Looking for models for Boolean operations - Forum Open Cascade Technology

Roman Lygin
Wed, 12/03/2008 - 09:28

Forums:

Hi folks,

I'm looking for a few good shape pairs that demonstrate slow performance of Boolean operations (BRepAlgoAPI). If you have comparative performance data with other CAD systems or 3D kernels on the same models, that would be a good addition.
The models should be complex enough, and a Boolean operation should take 30 seconds or more.

I'd like to try out our performance analysis tools on this scenario.

BOPs were blamed for slow performance on a regular basis on this forum, so models should be easily available ;-).
Please send me 1 or 2 pairs, either as compressed zip to roman dot lygin at gmail dot com, or send me a link to download.

Thank you and looking forward to hearing from you.
Roman

Evgeny Lodyzhehsky
Wed, 12/03/2008 - 16:54

Dear Roman Lygin.

Here is the link

http://rapidshare.com/files/169838995/ab.zip.html

try to do r=cut(a,b)

Regards

Roman Lygin
Wed, 12/03/2008 - 19:11

Evgeny Nicolaich,

Thanks for sending but I can't access rapidshare from the office. Could you please resend via email (the file seems to be small) ?
Thanks.
Roman

Evgeny Lodyzhehsky
Thu, 12/04/2008 - 07:49

Dear Roman Lygin

Sorry, but I do not prefer to use e-mail for the goals like that.

Roman Lygin
Thu, 12/04/2008 - 08:23

No problem. I have downloaded being at home ;-).

Roman Lygin
Sat, 12/06/2008 - 11:10

Thanks to all who sent me models. I have just concluded my experiments and am writing an article for my blog on Open CASCADE (opencascade.blogspot.com). It should be out in a couple of days. So, stay tuned !
Roman

Roman Lygin
Sat, 12/06/2008 - 11:18

Ah, forgot to mention - gained speed up is about 4x-20+x depending on a model... Not something I would have ever hoped for ;-)

---
opencascade.blogspot.com - blog on Open CASCADE

Svetlozar Kostadinov
Sat, 12/06/2008 - 15:53

How do you gain this speed-up? With the Intel software?

Roman Lygin
Sat, 12/06/2008 - 20:28

Yep, I used Intel tools to identify bottlenecks but I did not add any Intel libraries to make OCC faster. It remains self-contained.

---
opencascade.blogspot.com - blog on Open CASCADE

Roman Lygin
Mon, 12/08/2008 - 13:44

Thanks to all who replied and provided the models. There are different paths inside BOPs leading to different bottlenecks, and I seem to cover only one of them (but likely quite frequent).

As promised, results have been posted on my blog. Reference thread on the OCC Forum is here - http://www.opencascade.org/org/forum/thread_15003/

---
opencascade.blogspot.com - my blog on Open CASCADE