POV-Ray : Newsgroups : povray.windows : POVBench - testing disk I/O Server Time
23 Nov 2024 18:53:39 EST (-0500)
  POVBench - testing disk I/O (Message 1 to 2 of 2)  
From: Reuben Pearse
Subject: POVBench - testing disk I/O
Date: 13 Apr 2004 04:45:00
Message: <407ba88c$1@news.povray.org>
Hi all,

We have some new Windows 2000 servers and we want to do a "soak test" on 
the servers (give the some long processing tasks), to check that they 
run OK before using them in a production environment.

Looking at the POVBench site 
(http://www.haveland.com/index.htm?povbench/index.php) all the 
benchmarks seem to be about CPU performance. Are there any long-running 
POV scripts that give the disk a bit of thrashing, as this is the main 
thing that I want to test?


--
Reuben
reu### [at] pearsecouk

The best use of life is love. The best expression of love is time. The 
best time to love is now.


Post a reply to this message

From: Chambers
Subject: Re: POVBench - testing disk I/O
Date: 13 Apr 2004 13:57:53
Message: <407c2a21$2@news.povray.org>
"Reuben Pearse" <reu### [at] pearsecouk> wrote in message
news:407ba88c$1@news.povray.org...
> Hi all,
>
> We have some new Windows 2000 servers and we want to do a "soak test" on
> the servers (give the some long processing tasks), to check that they
> run OK before using them in a production environment.
>
> Looking at the POVBench site
> (http://www.haveland.com/index.htm?povbench/index.php) all the
> benchmarks seem to be about CPU performance. Are there any long-running
> POV scripts that give the disk a bit of thrashing, as this is the main
> thing that I want to test?


It's easy enough to write your own script for it - all you have to do is use
an insane number of objects, until VM kicks in.

Try playing around with the following:

////
#declare MAX_OBJECTS = 10000;
#declare MAX_EXTENT = 10000;
#declare MAX_SIZE = 500;

#declare rnd=seed(0);

#declare i=0;
#while (i < MAX_OBJECTS)
   #declare LOC=<rand(rnd), rand(rnd), rand(rnd)>*MAX_EXTENT*2-MAX_EXTENT;
   #declare SIZ=rand(rnd)*MAX_SIZE;
   sphere {LOC, SIZ pigment {color transmit 1} finish {reflection 1} }
#declare i=i+1;
#end

camera {
   location -z*MAX_EXTENT*3
   look_at 0
}

light_source {MAX_EXTENT*2 color rgb 1}

/////

Using the above settings, increasing the MAX_OBJECTS should cause a linear
increase in memory usage.  A value of 10000, as used above, uses a little
less than 10 megabytes.  So, if your machines have 512MB of RAM, you could
try a setting of 1000000 (to ensure swapping).

-- 
...Chambers
http://www.geocities.com/bdchambers79


Post a reply to this message

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