Graphics Performance

Carsten Haitzler (The Rasterman) raster at rasterman.com
Mon Apr 6 18:35:13 CEST 2009


On Thu, 02 Apr 2009 21:19:45 -0400 "Iain B. FIndleton"
<ifindleton at videotron.ca> said:

> Well, that clears things up a bit. So, there is no way to get rid of the 
> draping one sees when the display is refreshed? My stuff uses double 
> buffering, but your comments appear to indicate that that is a waste of  
> time.

none. there is no vblank interrupt so you cant do "have front and back buffer
in fullscreen mode and just swap fb pointers on vblank" and writes to the
frontbuffer are slower than the refresh so you will "watch" it draw as it takes
longer to copy to screen than refresh the screen by a large margin (whereas on
most other gfx systems ever since.. it has been faster to write than refresh -
so you can wait for vblank then quickly write knowing you will stay ahead of
the rescan).

> Carsten Haitzler (The Rasterman) wrote:
> > On Thu, 2 Apr 2009 18:23:38 +0200 Tobias Diedrich <ranma
> > +openmoko at tdiedrich.de> said:
> >
> >   
> >> Iain B. FIndleton wrote:
> >>     
> >>> A significant issue for me is the performance of the graphics display on 
> >>> the FR. I recall some discussions a while back about making use of the 
> >>> XGlamo acceleration features. Has any progress been made here? It 
> >>> appears to me that the graphics performance on the FR is poor compared 
> >>> to, for instance, the iPhone or iTouch, both of which have slower CPUs. 
> >>> When applications running on the FR have their X output routed to a 
> >>> machine with accelerated graphics, it is apparent that the FR processor 
> >>> can deliver the X events fast enough, but the FR graphics chip interface 
> >>> can't keep up.
> >>>       
> >> Isn't the glamo supposed to have one (or more?) OpenRISC cores?
> >> It would be nice to have a documented way to upload code to the
> >> core, that way it might be possible to implement the Bling on the
> >> graphics chip directly...
> >> I mean, since OpenRISC has a documented instruction set (unless
> >> they've augmented it) set I'd figure the only thing missing would
> >>  be where to put the code and how to start it...
> >>     
> >
> > this information is not even in the docs openmoko had on the glamo. there
> > is no known way to play with this core. my understanding is that it is
> > actually a relatively slow core (50mhz) and is only really for higher level
> > management of sub-systems on the glamo.
> >
> > of course here is your big problem.. you can do all this for the glamo and
> > it will never work anywhere else. it is a 1 off for 1 chip that will never
> > see the light of day in another product.
> >
> >   
> >> So, just like with the mpeg4 decoding unit, wouldn't it be
> >> possible for someone with access to the NDA documentation to write
> >> an example program that just shows how to run a simple program (e.g.
> >> bitblt) on the OpenRISC processor?
> >>     
> >
> > no. as those docs are not even in the nda docs. other than that.. bitblit is
> > documented and not related to the risc core. there is a blitter there.
> > xglamo uses even. xglamo *IS ACCELERATED* it's about as accelerated as most
> > x drivers (fills, blits). it has no accel for xrender (xglamo doesnt
> > implement enough of xrender's operations to make it worth it - again see my
> > previous mail. you'll be writing fallback software code and end up no
> > faster than where you started).
> >
> > if you want decent speed - drop to qvga. thats what glamo was really
> > designed to handle. even the 2442 (cpu) is pushing it to deal with vga
> > nicely. it can. but that generation of cpu is more geared to qvga
> > resolutions.
> >
> > the gta02 is a ferrari body (vga screen) with a lawnmower engine under it
> > (2442 +glamo). you need to drive it like a lawnmower - and then only expect
> > it to be as good as a lownmower. it looks nice parked on the street (still
> > photos) but if it moves... it will show its true nature. remove the heavy
> > ferrari body and drive it like a go-kart and you'll have more fun.
> >
> >   
> >> -- 
> >> Tobias						PGP:
> >> http://9ac7e0bc.uguu.de
> >>
> >> _______________________________________________
> >> Openmoko community mailing list
> >> community at lists.openmoko.org
> >> http://lists.openmoko.org/mailman/listinfo/community
> >>
> >>     
> >
> >
> >   
> 
> 
> _______________________________________________
> Openmoko community mailing list
> community at lists.openmoko.org
> http://lists.openmoko.org/mailman/listinfo/community
> 


-- 
------------- Codito, ergo sum - "I code, therefore I am" --------------
The Rasterman (Carsten Haitzler)    raster at rasterman.com





More information about the community mailing list