POV-Ray : Newsgroups : povray.newusers : Linux VS Xp : Re: Linux VS Xp Server Time
29 Jul 2024 12:15:20 EDT (-0400)
  Re: Linux VS Xp  
From: Patrick Elliott
Date: 21 Dec 2005 16:00:23
Message: <MPG.1e137005c397bde2989ea3@news.povray.org>
In article <43a90e97@news.povray.org>, war### [at] tagpovrayorg says...
> Edee <rah### [at] yahoocom> wrote:
> > Is there a significant improvement in rendering time by switching to Linux
> > from XP?
> 
>   No, not really. POV-Ray doesn't do much that would require an efficient OS
> (even the new 3.7, which requires SMP support from the OS, should run ok in
> both systems).
> 
That may not be 100% true with Dual Core systems though. I just read 
this:

http://blogs.pcworld.com/tipsandtweaks/archives/001073.html

Seems if your dumb enough to install SP2, it produces what MS calls, 
"Unexpected Behaviour", on such systems, which can only be partially 
fixed with one of their useless and not 100% publicly available patches 
(you have to ask them for it, so 99% of the people who have to problem 
will never know an incomplete patch for their sudden fucked up 
performance even exists...).

Not an issue for the guy your replying to, but at least some people here 
are bound to run into it and its going to make serious attempts to test 
POV-Ray's speed on dual cores real interesting once optimizations start 
to be made. This makes for a heads up. The OS *might* mess it up badly 
enough that an optimization that works for user A will see thing actually 
go down hill for user B, just because of which lame ass patches are 
installed. And since the patch for SP2 to restore *some* lost function 
doesn't fix "all" of it, we are looking at what may as well be six 
different configurations:

1. Non-Duel core without SP2.
2. Non-Duel core with SP2.
3. Non-Duel core with SP2, incorrently patched.
4. Duel core/processor without SP2.
5. Duel core/processor with SP2.
6. Duel core/processor with SP2, patched with the hotfix.

And that is just now, never mind later when they come up with yet another 
buggy, but maybe in some ways slightly more effective, patch to fix the 
issue. And all of them may screw up benchmarking in a way that makes 
clear determinations based on processor speeds problematic to say the 
least.

Starting to wonder who I know that I can get a copy of VM Ware through, 
so I can use the damn OS when I need to, but not rely on it for 
anything... lol

-- 
void main () {

    call functional_code()
  else
    call crash_windows();
}


Post a reply to this message

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