Intel Destroy The Casttle demo powered by Newton

Share with us how are you using the powerrrr of the force

Moderator: Alain

Intel Destroy The Casttle demo powered by Newton

Postby Julio Jerez » Thu Jul 17, 2008 1:55 pm

this is a demo that was played at DCD two a year ago using an open source physics engine.
http://www.newtondynamics.com/downloads ... untime.zip

I came across it searching over the Intel website for optimization techniques for multi threading.
It was a nice surprised seeing they realize the source. because this demo created lot of controversia among the gratuitos physics engine appraisals in some open source graphics and physics forums.
Basically is was used as prof of the superiority of open source just because a big company like Intel selected a popular open source physics engine to make it, which of course it a huge Fallacy.

As you can see the demo could easily be executed by any ordinary physic engine.
The demo was not made to demonstrate how good the the physics engines was at mutithreading,
It was made by Intel to show the versatility of multicores CPUs.
But the self appointed experts in open source physics and graphics forums took it as a sign
that Intel was saying these engines was the better for Multicores.
Julio Jerez
Moderator
Moderator
 
Posts: 11039
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Intel Destroy The Casttle demo powered by Newton

Postby agi_shi » Thu Jul 17, 2008 3:35 pm

Haha, the original version falls apart on its own without firing the cannon :roll:. Nothing seems stable, everything is jittering. The newton version is rock stable, no wiggling and it most definitely doesn't eventually fall on itself. Awesome! :D
agi_shi
 
Posts: 263
Joined: Fri Aug 17, 2007 6:54 pm

Re: Intel Destroy The Casttle demo powered by Newton

Postby Leadwerks » Thu Jul 17, 2008 4:14 pm

This is a really good comparison.
User avatar
Leadwerks
 
Posts: 531
Joined: Fri Oct 27, 2006 2:54 pm

Re: Intel Destroy The Casttle demo powered by Newton

Postby vectrex » Fri Jul 18, 2008 2:43 am

Nice. To be fair autofreeze should be off on both. Demos like this are all that's needed to evaluate different physics engines. Personal opinion is pretty useless since everyones requirements are different. I'm always amazed at how people argue about physics/gfx engines like they're arguing over music preference or something :) Let the demos speak for themselves.
vectrex
 
Posts: 21
Joined: Sat Oct 01, 2005 10:15 am
Location: Australia

Re: Intel Destroy The Casttle demo powered by Newton

Postby Julio Jerez » Fri Jul 18, 2008 11:13 am

Hi Vectrex.
vectrex wrote:To be fair autofreeze should be off on both.

I did not altered anything in the original, but I beleive that for I saw the original demo do have outofrezze on.
The source code is valiable for these using Archemedia tha way maybe you can verify if I am did something wrong.
also you can try setting auto freeze off in the newton world and see teh difference.
Julio Jerez
Moderator
Moderator
 
Posts: 11039
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Intel Destroy The Casttle demo powered by Newton

Postby vectrex » Fri Jul 18, 2008 12:38 pm

hehe I know newton v2 would be stable because I always turn off auto freeze when testing demos etc and newton demos are amazingly stable. Some slight movement but nothing compared to others. The ODE version looks to not have it on as it falls apart. If it IS on then it's broken :)
I just say turn auto freeze off to stop people accusing it of not being a fair comparison. Because it's a nice demo to show.
vectrex
 
Posts: 21
Joined: Sat Oct 01, 2005 10:15 am
Location: Australia

Re: Intel Destroy The Casttle demo powered by Newton

Postby Julio Jerez » Fri Jul 18, 2008 1:09 pm

how about a key option uption?
I will also add the ambient sound.

vectrex wrote:I just say turn auto freeze off to stop people accusing it of not being a fair comparison. Because it's a nice demo to show.

To tell you the truth I gave up in trying to satidfy the poeple, it does not matter what I do i will never be able to satisfy evrery body. There are peoeple in many places that seems to be very mad and angry at me, just because I made a physcis engine, It is madness.
Julio Jerez
Moderator
Moderator
 
Posts: 11039
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Intel Destroy The Casttle demo powered by Newton

Postby vectrex » Fri Jul 18, 2008 1:31 pm

Julio Jerez wrote:how about a key option uption?
I will also add the ambient sound.

vectrex wrote:I just say turn auto freeze off to stop people accusing it of not being a fair comparison. Because it's a nice demo to show.

