POV-Ray : Newsgroups : povray.general : PS3 Folding@Home stats show it may be an interesting Pov platform : Re: PS3 Folding@Home stats show it may be an interesting Pov platform Server Time
31 Jul 2024 12:19:16 EDT (-0400)
  Re: PS3 Folding@Home stats show it may be an interesting Pov platform  
From: Sabrina Kilian
Date: 25 Mar 2007 19:16:07
Message: <460710c7$1@news.povray.org>
Andrew Price wrote:
> Hi folks!
> 
> I was just looking at this page:
>
http://gizmodo.com/gadgets/home-entertainment/breaking-ps3-folding-ps3-triples-folding-at-homes-computing-power-to-over
> -500-tflopspflops-in-spitting-range-246664.php
> 
> It would seem that each PS3 is contributing the same CPU horsepower as 24.6
> pcs on average.
> 
> Since this task is similar in resource needs to a pov render -- mostly
> needing CPU and some memory, with very little IO; and since pov rendering
> is also highly parallelalizable (sp?) it seems to me that a native PS3 port
> of pov might be highly desirable around here.
> 
> What do others think? Is there some way I can help make such a thing come
> into being?
> 
> Cheers
> 
> Andy
> 
> 

24.6 'average' Windows PCs. What's the average PC that is running
Folding@Home? All it would take to throw that average a bit is a few
people with free electricity and a Pentium Pros they don't mind keeping
around. The PS3 has a nice advantage of being pretty uniform, so their
average is affected more by how many people are playing games then which
people are running on slower hardware. I'm also not sure about the PS3,
but I know PCs are able to run BOINC (SETI@Home and others) while
running other software, which would also result in lower scores for
active PCs. If the PS3 only runs Folding@Home while it is not playing
games, that would be another thing to boost it's apparent rating.

As for POV-Ray, it might not even be worth porting to the PS3. Quoting
wiki on the cell arch:

"The PPE's VMX (AltiVec) unit is fully pipelined for double precision
floating point and each SPU can complete two double precision operations
per clock cycle, which translates to 6.4 GFLOPS at 3.2 GHz"

That maximum doesn't mention anything about it's actual speed, just the
theoretical top end. Also from wiki:

"PlayStation 3's Cell CPU achieves 204 GFLOPS single precision float and
15 GFLOPS double precision"

That sounds a lot more reasonable, when you count in how many clock
cycles multiplying really takes.

15 GLOPS is still a good speed, but I don't know that porting POV-Ray to
the PS3 is going to happen before some quad core desktop matches or
passes that speed. If gcc gets to the point that it can take SMP POV-Ray
(beta or 4.0 eventually) and convert that to the Cell's unique
abilities, great. I'm putting my money on the x86 though.


Post a reply to this message

Copyright 2003-2023 Persistence of Vision Raytracer Pty. Ltd.