 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Hey all.
I recently upgraded my CPU from an Athlon XP 1800+ to 3000+. Of course
I wanted to try benchmarking my system using POV-Ray. I had stats for a
few old scenes of mine, so I chose one and rendered it in the latest
version of POV-Ray (3.6.1b). Nice and fast, went from 1:04 down to
0:51. Then I tried rendering it again, and it took 1:14. The next
time, it took 1:25, then 1:34, then 1:41, 1:45. Then it seemed to hover
around that value. When I restart my computer and repeat the process,
it does a similar sequence. The rendering time is always nearly the
same, but the parsing time gradually increases to a saturation point. I
at first thought it might be an issue with the CPU, or memory, or their
operating at different clock frequencies. That doesn't seem to be the
main issue, since it behaves in a similar way on our other computer, a
Dell, too. I also did lots of diagnostic tests, and everything checks
out fine. I think it may be an issue with POV-Ray itself. Is it not
freeing up memory? What's going on in there? It's a fairly simple
little scene. Other scenes where it's mostly rendering and doing little
parsing are a lot more consistent. It doesn't seem to have anything to
do with the random number generation in the scene, because I tried it
without it, too. If anyone can offer some suggestions or answers as to
what is going on, I'd greatly appreciate it.
------------------------------
Ben Scheele
Silmarillion: Ben's World
http://www.benscheele.com
------------------------------
Here's the scene: (I've been rendering it at 320x240, antialias at 0.3)
camera{
location 120*y
look_at 0
angle 40
}
light_source{ <50,300,20> rgb 1 }
background{ rgb .1 }
#declare r1 = seed(150);
#declare r2 = seed(200);
#declare r3 = seed(175);
#declare s = .35;
#declare d = .175;
#declare step = .25;
#declare E = 54;
#declare X = -E;
#declare EndX = E;
#while(X < EndX)
#declare Z = -E;
#declare EndZ = E;
#while(Z < EndZ)
#declare R = sqrt(X*X + Z*Z);
#if (R <= E)
#declare rot = 30*sin(d*R);
box { -s, s
pigment
{rgb<1+sin(d*R)+.3*rand(r1),.3+.3*rand(r2),1-sin(d*R)+.3*rand(r3)>}
finish {phong 0.3 ambient 0.35}
scale <step,1.5*rot+.001,step>
translate <X,0,Z>
rotate rot*y
}
#end
#declare Z = Z + step;
#end
#declare X = X + step;
#end
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Ben Scheele nous apporta ses lumieres en ce 30/03/2006 13:47:
> Hey all.
> I recently upgraded my CPU from an Athlon XP 1800+ to 3000+. Of course
> I wanted to try benchmarking my system using POV-Ray. I had stats for a
> few old scenes of mine, so I chose one and rendered it in the latest
> version of POV-Ray (3.6.1b). Nice and fast, went from 1:04 down to
> 0:51. Then I tried rendering it again, and it took 1:14. The next
> time, it took 1:25, then 1:34, then 1:41, 1:45. Then it seemed to hover
> around that value. When I restart my computer and repeat the process,
> it does a similar sequence. The rendering time is always nearly the
> same, but the parsing time gradually increases to a saturation point. I
> at first thought it might be an issue with the CPU, or memory, or their
> operating at different clock frequencies. That doesn't seem to be the
> main issue, since it behaves in a similar way on our other computer, a
> Dell, too. I also did lots of diagnostic tests, and everything checks
> out fine. I think it may be an issue with POV-Ray itself. Is it not
> freeing up memory? What's going on in there? It's a fairly simple
> little scene. Other scenes where it's mostly rendering and doing little
> parsing are a lot more consistent. It doesn't seem to have anything to
> do with the random number generation in the scene, because I tried it
> without it, too. If anyone can offer some suggestions or answers as to
> what is going on, I'd greatly appreciate it.
>
> ------------------------------
> Ben Scheele
> Silmarillion: Ben's World
> http://www.benscheele.com
> ------------------------------
>
Trying it with the current beta. My results using +b2
First render: Parse=26.978, bounding=15.011, trace=7:21
Second render:parse=26.788, bounding=13.859, trace=6:58
On an aging Athlon tbird 1400 and 512 Mb
--
Alain
-------------------------------------------------
I drive way too fast to worry about cholesterol.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Interesting. Thanks for checking it out. So it rendered faster the
second time? Hmm. Yeah, I remember when it used to take me 6 minutes
to render it. It helps to have a few hundred MB of RAM free for doing
that scene.
-Ben
Alain wrote:
> Ben Scheele nous apporta ses lumieres en ce 30/03/2006 13:47:
>
>> Hey all.
>> I recently upgraded my CPU from an Athlon XP 1800+ to 3000+. Of
>> course I wanted to try benchmarking my system using POV-Ray. I had
>> stats for a few old scenes of mine, so I chose one and rendered it in
>> the latest version of POV-Ray (3.6.1b). Nice and fast, went from 1:04
>> down to 0:51. Then I tried rendering it again, and it took 1:14. The
>> next time, it took 1:25, then 1:34, then 1:41, 1:45. Then it seemed
>> to hover around that value. When I restart my computer and repeat the
>> process, it does a similar sequence. The rendering time is always
>> nearly the same, but the parsing time gradually increases to a
>> saturation point. I at first thought it might be an issue with the
>> CPU, or memory, or their operating at different clock frequencies.
