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.

DLL Talk / Max. number of Newton objects?

Author
Message
RaceGT
19
Years of Service
User Offline
Joined: 15th Aug 2005
Location:
Posted: 13th Jun 2006 12:46
A question regardiding the maximum number of physics objects that would be feasible in a project:

I am using Newton and I find that I can't really go above 50-70 objects before I start to see the program lagging. I have only a 2.0 Ghz processor, but I'm sure there are others out there with newer ~4 Ghz machines that can feasibly handle more (physics objects).

So I'm wondering, either with Newton or any other available physics package, what would be the ceiling for the max number of objects before it would start to start to affect any machine?
I have around 50 objects right now (optimized as well as I could, and using the optimized 'cylinder' solid definiton), but I envision my project needing around at least 1000. Would this be feasible on any machine, and/or with a different physics package? It only needs to be rigid bodies for my simulation, no ragdolls or vehicles or anything, just rigid bodies.

I know the CPU intensity really kicks up when this many objects start coming to rest and laying/piling on top of one another. There are so many collisions and contact points to compute that I can hardly see any system being able to handle it, so I wonder if it would be possible.

http://www.geocities.com/crmnlelmnt/
EsteemDE
19
Years of Service
User Offline
Joined: 5th Aug 2005
Location:
Posted: 13th Jun 2006 15:39
Once I saw this line...

Quote: "with newer ~4 Ghz machines that can feasibly handle more"


I stopped reading because you have to think about what you are doing first. Very few people have 4ghz machines and the people that do wont want to play a game made in DarkBASIC unless 1) they use DarkBASIc, 2) they know you, or 3) they got bored of all other games and decided to find some random indy game some one made, even then its unlikely they will find yours.

Aiming for 4ghz machines is like aiming for an X1900 or 7900 card... why would you do that? If your 2ghz machine cant handle it that well then either stop there or lower it. 2ghz isnt that bad... a lot of people are still 1.6 and 1.8.

RaceGT
19
Years of Service
User Offline
Joined: 15th Aug 2005
Location:
Posted: 13th Jun 2006 21:15
Yeah thats what I's saying, is it even possible? What are the limits that different machines can handle?

http://www.geocities.com/crmnlelmnt/
EsteemDE
19
Years of Service
User Offline
Joined: 5th Aug 2005
Location:
Posted: 13th Jun 2006 23:03
If you have the expansion pack then just get the CPU speed of the person's computer and base the max ammount of objects on the result. That would probably be very very hard to code... and it might not even be possible... but that is the only solution I can think of right now. That is, unless you have lots of testers with good and bad PC's you will never know how much certain CPU's at certain speeds can handle.

And you say your PC can handle 50-70. That is not bad at all... actually that is great. I remember playing with the newton demos on my 1.8 PC (one I am on now) and just kept spitting out more and more objects watching them bounce around. I got about 150-200 on screen if I recall correctly.... I know it was 150 or more. But my 150-200 and your 50-70 (well at least I think so) did not have a full game engine in the background. Having a whole engine with all your graphics, NPC's, players, items, etc processing it will slow it down even more.

I suggest going down to 20-30 objects and see how it performs in a game. 20-30 is still a lot.... not many games have more than 5-12 on screen anyways... and that is with modern games.

RaceGT
19
Years of Service
User Offline
Joined: 15th Aug 2005
Location:
Posted: 14th Jun 2006 17:18
Quote: "you will never know how much certain CPU's at certain speeds can handle."




Quote: "I suggest going down to 20-30 objects and see how it performs in a game. 20-30 is still a lot.... "


Thanks, this simulation needs the number of objects I quoted...

http://www.geocities.com/crmnlelmnt/
Xarshi
18
Years of Service
User Offline
Joined: 25th Dec 2005
Location: Ohio
Posted: 14th Jun 2006 18:53
My computer is horrible,and it can support around 200 bodies without lag. Use box collisions where possible,those tend to not be as hard to handle for the computer(I may just be being a moron right now)

all ben needs is his band,his guitar,and his computer
EsteemDE
19
Years of Service
User Offline
Joined: 5th Aug 2005
Location:
Posted: 14th Jun 2006 21:13
Quote: "you will never know how much certain CPU's at certain speeds can handle."


