g = req->bg;
b = req->bb;
} else if (src[i] == 255) {
r = req->fr;
g = req->fg;
b = req->fb;
} else {
r = req->fr * src[i] +
req->br * (255 - src[i]);
r /= 256;
g = req->fg * src[i] +
req->bg * (255 - src[i]);
g /= 256;
b = req->fb * src[i] +
req->bb * (255 - src[i]);
b /= 256;
}
val = (r << 16) | (g << 8) | b;
((uint16_t*)dst)[i] = xrgb32_to_device(disp, val);
((uint16_t*)dst)[i] =
xrgb32_to_device(disp, val);
}
dst += disp->fbdev.stride;
src += req->buf->stride;
@@ -682,17 +737,28 @@ static int display_blendv(struct uterm_display *disp,
} else if (disp->fbdev.Bpp == 4) {
while (height--) {
for (i = 0; i < width; ++i) {
r = req->fr * src[i] +
req->br * (255 - src[i]);
r /= 256;
g = req->fg * src[i] +
req->bg * (255 - src[i]);
g /= 256;
b = req->fb * src[i] +
req->bb * (255 - src[i]);
b /= 256;
if (src[i] == 0) {
r = req->br;
g = req->bg;
b = req->bb;
} else if (src[i] == 255) {
r = req->fr;
g = req->fg;
b = req->fb;
} else {
r = req->fr * src[i] +
req->br * (255 - src[i]);
r /= 256;
g = req->fg * src[i] +
req->bg * (255 - src[i]);
g /= 256;
b = req->fb * src[i] +
req->bb * (255 - src[i]);
b /= 256;
}
val = (r << 16) | (g << 8) | b;
((uint32_t*)dst)[i] = xrgb32_to_device(disp, val);
((uint32_t*)dst)[i] =
xrgb32_to_device(disp, val);
}
dst += disp->fbdev.stride;
src += req->buf->stride;