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.

Dark Physics & Dark A.I. & Dark Dynamix / PPU Benchmark

Author
Message
AVC
18
Years of Service
User Offline
Joined: 15th Aug 2006
Location:
Posted: 16th Aug 2006 04:24
Link
I am really new to programming in Dark basic and Dark Physics. I have an AGEIA Physx card and have run some of the demos (like water) and see that you can specify to run in hardware or software, but I don't really see a performance difference between the two.

Can someone show me a quick example of using a bunch of objects, say 2 or 3 thousand, running in software only mode then running in hardware mode so I can see what the fps is with or without?

Thank you.

AVC
Back to top
Profile PM
Mike Johnson
TGC Developer
22
Years of Service
User Offline
Joined: 13th Sep 2002
Location: United Kingdom
Posted: 16th Aug 2006 10:20
Link
Try the Dark Physics\Demos\PhysX\Bricks demo. The demo creates a large quantity of rigid body boxes that construct 4 walls. A further set of rigid bodies are created and dropped into the scene. Initially this process starts off in software and if you let it run to the end you will get a good idea of the typical frame rate. Once everything has collapsed press the space bar to switch to hardware mode. There will be a significant increase in the frame rate when running in hardware mode.
Back to top
Profile PM Email Website
RickV
TGC Development Director
24
Years of Service
User Offline
Joined: 27th Apr 2000
Location: United Kingdom
Posted: 16th Aug 2006 15:33
Link
Also be aware that once the physics is being handled by the PPU, you CPU is freed up for other tasks!

Financial Director
TGC Team
[Check out Jed McKenna - http://www.wisefoolpress.com/]
Back to top
Profile PM Email Website
Fallout
22
Years of Service
User Offline
Joined: 1st Sep 2002
Location: Basingstoke, England
Posted: 16th Aug 2006 16:27
Link
Actually, that just made me think of something. When we quote minimum specs for our games now, it'll probably make sense to quote different specs with or without a PPU. Instead of just saying PPU supported, I suppose a game that needs a 3ghz CPU might only need a 2GHZ CPU with a PPU equiped.

Back to top
Profile PM
AVC
18
Years of Service
User Offline
Joined: 15th Aug 2006
Location:
Posted: 16th Aug 2006 19:37
Link
This is perfect. Thank you all very much.
Back to top
Profile PM
AVC
18
Years of Service
User Offline
Joined: 15th Aug 2006
Location:
Posted: 16th Aug 2006 20:18
Link
Actually I have a few questions about this demo. I bumbed the amount of bricks up to 2852, and in software I see that it starts running at about 11fps, but as the bricks begin to tumble the fps goes up to 23fps. Shouldn't this be the opposite? I would think that when the bricks are in motion, that the cpu would have to work harder to calculate the motion and collisions now going on vs. when they are just sitting there. When I run the same test in Hardware, I start at 23fps, and it stays there through out the whole simulation.

I also noticed that my cpu usage in software was on avg 58% and when running in Hardware was about 49%. That's nut much of a difference. Is there something in the code that may be incorrect? It doesnt look it's using the card to it's potential. Is it that there is a gpu limitation? I see that you can have a fps counter, is there a way to display how hard the ppu, gpu, and cpu are working?

Thank you for your time.
Back to top
Profile PM
AVC
18
Years of Service
User Offline
Joined: 15th Aug 2006
Location:
Posted: 16th Aug 2006 21:44
Link
I am also getting this error quite a bit when I try to run the hardware version.

PPU Assert: ASSERT [../../core/dma.c 364]: bytes < (64*KILO) called from 0x16a

Maybe my hardware isn't working properly.
Back to top
Profile PM
bosskeith
18
Years of Service
User Offline
Joined: 5th Dec 2005
Location:
Posted: 17th Aug 2006 04:14 Edited at: 17th Aug 2006 04:17
Link
actualy when they are all just sitting in there stacks is when iit is going to use the most processing because of the proximatey when they start to pull away and fall apart they no longer have any vector forces between them that needs to be calculated.

think of the old archway and keystone construction method...the entire arch is held together by 1 wedge stone at the top because of this one stone the other stacks of stones do not fall therefore it has an effect across all of the stones beside and below it...if you remove the stone it's effect is no longer applied and the whole thing topples

Ink all over the screen...and dang it! Signature still does not show.
Back to top
Profile PM Email Website
AVC
18
Years of Service
User Offline
Joined: 15th Aug 2006
Location:
Posted: 17th Aug 2006 04:27
Link
That makes perfect sense.

soooo why such a low delta between hardware and software? Does it have something to do with the error message I posted? I also tried one of your other demos, "Rigid Bodis" It said in the notes that there should be a 30% difference. I got none. Started at 25 then dropped down to 12 on both cases.

Am I doing something wrong?
Back to top
Profile PM
Dark Eternity
20
Years of Service
User Offline
Joined: 10th Jul 2004
Location: Vana Diel
Posted: 17th Aug 2006 19:02
Link
Quote: "That makes perfect sense.

soooo why such a low delta between hardware and software? Does it have something to do with the error message I posted? I also tried one of your other demos, "Rigid Bodis" It said in the notes that there should be a 30% difference. I got none. Started at 25 then dropped down to 12 on both cases.

Am I doing something wrong?"


i think if people post their specs it is easier to understand this, maybe your graphic card cant keep up?

Scorched Skies is waiting,can you help

AMDAthlonFX-62,NvidiaGeforce7900 512MB X2,RAM4GB
Back to top
Profile PM Email Website
AVC
18
Years of Service
User Offline
Joined: 15th Aug 2006
Location:
Posted: 18th Aug 2006 07:27
Link
Sure I am using a Shuttle with a dual core fx-62 with 2 gigs of ram and a 7900gt.

Is there any feature of dark physics that can display either the physics card usage, or especially the gpu performance so I can see where the bottle neck is?
Back to top
Profile PM
Denis
20
Years of Service
User Offline
Joined: 18th Jan 2004
Location:
Posted: 19th Aug 2006 14:33
Link
I am also getting several errors (crashes)when running in hardware, but I run tests on the Ageia PPU card, all chips appear to be OK. If I reset the card, demos like slime run fine for a couple of seconds then fail with "Internal Driver Error:" but no # after the colon.

Looks like the SDK might have some problems .......
Back to top
Profile PM
AVC
18
Years of Service
User Offline
Joined: 15th Aug 2006
Location:
Posted: 22nd Aug 2006 01:15
Link
Also, just for the Record. "I LOVE DARK PHYSX" It's so easy to use and I love the stuff you can do with it.

Thank you for making it!!!
Back to top
Profile PM
Back to top

Login to post a reply

Server time is: 2024-11-24 15:06:39
Your offset time is: 2024-11-24 15:06:39