What I mean is you wont be able to test both AMD and Intel CPU's with different CPU speeds.

Quote: "My computer is horrible,and it can support around 200 bodies without lag."


Yeah.... like I said I could handle 150-200 also. I do not know why he could only handle 50-70...

Quote: "Thanks, this simulation needs the number of objects I quoted..."


If it needs 50-70 objects then just keep trying different code. Just rewrite some of your programs code and keep rewriting it till it is finished. My 1.8 could handle 240-245 active physics objects before going under 30fps at 1024x768x32 in the Newton demos (I just tested). I do not know why you can only handle 50-70...

What else is going on in your program and what is your program like graphics wise? You never said what else you have in place yet... but maybe it isnt just newton you have a problem with because if I can handle around 240 objects you should be able too also.

RaceGT
19
Years of Service
User Offline
Joined: 15th Aug 2005
Location:
Posted: 15th Jun 2006 21:10
Well, I don't have a dedicated video card either, just ATI 200 onboard graphics. Though I was surprised, it performed as well as the machine it was replacing, with a GeForce2 card that was the top of the line, in its day. I'd like to get a new PCI express card, but for now it's just the onboard graphics.

Pretty much, the only thing happening in the program is the Newton objects colliding with themselves, a (ground) object, and about 30 convex hull Newton objects. The convex hulls are not very intricately shaped, but they do add to the Newton calculations.

I can actually handle more objects, 100, 200, 300 like was mentioned, and when these objects not very densely packed together and colliding, then it's mostly fine. It's when the objects must all land on the ground, near each other, and they start having to lay atop one another. Well, I tried it at 800 objects today and got about 2 FPS. I will have to find a way to use the physics sparingly, because I don't think there would be a way to handle 800 objects or more, that's quite a bit to ask for a physics simulation. Would be nice though!

http://www.geocities.com/crmnlelmnt/
Xarshi
18
Years of Service
User Offline
Joined: 25th Dec 2005
Location: Ohio
Posted: 15th Jun 2006 21:42
I wish I had a good computer...

all ben needs is his band,his guitar,and his computer
RaceGT
19
Years of Service
User Offline
Joined: 15th Aug 2005
Location:
Posted: 17th Jun 2006 06:18
Quote: "I wish I had a good computer..."


I think that's mostly what it is, so that's why I'm trying to find out if it's even possible for any machine to run this many objects....

http://www.geocities.com/crmnlelmnt/
General Reed
18
Years of Service
User Offline
Joined: 24th Feb 2006
Location:
Posted: 18th Jun 2006 21:59 Edited at: 18th Jun 2006 22:01
Your pc should be able to handle 600+ objects RACEGT. My 2.3GHz athlon handels 700 objects easily. But maybe its the memory???

Why dont you delete the physical objects when the user cannot see them and the objects are still, but store their positions?

AMD Athlon 64 3200+, 1GB ram, Geforce 6600 GTX, 250GB HDD
Vist www.scratchyrice-dev.co.uk
RaceGT
19
Years of Service
User Offline
Joined: 15th Aug 2005
Location:
Posted: 19th Jun 2006 14:27
I think you're right, General. I will need to find a way to not have the physics on "full time", but instead make the program pick and choose at what times the physics will be on. Unfortunately, most all the objects will be seen mostly all the time, but I can try to work something out.

http://www.geocities.com/crmnlelmnt/
RaceGT
19
Years of Service
User Offline
Joined: 15th Aug 2005
Location:
Posted: 19th Jun 2006 23:24
Well, whatever the outcome (I'll get something worked out someday...), I appreciate you guys stopping in to take a look at my question.

http://www.geocities.com/crmnlelmnt/

Login to post a reply

Server time is: 2024-11-22 06:18:54
Your offset time is: 2024-11-22 06:18:54