[directfb-users] (video over SDL over DirectFB) VERSUS (video over DirectFB)

SDL Development
1

Currently all blits to/from system surfaces will be done using the CPU. I
did some testing with AGP surfaces which would help, but nothing
production quality.

Why don’t you try something like this:

  • capture into a system surface
  • process the image
  • Blit into a video surface (same size as the system surface)
    Avoids using the CPU to scale the image.
  • StretchBlit into the primary video surface. Nice and fast.

That is what I’m currently doing (well except I’m not capturing anything).
In my case I can also clip the blit from system to video mem. I haven’t
actually benchmarked it but I would think it’s a good approach.On Sat, Jun 22, 2002 at 01:52:07PM -0400, Bilal Khan wrote:

Hi,

–
Ville Syrj?l?
@Ville_Syrjala
http://www.sci.fi/~syrjala/

2

Quoting Ville Syrj?l? (syrjala at sci.fi):> On Sat, Jun 22, 2002 at 01:52:07PM -0400, Bilal Khan wrote:

Hi,

Currently all blits to/from system surfaces will be done using the CPU. I
did some testing with AGP surfaces which would help, but nothing
production quality.

Why don’t you try something like this:

  • capture into a system surface
  • process the image
  • Blit into a video surface (same size as the system surface)
    Avoids using the CPU to scale the image.
  • StretchBlit into the primary video surface. Nice and fast.

That’s the best approach currently. With AGP it would look like:

  • capture into a auto-video surface
  • process the image
  • StretchBlit into the primary video surface. Nice and fast.

Because of the skipped copy (blit) from system to video memory
the frame rate increases much, I’m really looking forward to it.

–
Best regards,
Denis Oliver Kropp

.------------------------------------------.
| DirectFB - Hardware accelerated graphics |
| http://www.directfb.org/ |
"------------------------------------------"

                        Convergence GmbH