 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Thorsten,
Here is what I see for optics.pov :
41.4% ==> gatherPhotonsRec
21.1% ==> sqrt
17.9% ==> counting routine for profiling
5.6% ==> fixDown
2.6% ==> fixUp
2.2% ==> more profiling overhead
2.1% ==> scattering_attenuation
1.5% ==> photonRgbe2colour
1.1% ==> PQDelMax
The rest is all below 1%
Bill P.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
William F. Pokorny <pok### [at] attglobal net> wrote:
> 41.4% ==> gatherPhotonsRec
I wonder if that happens at parse time or at rendering time. If it happens
at rendering time, then it may be a bit alarming.
--
#macro N(D)#if(D>99)cylinder{M()#local D=div(D,104);M().5,2pigment{rgb M()}}
N(D)#end#end#macro M()<mod(D,13)-6mod(div(D,13)8)-3,10>#end blob{
N(11117333955)N(4254934330)N(3900569407)N(7382340)N(3358)N(970)}// - Warp -
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
In article <3D735482.B94530A4@attglobal.net> , "William F. Pokorny"
<pok### [at] attglobalnet> wrote:
> Thorsten,
> Here is what I see for optics.pov :
>
> 41.4% ==> gatherPhotonsRec
> 21.1% ==> sqrt
> 17.9% ==> counting routine for profiling
> 5.6% ==> fixDown
> 2.6% ==> fixUp
> 2.2% ==> more profiling overhead
> 2.1% ==> scattering_attenuation
> 1.5% ==> photonRgbe2colour
> 1.1% ==> PQDelMax
Thank you very much! This is very useful indeed.
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
|
|
| |
| |
|
|
|
|
| |
| |
|
|
In article <3d73558d@news.povray.org> , Warp <war### [at] tagpovrayorg> wrote:
> I wonder if that happens at parse time or at rendering time. If it happens
> at rendering time, then it may be a bit alarming.
At render time. The trick is to remove the square root from each recursive
call. This can be done by calculating it in PQInsert and making sqrt_dmax_s
a global variable just like dmax_s. This is because dmax_s seems to only
change in PQInsert as far as I can tell...
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
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Thorsten, I am happy to be of help. Bill P.
> Thank you very much! This is very useful indeed.
>
> Thorsten
>
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Thorsten,
I got around to trying this last night and after figuring out that we must catch
places in PQDelMax and gatherPhotons in addition to PQInsert I found it
cuts the run time for optics.pov almost exactly in half.
One very interesting thing to me I'll mention too. I started by wondering
what would happen just by making sqrt_dmax_s global. It turns out both
the gcc and vac compilers did really well and were able to optimize within
about 5% of the recoded version of photons.cpp!!!
Bill P.
>
> At render time. The trick is to remove the square root from each recursive
> call. This can be done by calculating it in PQInsert and making sqrt_dmax_s
> a global variable just like dmax_s. This is because dmax_s seems to only
> change in PQInsert as far as I can tell...
> ________________________________________
> Thorsten Froehlich, Duisburg, Germany
>
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
