If you use LINUX, you can reduce the total rendertime
up to 50%-65% in a very simple way.

First split your animation in two(or more) segments
of Subset_Frames

i.e.
If you want to render an animation with 100 picture, create
two INI-Files.
The first ini-file controls the rendering of the pictures 1-50
the second .INI-File controls the rendering of the pictures 51-100
by setting subset_frames.

Then create a small shellscript with this two commandlines

 povray scene1_0.ini&
 povray scene1_1.ini

and start the script.

The "magic Point" is the "&"-sign at the end of the first
script line. This allow to run more than one instances
of povray *parallel*. Because the instance of povray runs
in a subshell in the background. And this parallel-rendering
brings a reduction of about 50%-65% of total rendertime.

Note!
The parallel-rendering does *NOT* increase the performance
in any way. It only uses CPU-power wich is normaly unused.

Post a reply to this message

Meothuru <nomail@nomail> wrote:
> And this parallel-rendering
> brings a reduction of about 50%-65% of total rendertime.

  And doubles memory consumption. Only feasible if rendering with one
instance takes less than half of your available RAM.

  The new pov3.7 will be able to use all cpus without increased memory
consumption.

-- 
                                                          - Warp

Post a reply to this message

Meothuru wrote:
> If you use LINUX, you can reduce the total rendertime
> up to 50%-65% in a very simple way.
> 

AND dual-code or SMP. Linux alone won't get any boost from running n+1
processes at the same time.

-- 
Eero "Aero" Ahonen
   http://www.zbxt.net
      aer### [at] removethiszbxtnetinvalid

Post a reply to this message

Eero Ahonen <aer### [at] removethiszbxtnetinvalid> wrote:
> AND dual-code or SMP. Linux alone won't get any boost from running n+1
> processes at the same time.

  In Pentium4 there may be a slight boost of about 10-20%. In other
single-core cpus there's probably no boost (unless there's a huge
amount of disk I/O).

-- 
                                                          - Warp

Post a reply to this message

Warp <war### [at] tagpovrayorg> wrote:
> Eero Ahonen <aer### [at] removethiszbxtnetinvalid> wrote:
> > AND dual-code or SMP. Linux alone won't get any boost from running n+1
> > processes at the same time.
>
>   In Pentium4 there may be a slight boost of about 10-20%. In other
> single-core cpus there's probably no boost (unless there's a huge
> amount of disk I/O).
>
> --
>                                                           - Warp

Sorry, I forgot...
I'm using a AMD64 x2 4400+ with 2GB RAM and a LINUX-Kernel
with Multiprocessor-Support.

And this  brings up to 65% ...believe it.

Post a reply to this message

Meothuru wrote:
> 
> Sorry, I forgot...
> I'm using a AMD64 x2 4400+ with 2GB RAM and a LINUX-Kernel
> with Multiprocessor-Support.

DualCore.

> And this  brings up to 65% ...believe it.

Yep, very possible, since AMD X2 processors have 2 cores, which makes
them a kind like a SMP system. One POV-process can utilize one core, so
running 2 processes gives a serial boost.

Like Warp said, Pov3.7 is going to have threading support, so it'll
remove the "need" for two different processes.

-- 
Eero "Aero" Ahonen
   http://www.zbxt.net
      aer### [at] removethiszbxtnetinvalid

Post a reply to this message

Meothuru wrote:
> If you use LINUX, you can reduce the total rendertime
> up to 50%-65% in a very simple way.

You can do the same in Windows BTW, just start two copies of POV and set 
them off rendering.

Post a reply to this message