Jim Henderson <nos### [at] nospamcom> wrote:
> I think that would depend on the comparitive niceness levels, though,
> wouldn't it?

  If two cpu-intensive processes have the same nice level, then the OS
will share the CPU among them about equally.
  Naturally if one of the processes has a higher nice value then the OS
will give less CPU to it and more to the other (I'm not exactly sure
about the formula the OS uses for that; could even depend on the OS;
it would be interesting to know what the linux kernel does).

  However, I got the impression from the original claim that if the
OS would run POV-Ray at maximum efficiency (ie. giving it almost 100%
of CPU time) then it would somehow be difficult for anything else to
do anything, and that everything else would go really slow. I got the
impression that he thought that if POV-Ray has (almost) 100% of CPU,
it would have something like "hijacked" the entire computer to itself
and that it would be difficult to get the resources back.
  But of course that's not the case in Unix systems (nor in Windows).
The task scheduler has full control of the machine and nothing can go
over it. It will always schedule tasks at regular intervals (I think
that in a typical linux OS it does so each millisecond) and there's
nothing a program can do about it. This means in practice that the
task scheduler can for example completely stop a program from running
if it wants (and it doesn't matter if that program was using 99.9% of
CPU or whatever). The task scheduler has complete control. It cannot
be "hijacked" by any program in any way.

-- 
                                                          - Warp

Post a reply to this message

you all run on Linux. I own a pentium 4 dual core Acer and currently running
3.61 on XP. Using 3.7 renders slower then the 3.61 version.
- Does it make a difference in speed if I use the Windows version (on XP) or
if I run it on Linux ?
- How can I check on Windows XP if the second core is indeed used ?

thanx

Adrie

Post a reply to this message

bagger <adr### [at] asmlcom> wrote:
> - How can I check on Windows XP if the second core is indeed used ?

  Just open the task manager (ctrl-alt-del) and go to the processor tab
(or whatever it was called).

-- 
                                                          - Warp

Post a reply to this message

Warp <war### [at] tagpovrayorg> wrote:
> bagger <adr### [at] asmlcom> wrote:
> > - How can I check on Windows XP if the second core is indeed used ?
>
>   Just open the task manager (ctrl-alt-del) and go to the processor tab
> (or whatever it was called).
>
> --
>                                                           - Warp

that only gives the percentage of CPU used. Does a 100% garantuee me that
both cores are used ? (by the way does every core have a seperate FPU ?)

Does anyone know if the Linux variant is faster then the Windows XP variant
?

Post a reply to this message

bagger <a.b### [at] chellonl> wrote:
> that only gives the percentage of CPU used. Does a 100% garantuee me that
> both cores are used ? (by the way does every core have a seperate FPU ?)

  Windows should show *two* graphs if you have two CPUs (or two cores,
which is the same thing).
  Are you only seeing one?

-- 
                                                          - Warp

Post a reply to this message

On 26 Jan 2007 10:30:12 -0500, Warp <war### [at] tagpovrayorg> wrote:

>  Windows should show *two* graphs if you have two CPUs (or two cores,
>which is the same thing).

Only if you have "One Grap Per CPU" selected in task manager, otherwise it will show
one graph for all CPUs.



Kyle

Post a reply to this message

Kyle <hob### [at] gatenet> wrote:
> Only if you have "One Grap Per CPU" selected in task manager, otherwise it will show
one graph for all CPUs.

  Well, if that's the case, I think the next step is quite obvious...

-- 
                                                          - Warp

Post a reply to this message

In my experience task-manager _can_ err in the other direction - e.g. a
single-core "hyperthreaded" shows up as two (virtual) cpu's so task-manager
 reports 50% overall usage while in reality the usage is closer to 100%.

As for 3.6 versus 3.7, what are you rendering? Meshes?  At this stage and
for my needs I get about twice the performance from multiple instances of
3.6 compared with running 3.7. Some results I got are here:

http://news.povray.org/povray.general/thread/%3Cweb.45a68007797c8d649356f54f0%40news.povray.org%3E/

Charles

Post a reply to this message

On 26 Jan 2007 11:12:37 -0500, Warp <war### [at] tagpovrayorg> wrote:

>Kyle <hob### [at] gatenet> wrote:
>> Only if you have "One Grap Per CPU" selected in task manager, otherwise it will
show one graph for all CPUs.
>
>  Well, if that's the case, I think the next step is quite obvious...

Indeed, it should be.  I only mentioned it for bagger to check while he's in task
manager if he only
sees one graph.

Also, what if you aren't using the SMP "driver" in XP?  I imagine that would also
cause only one
graph to be displayed.


Kyle

Post a reply to this message

bagger wrote:
  > that only gives the percentage of CPU used. Does a 100% garantuee me 
that
> both cores are used ? (by the way does every core have a seperate FPU ?)

IIRC, it should show 50% if only one of two cores is used - however, I'm 
not running a dual core machine, so take my advice with a grain of salt.

Yes, every core has a separate FPU.

> 
> Does anyone know if the Linux variant is faster then the Windows XP variant
> ?

For what POV-Ray is doing, the choice is OS is (almost) irrelevant.

...Chambers

Post a reply to this message