Sorry your browser is not supported!

You are using an outdated browser that does not support modern web technologies, in order to use this site please update to a new browser.

Browsers supported include Chrome, FireFox, Safari, Opera, Internet Explorer 10+ or Microsoft Edge.

Program Announcements / DBP/DBC Collision DLL - FREE

Author
Message
nuclear glory
21
Years of Service
User Offline
Joined: 2nd Oct 2003
Location:
Posted: 13th Mar 2006 21:58
Link
Yes, and shave off about anywhere from 60% to 80% of our load if done right, lol.

Powerful Collision DLL for DBPro and DBC: http://www.nuclearglory.com
Back to top
Profile PM Email Website
Pulsar Coder
20
Years of Service
User Offline
Joined: 3rd Jan 2004
Location:
Posted: 13th Mar 2006 22:03
Link
ok, but are you re-arranging the 8 childnodes in a binary way or re-calculating everything with a partition plain -like BSP?

Take into account that the binary tree should be balanced in order to get an increase in performance.
Back to top
Profile PM Email Website
Cellbloc Studios
20
Years of Service
User Offline
Joined: 15th Mar 2004
Location: Atlanta, GA
Posted: 13th Mar 2006 22:04
Link
@NG:

So are you working on a 4.02?

-This...is my boomstick!
Support your local Riker 9 Chapter.
Back to top
Profile PM Website
Paul Johnston
TGC Developer
22
Years of Service
User Offline
Joined: 16th Nov 2002
Location: United Kingdom
Posted: 13th Mar 2006 22:15
Link
Quote: "ok, but are you re-arranging the 8 childnodes in a binary way or re-calculating everything with a partition plain -like BSP?"


It's like BSP, attempting to sucessively cut the data set in half.

Back to top
Profile PM Email
nuclear glory
21
Years of Service
User Offline
Joined: 2nd Oct 2003
Location:
Posted: 14th Mar 2006 04:43
Link
Relax I'm doing a dirty test right now (yes, for a v4.02 release). Lost in thought has already confirmed a 7.25 ms speed on my optimized ray routine (before the trees I'm working on now).

Powerful Collision DLL for DBPro and DBC: http://www.nuclearglory.com
Back to top
Profile PM Email Website
Lost in Thought
20
Years of Service
User Offline
Joined: 4th Feb 2004
Location: U.S.A. : Douglas, Georgia
Posted: 16th Mar 2006 13:06
Link
New test ran with beta Matthew (nuclear glory) sent me are really close. Here are the averages per cast out of 81 samples on a 130050 poly object in differnt positions and directions. Each test used the same sets of positions and directions. A 3 by 3 point grid evenly spaced on the object and ran rays from each point's x and z to each other in all combinations (even straight down) making a 3D web mesh pattern through the object. Shoting from 10 units over the highest point in the object to 10 units below the lowest point in the object.

Sparky's : 0.508 ms
NGC 4.02 Beta : 0.783 ms
MGC 2.03 : 22.271ms

NGC 4.02 beta was faster than Sparky's on some samples and Sparky's fatser than NGC 4.02 beta on others. If the sample were shifted a bit the times may have shifted the other way they were so close on some. Looking really good.

As Sparky said :
Quote: "the users are the real winners"


Back to top
Profile PM Email
nuclear glory
21
Years of Service
User Offline
Joined: 2nd Oct 2003
Location:
Posted: 16th Mar 2006 14:23
Link
I just sent Lost in Thought a new version. So, we'll see if that changes anything. I'm seeing a significant decrease in load on the latest v4.02 beta (considering I just rewrote it again, lol)

Powerful Collision DLL for DBPro and DBC: http://www.nuclearglory.com
Back to top
Profile PM Email Website
Lost in Thought
20
Years of Service
User Offline
Joined: 4th Feb 2004
Location: U.S.A. : Douglas, Georgia
Posted: 16th Mar 2006 15:07
Link
0.562 ms for new 4.02 beta average with those settings and about 547 for Sparky's (my computer must be slowing down after running so many tests)

Sparkys test


NGC 4.02 beta test


You can copy them in 2 text windows to compare each test side by side as they were ran in the same order with all the same end and start points for each ray.

Back to top
Profile PM Email
Lost in Thought
20
Years of Service
User Offline
Joined: 4th Feb 2004
Location: U.S.A. : Douglas, Georgia
Posted: 16th Mar 2006 15:18
Link
Keeps locking up when editing that last post so... Just found out the threshold was locked at 1000 with the latest 4.02 beta (was using 450 beore). When setting Sparky's to 1000 it averages 1.1ms per cast compared to NGC's 0.562. I'll have to wait a bit before setting them both back to 450. I use 450 as there are 450 polys per limb and the trees would be more even.

Back to top
Profile PM Email
nuclear glory
21
Years of Service
User Offline
Joined: 2nd Oct 2003
Location:
Posted: 16th Mar 2006 23:43
Link
That's better.

Okay, everyone who lost faith in NG raise your hands high so I can see you.

Powerful Collision DLL for DBPro and DBC: http://www.nuclearglory.com
Back to top
Profile PM Email Website
Pulsar Coder
20
Years of Service
User Offline
Joined: 3rd Jan 2004
Location:
Posted: 16th Mar 2006 23:51
Link
So, ETA?
Back to top
Profile PM Email Website
nuclear glory
21
Years of Service
User Offline
Joined: 2nd Oct 2003
Location:
Posted: 17th Mar 2006 00:06
Link
I'll tell you tomorrow. I'm taking the day off. Right now, I'm estimating no later than next week Friday (the 24th).

Powerful Collision DLL for DBPro and DBC: http://www.nuclearglory.com
Back to top
Profile PM Email Website
Baggers
20
Years of Service
User Offline
Joined: 31st May 2004
Location: Yonder over dem dere hills
Posted: 17th Mar 2006 08:19
Link
Ok guys this is verging on hijacking. Please move to another thread for discussing Nuclear Glory....even comparig these two plugins can be done there, just at the moment people with questions for sparky will get swamped with Nuclear Glory talk.
thanks

M.I.A is pending
Back to top
Profile PM Email
Lost in Thought
20
Years of Service
User Offline
Joined: 4th Feb 2004
Location: U.S.A. : Douglas, Georgia
Posted: 17th Mar 2006 11:01 Edited at: 17th Mar 2006 11:02
Link
Yeah, sorry about that I was about to post the same thing, it's my fault. The results posting has already been moved to the NGC forum under their benchmarking thread and a new thread will be created in the Program Announcements for NGC 4.02 when it is released which will contain the final results. The tests will be quite extensive not to favor either side. I will post the best, worse, and averages of each. So if anyone wants any info on this test will you please go to and post about it here: http://www.nuclearglory.com/php-bin/forums/view_topic.php?id=392&forum_id=2 Thank you.

Back to top
Profile PM Email
Back to top

Login to post a reply

Server time is: 2024-11-22 19:48:38
Your offset time is: 2024-11-22 19:48:38