>> That doesn't seem to be the main issue, since it behaves in a similar
>> way on our other computer, a Dell, too. I also did lots of diagnostic
>> tests, and everything checks out fine. I think it may be an issue
>> with POV-Ray itself. Is it not freeing up memory? What's going on in
>> there? It's a fairly simple little scene. Other scenes where it's
>> mostly rendering and doing little parsing are a lot more consistent.
>> It doesn't seem to have anything to do with the random number
>> generation in the scene, because I tried it without it, too. If
>> anyone can offer some suggestions or answers as to what is going on,
>> I'd greatly appreciate it.
>>
>> ------------------------------
>> Ben Scheele
>> Silmarillion: Ben's World
>> http://www.benscheele.com
>> ------------------------------
>>
> Trying it with the current beta. My results using +b2
> First render: Parse=26.978, bounding=15.011, trace=7:21
> Second render:parse=26.788, bounding=13.859, trace=6:58
> On an aging Athlon tbird 1400 and 512 Mb
>
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Hi.
On my Athlon 64 3700+ it takes 40 seconds using a 32bit Debian Linux.
Of course as on Linux there isn't a frontend with POV, it exits
comletely when the render is done, so subsequent renders take the same time.
Manuel
#####################################################################
Total Scene Processing Times
Parse Time: 0 hours 0 minutes 12 seconds (12 seconds)
Render Time: 0 hours 0 minutes 28 seconds (28 seconds)
Total Time: 0 hours 0 minutes 40 seconds (40 seconds)
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Good, more information; thanks Manuel. I tried restarting the program,
and that reset the time back to around my minimum. That wouldn't help
for animations, but it's good to know that that works, and I don't need
to restart my whole computer to get that effect. I still would like to
know what's going on within the program that would cause that, but that
would require insider information, it would seem.
------------------------------
Ben Scheele
Silmarillion: Ben's World
http://www.benscheele.com
Manuel Kasten wrote:
> Hi.
>
> On my Athlon 64 3700+ it takes 40 seconds using a 32bit Debian Linux.
> Of course as on Linux there isn't a frontend with POV, it exits
> comletely when the render is done, so subsequent renders take the same
> time.
>
> Manuel
>
> #####################################################################
> Total Scene Processing Times
> Parse Time: 0 hours 0 minutes 12 seconds (12 seconds)
> Render Time: 0 hours 0 minutes 28 seconds (28 seconds)
> Total Time: 0 hours 0 minutes 40 seconds (40 seconds)
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Ben Scheele wrote:
> to restart my whole computer to get that effect. I still would like to
> know what's going on within the program that would cause that, but that
> would require insider information, it would seem.
I tried this with the current 3.6.1c and cannot replicate the issue. Render
time stayed the same for each render. Given 3.6.1c doesn't fix any issues
related to memory or render time (vs 3.6.1b) I can't explain the cause of
your problem.
-- Chris
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Thanks for looking into this issue as well, Chris. I tried installing
this latest version and it still has that issue for me. Here is a set
of render stats for successive renderings.
Parse + Render = Total
16 + 36 = 0:52
41 + 37 = 1:18
49 + 36 = 1:25
56 + 37 = 1:33
60 + 36 = 1:36
63 + 36 = 1:39
68 + 37 = 1:45
75 + 36 = 1:51
76 + 36 = 1:52
The render time is fairly constant, but the parsing time is increasing
until it reaches a saturation point eventually. That does seem like an
odd behavior, doesn't it? Would you please describe your computer's
setup, Chris (RAM, CPU, etc.)? Thanks.
-Ben
Chris Cason wrote:
> I tried this with the current 3.6.1c and cannot replicate the issue. Render
> time stayed the same for each render. Given 3.6.1c doesn't fix any issues
> related to memory or render time (vs 3.6.1b) I can't explain the cause of
> your problem.
>
> -- Chris
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Ben Scheele wrote:
> The render time is fairly constant, but the parsing time is increasing
> until it reaches a saturation point eventually.
You you perhaps have a virus scanner installed that is interfering with the
reading of files?
Thorsten
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Thorsten Froehlich nous apporta ses lumieres en ce 10/04/2006 14:31:
> Ben Scheele wrote:
>
>> The render time is fairly constant, but the parsing time is increasing
>> until it reaches a saturation point eventually.
>
>
> You you perhaps have a virus scanner installed that is interfering with
> the reading of files?
>
> Thorsten
I don't think that a virus scanner would cause progressively increasing parse time. It
would tend to
increase the first render time, then have a decreased, constant, effect on successive
render. RAM
fragmentation can cause this progressive increasing in parse time.
--
Alain
-------------------------------------------------
Fundamentalism #3: Shit must be born again.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Hey, thanks Thorsten and Alain. I don't use an antivirus program, so
that couldn't be an issue. I researched RAM fragmentation and that is
likely the culprit. Aside from paying $15 for some RAM defragmenter
that may or may not improve performance, is there some tweak that can be
done in POV-Ray to get it to behave properly?
-Ben
Alain wrote:
> Thorsten Froehlich nous apporta ses lumieres en ce 10/04/2006 14:31:
>
>> Ben Scheele wrote:
>>
>>> The render time is fairly constant, but the parsing time is
>>> increasing until it reaches a saturation point eventually.
>>
>>
>>
>> You you perhaps have a virus scanner installed that is interfering
>> with the reading of files?
>>
>> Thorsten
>
> I don't think that a virus scanner would cause progressively increasing
> parse time. It would tend to increase the first render time, then have a
> decreased, constant, effect on successive render. RAM fragmentation can
> cause this progressive increasing in parse time.
>
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
|
|
