POV-Ray : Newsgroups : povray.off-topic : solid assignment Server Time
4 Nov 2024 17:36:32 EST (-0500)
  solid assignment (Message 1 to 5 of 5)  
From: Meothuru
Subject: solid assignment
Date: 4 Mar 2008 01:00:01
Message: <web.47cce4aa5d6af01e2a42a77a0@news.povray.org>
Currently I write a simple tool, which allows to split scenes in up
to 16 segments and let them render parallel(See the snapshot below).
After all segments has been rendered, the segments are automatic
build up to the general view.


In the past I wrote a similar tool for animations...so this is *not*
the problem to realize  it.

My problem is:
The program must check all running tasks. This is no problem, as long only
one scene is rendered. But I sometimes rendering more the one scene/animation
parallel. And so with this "Splitter"-tool something like this
will be "normal".

Example:

 scene 1  - running with 2 tasks  (PID 1202 and PID 1203)
 sence 2 -  runnning with 1 task  (PID 2078)
 scene 3 - running with 4 tasks    (PID's 3207 .. 3210 )
 animation - running with 2 tasks   PID's  3921 and 3922)
 ...


and all this is running at the same time.
To read out the PID of a task is no problem, but the method to identify and
assign a specific task to a specific scene in a *solid* way...this is my
problem.

So my question:
How can assign in a *solid* way, the  tasks to a specific scene/animation ?
Or does anybody know, where I can finde more infos (links, books) ?



btw.
I'm using LINUX


Post a reply to this message


Attachments:
Download 'snapshot.png' (34 KB)

Preview of image 'snapshot.png'
snapshot.png


 

From: Warp
Subject: Re: solid assignment
Date: 4 Mar 2008 03:59:17
Message: <47cd0f65@news.povray.org>
Meothuru <nomail@nomail> wrote:
> Currently I write a simple tool, which allows to split scenes in up
> to 16 segments and let them render parallel(See the snapshot below).
> After all segments has been rendered, the segments are automatic
> build up to the general view.

  I assume that means that rendering the scene takes 16 times as much
memory as usual?

  Isn't this kind of obsolete, as the new POV-Ray 3.7 supports parallel
rendering (and without taking any additional memory for that)?

-- 
                                                          - Warp


Post a reply to this message

From: Meothuru
Subject: Re: solid assignment
Date: 5 Mar 2008 00:55:00
Message: <web.47ce3504bb3159a22a42a77a0@news.povray.org>
Warp <war### [at] tagpovrayorg> wrote:

>
>   I assume that means that rendering the scene takes 16 times as much
> memory as usual?

May be on Windows...but not on LINUX.
I never have seen more than 1.5 GB RAM in usage (AFAIR running 32 task)



>   Isn't this kind of obsolete, as the new POV-Ray 3.7 supports parallel
> rendering (and without taking any additional memory for that)?
>

No it isn't.


Post a reply to this message

From: Jim Henderson
Subject: Re: solid assignment
Date: 5 Mar 2008 00:59:37
Message: <47ce36c9$1@news.povray.org>
On Tue, 04 Mar 2008 03:59:17 -0500, Warp wrote:

>   Isn't this kind of obsolete, as the new POV-Ray 3.7 supports parallel
> rendering (and without taking any additional memory for that)?

Not if you have multiple machines you can use for rendering. :-)  True 
that the tool doesn't seem to have that functionality yet, but that would 
be my first enhancement request - a job control system that let you farm 
the job out to multiple networked machines.  (I know that's not a new 
idea, though).

If you run in an environment that supports true clustered computing, 
though, I could see this being helpful even without a job control 
system.  I want to say Mepis provides this functionality (clustered 
processing), but I might be thinking of something else.

Jim


Post a reply to this message

From: Warp
Subject: Re: solid assignment
Date: 5 Mar 2008 03:01:32
Message: <47ce535b@news.povray.org>
Meothuru <nomail@nomail> wrote:
> >   I assume that means that rendering the scene takes 16 times as much
> > memory as usual?

> May be on Windows...but not on LINUX.
> I never have seen more than 1.5 GB RAM in usage (AFAIR running 32 task)

  Well, if rendering a particular scene with one instance of POV-Ray
requires 44 MB of memory, rendering with 32 instances requires 1.5 GB.
It's not a question of how much the total memory usage is, but how much
it is compared to running only one instance.
  POV-Ray 3.7 can render that same scene using 32 threads and it will
only require the 44 MB of memory.

  Or are you saying that you can run 32 instances of POV-Ray 3.6 without
it requiring 32 times as much memory as running 1 instance? Exactly how
does it achieve that feat?

> >   Isn't this kind of obsolete, as the new POV-Ray 3.7 supports parallel
> > rendering (and without taking any additional memory for that)?

> No it isn't.

  What does it do that POV-Ray 3.7 doesn't (besides consuming more memory)?

-- 
                                                          - Warp


Post a reply to this message

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