hi,

"Srihari" <sri### [at] yahoocom> wrote:
> ...
> Based on your inputs I tried to run with and without +WT1 option and check how
> many no of threads of running
>
> I could see that multiple threads were getting opened. But with +WT1 option one
> thread was having like >12% CPU utilisation and others less than < 0.01.
> Meanwhile with +I options I got two threads > 12% utilisation and others < 0.01

disclaimer - not an "expert" in things POV-Ray internals.  when you look at the
(text) output, you can see the various stages.  first parsing, that is always
one thread "working", next stage stuff I know little about like photons +
radiosity, but, as you observed, multiple threads involved.  last the actual
render, and you tell POV-Ray to use 'wt' threads at most to do that.  the
remaining threads are (the "main" thread aside) to service the console IO etc.

in a typical scene where more than just a test render is done, you'll see CPU
utilisation "at capacity" unless you use options; so eg for an old Core2 (== 4
h/ware threads), I see 397/398%, solid, until POV's done.


> Assuming that the CPU threads refers to the trace time, is there any similar
> ways to minimise number of threads or impact of threads on photon time and
> radiosity time. Screenshots of CPU Utilisation are attached for reference

not that I'm aware of, but hope someone in-the-know will butt in.


regards, jr.

Post a reply to this message