Thorsten,
Here is what I see for benchmark.pov :

19.3% ==> A profiling counter routine.
15.6% ==> Noise
11.1% ==> DNoise
10.9% ==> __itrunc
6.9%  ==> POVFPU_RunDefault
2.3%  ==> More profiling
1.9%  ==> f_ridged_mf
1.5%  ==> MInvTransPoint
1.3%  ==> Compute_pigment
1.2%  ==> Warp_EPoint
1.0%  ==> sample_media
1.0%  ==> DTurbulence
1.0%  ==> Evaluate_TPat
0.9%  ==> drem.c
0.9%  ==> __start
0.9%  ==> wrinkles_pattern
0.8%  ==> exp.c
0.8%  ==> IsoSurface_Function_Find_Root_R
0.8%  ==> Simulate_Media
0.8%  ==> Check_And_Enqueue
0.8%  ==> Intersect_BBox_Tree
0.8%  ==> sqrt

Given the "Time for Photon" below I expected more photon related routines near
the top??? And to answer Warp's query about optics.pov almost all of the time
was spent in the rendering phase.

Time For Parse:    0 hours  0 minutes  21.0 seconds (21 seconds)
Time For Photon:   0 hours  4 minutes   5.0 seconds (245 seconds)
Time For Trace:    0 hours  8 minutes  51.0 seconds (531 seconds)
    Total Time:    0 hours 13 minutes  17.0 seconds (797 seconds)

Bill P.

Post a reply to this message

"William F. Pokorny" wrote:
> 
> Thorsten,
> Here is what I see for benchmark.pov :
> 
> [...]
> 

Interesting, i would have expected to see crackle_pattern() occur
somewhere, after all it is used for the textures of a lot of objects.

> Given the "Time for Photon" below I expected more photon related routines near
> the top??? And to answer Warp's query about optics.pov almost all of the time
> was spent in the rendering phase.

I suppose most of the time shooting photons is not spent in photon
specific code but in the general tracing routines.  The scene is much less
photons intensive than 'optics.pov' during rendering, try activating
photons on the water and this will probably change...

Christoph

-- 
POV-Ray tutorials, IsoWood include,                 
TransSkin and more: http://www.tu-bs.de/~y0013390/  
Last updated 13 Aug. 2002 _____./\/^>_*_<^\/\.______

Post a reply to this message

Christoph,
crackle_pattern was at 0.1 %
Bill P.

>
> Interesting, i would have expected to see crackle_pattern() occur
> somewhere, after all it is used for the textures of a lot of objects.
>

Post a reply to this message

In article <3D735F82.3E0FD4F0@attglobal.net> , "William F. Pokorny" 
<pok### [at] attglobalnet> wrote:

> Given the "Time for Photon" below I expected more photon related routines near
> the top???

Hmm, not really.  Photon tracing disappears mostly in the other tracing
statistics (calls to Trace) it seems.

    Thorsten

____________________________________________________
Thorsten Froehlich, Duisburg, Germany
e-mail: tho### [at] trfde

Visit POV-Ray on the web: http://mac.povray.org

Post a reply to this message

William F. Pokorny wrote:
> 15.6% ==> Noise
> 11.1% ==> DNoise

What exactly is noise and where is it everywhere used? For some patterns? 
Is it used for 'turbulence' (if not how is that implemented)? 

Can anybody give me some pointers about this rendertime-consuming thing?

Thanks

- Micha

-- 
http://objects.povworld.org - the POV-Ray Objects Collection

Post a reply to this message

Micha Riser wrote:
> 
> What exactly is noise and where is it everywhere used? For some patterns?
> Is it used for 'turbulence' (if not how is that implemented)?
> 
> Can anybody give me some pointers about this rendertime-consuming thing?
> 

It is used for most patterns with random structure (like bozo/bumps,
dents, wrinkles, granite, etc. but not in crackle), in turbulence and in
some functions.  

The benchmark scene is heavily using the noise function, especially
because of the media sky and the RMF isosurface ground.

Christoph

-- 
POV-Ray tutorials, IsoWood include,                 
TransSkin and more: http://www.tu-bs.de/~y0013390/  
Last updated 13 Aug. 2002 _____./\/^>_*_<^\/\.______

Post a reply to this message

In article <3d7613b9@news.povray.org> , Micha Riser <mri### [at] gmxnet>  wrote:

> Can anybody give me some pointers about this rendertime-consuming thing?

If you have the Windows version of the source code you find specific
optimizations for noise on P4s contributed by Intel.  There is very little
left to optimize in the noise functions without changing the result :-(

    Thorsten

____________________________________________________
Thorsten Froehlich, Duisburg, Germany
e-mail: tho### [at] trfde

Visit POV-Ray on the web: http://mac.povray.org

Post a reply to this message

Christoph Hormann wrote:

> 
> It is used for most patterns with random structure (like bozo/bumps,
> dents, wrinkles, granite, etc. but not in crackle), in turbulence and in
> some functions.

Is the idea of noise to provide a 3d-field of random values which change 
smoothly (order 1 or 2 differentability)? I think there has been a link to 
an article posted recently.. I'll search for it.

- Micha

-- 
http://objects.povworld.org - the POV-Ray Objects Collection

Post a reply to this message

Thorsten Froehlich wrote:

> In article <3d7613b9@news.povray.org> , Micha Riser <mri### [at] gmxnet> 
> wrote:
> 
>> Can anybody give me some pointers about this rendertime-consuming thing?
> 
> If you have the Windows version of the source code you find specific
> optimizations for noise on P4s contributed by Intel.  There is very little
> left to optimize in the noise functions without changing the result :-(

This is probably "fasternoise.h" then. Why are these optimisation in the 
Windows version only? Is it assebler code? BTW is the official windows 
binary compiled with icc then?

- Micha


-- 
http://objects.povworld.org - the POV-Ray Objects Collection

Post a reply to this message

Micha Riser wrote:

I have downloaded the windows sources now.

> 
> This is probably "fasternoise.h" then. 

Yep, so is it.

> Why are these optimisation in the
> Windows version only? Is it assebler code? 

I see now, it has some Windows-specific part. With some modification I got 
it to compile with the unix source. Now I wonder if it will actually run on 
a P4.. we will see.


-- 
http://objects.povworld.org - the POV-Ray Objects Collection

Post a reply to this message