Shachar Kaufman wrote:
> gAcquire sets
> gfxs->dst_org[0] = state->dst.addr;
>
> Aop_xy sets
> gfxs->Aop[0] = gfxs->dst_org[0];
>
> ABacc_prepare does
> void *ABstart = D_MALLOC( size * sizeof(GenefxAccumulator) * 2
> + 31 );
>
> So as far as I can see, genefx uses direct virtual address of the
> surfaces in order to read/write data. It usually reads data into an
> internal accumulation buffer which is malloced off the system heap, does
> pixel data processing ("run pipeline") there (at 16bit per component),
> and finally packs it back into the destination surface again with direct
> virtual address access. All this per span.
>
> What you want sounds like EXA pixmap migration in XOrg. Doesn't seem
> like there's an equivalent in DFB.
I had a quick look at the 2001(?) code and it looks like what has been in
DirectFB up til 1.0.x but does not in any way compare to the abstract surface
pool mechanism with the transfers using bridges.
--
Best regards,
Denis Oliver Kropp
.------------------------------------------.
| DirectFB - Hardware accelerated graphics |
| http://www.directfb.org/ |
"------------------------------------------"
_______________________________________________
directfb-users mailing list
directfb-users@directfb.org
http://mail.directfb.org/cgi-bin/mailman/listinfo/directfb-users
