Asdf Ghjk wrote:
This might be more related to blitting, hardware acceleration and the
Linux framebuffer in general, but as the answers might be interesting
to other people using DirectFB, I thought I'd ask anyway.
1. When your source data is in main memory, what does
hardware-accelerated blitting do differently to achieve higher
performance? If the operation is just to copy data from main memory to
vram, then how could this ever be "accelerated"?
it could use DMA. If you simply use the 'bare' framebuffer device, which
is available via mmap, then the data is always handled by the CPU.
However, in general if you use acceleration, you stick the data in VRAM
and use the video card hardware for moving-blitting-etcetera.
3. Is it possible to do a "real" page-flip on just the framebuffer
device without hardware acceleration, or will this just be implemented
by copying data from one region to another?
We rely on frame buffer ioctls to do this - the default fb driver
creates a buffer twice the size of the display and does an ioctl-pan to
and fro to simulate a flip. If this doesn't work we of course must copy
using memcpy or equivalent.
Thanks!
------------------------------------------------------------------------
_______________________________________________
directfb-users mailing list
[email protected]
http://mail.directfb.org/cgi-bin/mailman/listinfo/directfb-users
--
.------------------------------------------.
| DirectFB - Hardware accelerated graphics |
| http://www.directfb.org/ |
"------------------------------------------"
_______________________________________________
directfb-users mailing list
[email protected]
http://mail.directfb.org/cgi-bin/mailman/listinfo/directfb-users