To tell you the truth I gave up in trying to satidfy the poeple, it does not matter what I do i will never be able to satisfy evrery body. There are peoeple in many places that seems to be very mad and angry at me, just because I made a physcis engine, It is madness.


key option would work. I wouldn't waste too much time on it though.
I'd would also avoid saying things like this "self appointed experts". As soon as you get personal (it doesn't matter who started it) things get ugly :) I find even if one person deserves it, they aren't alone, other good people who didn't deserve it are on that team and they will see comments like this out of context and take great offense even though that wasn't the intent. Now they will hold a grudge against newton and it spreads and spreads. No good can come of it. That's why I like demos like this, open source, no comment needed. If someone on another forum suggests newton can't do something or some engine is 'the best', just post the demo link with zero comments. No one can deny a real demo, whereas personal speculation about people motives are useless and cause misunderstanding and anger. It's taken me a long time to get out of the habit myself 8)
vectrex
 
Posts: 21
Joined: Sat Oct 01, 2005 10:15 am
Location: Australia

Re: Intel Destroy The Casttle demo powered by Newton

Postby Julio Jerez » Sat Jul 19, 2008 1:45 pm

Ok I fixed a few things I did not had time last weekend in the demo.

-I am now passing the position and the type of sound to the sound library.
-The cannon ball when created it set to a huge force value, but when I use the value in the original demo, it did not reall wiork in newton because I did nto wna to coem up with elaborate call back to just apply teh fire implu ot the ball.
instead I multiplied teh force bu teh time step and applied and impulse, but is was still to strong.
Then a realized that the demo uses centimeters units but the gravity is in miter/sec2 (have no idea how they mix units like that, built obviously is works for them)
I scale teh gemetry properlly to metters but I forgot to scale the cannon ball impulse to meters as well, therefore it was travelling 100 time faster than in the original demo making the castle blowup with a single shot.
Scaling teh impule properlly to meter * mass / sec mak eteh shot like in the original demo.

Now the sounds and particles show up, which is nice, and also cattle can take a big beating without collapsing to the ground.

Vertrex.
I think that is better than hacking the original code to add a user interface command which I do not really want to do.
plus sinc eteh catle cat take many hit without collasing I think it reflect a more real game scenarion.
Julio Jerez
Moderator
Moderator
 
Posts: 11039
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Intel Destroy The Casttle demo powered by Newton

Postby JernejL » Sat Jul 19, 2008 1:55 pm

is the reset scene fixed as well? in intel version pressing space reset the castle, in newton it didn't.
Help improving the Newton Game Dynamics WIKI
User avatar
JernejL
 
Posts: 1492
Joined: Mon Dec 06, 2004 2:00 pm
Location: Slovenia

Re: Intel Destroy The Casttle demo powered by Newton

Postby Julio Jerez » Sat Jul 19, 2008 2:59 pm

Oh I did no tknwo that,
I just tested it and when I hit space the scene fade out and re inilliaze the cattle, isn't that was it supposet to do?
could you test it again?
Julio Jerez
Moderator
Moderator
 
Posts: 11039
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles

Re: Intel Destroy The Casttle demo powered by Newton

Postby Dave Gravel » Sat Jul 19, 2008 3:12 pm

Yes the space working ok for me too.

I think it reflect a more real game scenarion.

[off-topic]
Yes but the threaded implementation from intel is wrong, if you like I can make a video to show where it is wrong.
[Edited] I can forget the video idea, my video recorder don't work good in fullscreen and Fraps make run the simulation so slow to show the problem...

This method add some flickering in the physics update, you can see this in the profiler.
If you moving fast or if the physics object moving fast when this flickering happening the visual don't follow the physics normally.
Exemple when the physics scene is really in action and if the application progress messages like mouse, or keyboard the physics waitting and it is not so good in a game or simulation.

If someone like to use this threading method in a game he need a better implementation that what the intel demo using.
When the threading is off in the demo the messages can progress normally and it don't add flickering in the physics update...
I have test a lot the intel demo and I have implement the same method in a demo and both get the same problem...

Here some explanation of what I trying to say in general...
http://www.gamasutra.com/features/20051 ... b_01.shtml
You search a nice physics solution, if you can read this message you're at the good place :wink:
OrionX3D Projects & Demos:
https://www.facebook.com/dave.gravel1
http://orionx3d.googlepages.com/
https://www.youtube.com/user/EvadLevarg/videos
User avatar
Dave Gravel
 
Posts: 713
Joined: Sat Apr 01, 2006 9:31 pm
Location: Quebec in Canada.

Re: Intel Destroy The Casttle demo powered by Newton

Postby dreamstyler » Mon Jan 19, 2009 4:24 am

Nicely done, gives a nice comparison.
Could you please add the missing source code require to compile it, ie the folder Dynamics since the compiler cant find "Dynamics::SDK_NEWTON". The only folder that is in the zip is the ParallelDemo folder.
Thanks
dreamstyler
 
Posts: 1
Joined: Mon Jan 19, 2009 3:47 am

Re: Intel Destroy The Casttle demo powered by Newton

Postby Julio Jerez » Mon Jan 19, 2009 12:55 pm

Ok done, you can download it again.
Julio Jerez
Moderator
Moderator
 
Posts: 11039
Joined: Sun Sep 14, 2003 2:18 pm
Location: Los Angeles


Return to User Gallery

Who is online

Users browsing this forum: No registered users and 0 guests