POV-Ray : Newsgroups : povray.general : Where to find a comparison of modern ray-tracers (speed > POVray) : Re: Where to find a comparison of modern ray-tracers (speed > POVray) Server Time
31 Jul 2024 04:19:49 EDT (-0400)
  Re: Where to find a comparison of modern ray-tracers (speed > POVray)  
From: Charles C
Date: 4 Nov 2007 18:35:00
Message: <web.472e56308f32c2582869ae640@news.povray.org>
Jim Henderson <nos### [at] nospamcom> wrote:
> On Sun, 04 Nov 2007 11:01:05 -0500, Warp wrote:
>
> > Jan Dvorak <jan### [at] centrumcz> wrote:
> >> trace frames 0,10,20,30... on PC 1, frames 1,11,21... on PC 2 etc.
> >
> >   Wouldn't it be simpler to simply trace eg. frames 0-9 in PC 1, 10-19
> > in PC 2, etc?
>
> Perhaps, but if you have 'n' machines, and render frames 1 through n-1
> simultaneously, then you can have a look to see if the first n frames
> animate properly.  If you do 1-n on one, n+1-2n on the next, then you
> have to wait for the first machine to finish all of its frames before
> combining them into an animation.
>
> Jim


Or, here's another approach:  Let all machines go at once & let SDL create "flag
files" and either render blank (read "easily sorted out") images or real frames
depending on whether a given frame number's "flag file" exists already or not.
 It's an SDL only approach.  It may be crude but it's funtional and all PC's
will finish around the same time.  It basically requires each machine (or
instance of POV-Ray) to have it's own unique output location, and for all
machines (or instances of POV-Ray) to have read/write access to a shared
network directory.

Demo scene for "SimpleNetAnimator":
http://news.povray.org/povray.binaries.scene-files/thread/%3Cweb.4668e8cb265a91e39e4bf5850@news.povray.org%3E/

Or here is "SimpleNetAnimator"  although I think what's in the first link is
actually a little more updated:
http://news.povray.org/./povray.binaries.scene-files/thread/%3Cweb.45b9a29a2c9faeaa9356f54f0%40news.povray.org%3E/

One interesting thing that happens when using this approach is that each
machine's reported pixels-per-second will indicate the average overall
performance of the cluster.

Charles


PS If you consolidate the good frames during processing, it's *very* important
to not consolidate them in one of the output directories.... If you happen to
have that machine crash and then re-start the rendering, it's too easy for that
machine to over-write hundreds of good frames with emty ones! :P


Post a reply to this message

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