[Bf-python] svg2obj script.
Martin Poirier
theeth at yahoo.com
Thu Mar 29 19:49:22 CEST 2007
--- jmsoler at free.fr wrote:
> Selon Martin Poirier <theeth at yahoo.com>:
>
> > You're sure it's that single if that causes the
> slow
> > down?
> >
>
> Nothing else was changed.
Heh, when you mentionned a slow down, I though you
meant something more than a 1% slow down on a size 10
sample with more than 1% variation.
> > I did some benchmarks with different size strings
> and
> > the first version (with the split find) is always
> > faster.
> >
> > But even on large strings, the execution time
> > difference was between 3.60951423645e-06 and
> > 5.2826166153e-06 seconds.
> >
> > Martin
>
> I did the tests in the script itself with a 800kos
> svg file
> exported from Adobe Illustrator.
>
> With Blender 2.43 CVS, win32 XP pro, python 2.5,
> imported objects remain in Blender'memory space.
>
> 10 Times without this modif :
> elapsed time : 3.02794895687
> elapsed time : 3.02382840209
> elapsed time : 3.0483079298
> elapsed time : 3.03878685169
> elapsed time : 3.08494582747
> elapsed time : 3.04972026395
> elapsed time : 3.13402334144
> elapsed time : 3.15297461674
> elapsed time : 3.05711943378
> elapsed time : 3.13330760271
>
> Average on 10 times = 3.07509
> Average on first 5 times = 3.04474
>
> 10 Time with this modif :
> elapsed time : 3.10750422266
> elapsed time : 3.07682870892
> elapsed time : 3.01555719385
> elapsed time : 3.0640956465
> elapsed time : 3.06022679437
> elapsed time : 3.02146382774
> elapsed time : 3.12042855888
> elapsed time : 3.11457905378
> elapsed time : 3.06329264794
> elapsed time : 3.05221140016
>
> Average on 10 times = 3.06961, faster but you have
> to load the same file 3
> times before to gain something .
> Average on 5 first times = 3.06483, slower ...
>
> New tests with official Blender 2.43, win32 XP pro,
> python 2.4,
> imported objects reamain in Blender's memory space.
>
> 10 Times without this modif :
> elapsed time : 2.55757148982
> elapsed time : 2.54529198096
> elapsed time : 2.55689776832
> elapsed time : 2.57920924172
> elapsed time : 2.5708110137
> elapsed time : 2.58500159065
> elapsed time : 2.62151425544
> elapsed time : 2.64284499019
> elapsed time : 2.58333532894
> elapsed time : 2.63685119103
>
> Average = 2.5879
>
> 10 Time with this modif:
> elapsed time : 2.54051702967
> elapsed time : 2.52347233528
> elapsed time : 2.5349952355
> elapsed time : 2.5584814291
> elapsed time : 2.57258579398
> elapsed time : 2.576736104
> elapsed time : 2.6136939851
> elapsed time : 2.61544180995
> elapsed time : 2.5731474679
> elapsed time : 2.66417633218
>
> Average = 2.5773
The difference is not very meaningful, it is less than
one standard deviation away in both cases. (I'm not
going to enter deeper into statistical theories, but
with a small sample like this, anything can happen on
both sides)
Frankly, on this one, I'd say do what you want, but I
really can't understand how you can figure that doing
the lookup twice is better than once, even if the
difference is minimal / non meaningful.
Martin
____________________________________________________________________________________
Looking for earth-friendly autos?
Browse Top Cars by "Green Rating" at Yahoo! Autos' Green Center.
http://autos.yahoo.com/green_center/
More information about the Bf-python
mailing list