Box StressTest, amazing....

ColeZero

02-06-2006 15:49:44

I've test nxOgre with many,many objects that are about 2500 boxes.
See screens. Yes PhysX is powerful. No Doubt about that.^^

Screen 1


Screen 2


Screen 3



Wow..

betajaen

02-06-2006 17:25:26

I keep telling people, but does anyone listen!?

BenO

03-06-2006 08:57:04

commercial usage is expensive ^^

betajaen

03-06-2006 09:43:15

I don't think it really matters if your a business because you'll make that up in sales of the game anyway.

Wretched_Wyx

03-06-2006 09:56:54

I think people need to actually read about something they are going to use. Not just go "Hey! Everyone uses this, so I will too!". Seriously *read* documentation. Read why a particular item is so cool (or not so cool). You don't need a commercial license to use PhysX, unless your product is going to be commercial. And, as betajaen stated, if you were making a *serious* commercial product, the $50k the PhysX license cost would be nothing.

It seems alot of people get scared away from PhysX because it's "commercial middleware". See, Planta just reached everyone's ears recently, and they are excited. But wait! It's a commercial product. So why don't people stray away from that? Well, lucky for us, there is a little thing called "free for non-commercial use". This same kick-ass thing happens to apply to the PhysX SDK.

So what's the formula here? Free game project = Free SDK. Commercial game project = Commercial license. So for a project using PhysX commercially, it would look like this: Profit = Overheads + Salaries + Game Dev + 50000 PhysX License - (unit sales * price). So again, the PhysX license is chump change to a serious game companie, and quite possibly even a serious indy game company that has a sponsor (or more) and a publisher.

Not only do you have the cool feature of physics powered by PhysX, but you can actually slap that nice logo on your box and state "Powered by PhysX". Heck, investing in PhysX may end up boosting your sales.

Oh, and BenO- please don't take this personally, and I'm not flaming. This has been something that has been itching at me for some time, as alot of people "complain" about PhysX and choose to use something else simply because they are scared of a commercial product.

[edit]
Heh. Totally forgot to say this, but ColZero- that looks sweet. Good job on that. Didn't take you long at all to dive in. Pretty easy huh?
[/edit]

BenO

03-06-2006 10:38:54

i won't take it personally, my first statement was really "light", but :

am in business.
am not doing games :o
my company can't afford physX.

and i hope my product could be considered as serious ^^

ColeZero

03-06-2006 11:09:13

Heh. Totally forgot to say this, but ColZero- that looks sweet. Good job on that. Didn't take you long at all to dive in. Pretty easy huh?
Yeah really the API of nxOgre is really simple, but powerful.

You can easly add your own particlesystem, like the smokeemitter if take a look into the code..The Source of nxogre is so self-explanatory. very very good job,to the nxOgre-Dev team.


So but it seems the question for this thread is: What is commercial and what is non-commercial..^^

BenO

03-06-2006 13:46:58

sorry, let's concentrate on the subject :o

what is the framerate (can't read it)?
how many cubes can u handle with 50 fps?
(could u describe ur configuration plz)

danharibo

03-06-2006 13:56:17

sorry, let's concentrate on the subject :o

what is the framerate (can't read it)?
how many cubes can u handle with 50 fps?
(could u describe ur configuration plz)
on he 3rd shot its 247 i think
(or is it 2.?)

ColeZero

03-06-2006 14:37:47

So that are 50*50 boxes, so that are 2500 boxes.
The Fps is about 3-4 fps, sadly not 200fps^^.
The demo runs on a Centrino 1.73Gh with 1GB RAM.
But i think 3 fps are alot, if you keep in mind that are 2500 boxes on the screen...^^

With my Ccmputer i can handle about 1000 Cubes, maybe more with a frame rate of 40-50.

Vectrex

03-06-2006 15:06:24

See, Planta just reached everyone's ears recently, and they are excited. But wait! It's a commercial product. So why don't people stray away from that?

$50,000 vs $800 that's why :) I'd LOVE to use Novodex (old school ;) ) but I can't think about that price. So I'm using Newton, which is definately slower but thankfully I don't need stacks of 100,000 boxes in my game :D

ColeZero

03-06-2006 15:17:55


I don't need stacks of 100,000 boxes in my game


So as i, but i don't use nxOgre because it can handle thousands of boxes
i use it because its so easy to use and its stable

Wretched_Wyx

03-06-2006 17:03:27

Yeah, I can get a stable 40-50 FPS with 1000 cubes as well. Now, this is without tons of optimized code, and some multi-threading going on. Pretty neat to think of what's possible...

I guess I won't mention anything about the qualification for a free PhysX license. Personally, I don't like anything outside of PhysX for a few reasons- Most are somewhat unstable, weak, lacking features... And not half as fun to work with as far as "ease of use" goes.

The fact that with NxOgre you can easily add more features is nice. Being able to add new particle emitter systems for instance. I don't mean making steam as opposed to smoke, but rather a complete new type, such as falling leaves. Perhaps you've noticed the "smokeEmitter", imagine being able to make a leavesEmitter, or dirtcloudEmitter for example. This is just a drop in the bucket really.

Then, the fact that new versions of the PhysX SDK come out regularly (which would also apply to Newton/ODE) is nice. And you know what's cool? I re-compiled the CVS version of NxOgre against the 2.3.3 version of the PhysX SDK with no hitches. Even saw slight performance and accuracy increases. I wouldn't doubt it will be this way when 2.4 comes around.

And lastly, betajaen has some cool things lined up for the next version of NxOgre. A good reference would be the addition of chemistry. Chemistry! I'm also pushing the idea of a new particle system for some sort of fluid system. That might be crazy though. They already outline a way to do this on the Ageia support site.

So when it comes down to it, I'll fully admit I'm in love with PhysX/NxOgre. I don't even listen to stuff about anything else anymore, because I know they just won't stand up strong in the world of physics. Though, I do fully understand and support the open-source ideals still, and of course people will have to go that route for specific reasons. You have to admit though, if you had the capital to become a licensee, you'd follow the path of PhysX.

ColeZero

03-06-2006 17:25:49

Hats off!!Yeah man you say it...^^

betajaen

03-06-2006 18:55:33

Wyx gets the NxOgre hat of the day award.

<~Creatine~>

23-09-2006 12:30:16

I've test nxOgre with many,many objects that are about 2500 boxes.
Awesome! :shock:
What PC configuration have you got? Have you got a Physx card?
And how Ogrenwt FPS speed with 2500 boxes on your machine?

betajaen

23-09-2006 18:45:37

I doubt he had a PhysX card then. I can get that with or without a PhysX card the only problem is that the render slows it down.

However it's quite unlikely to have that many objects on a screen at once or if they are they aren't there for long.

<~Creatine~>

23-09-2006 21:16:09

I doubt he had a PhysX card then. I can get that with or without a PhysX card the only problem is that the render slows it down.

However it's quite unlikely to have that many objects on a screen at once or if they are they aren't there for long.

Very interesting dilemma with PPU card because to be some professional engines and Ageia card prefers PhysX only. I saw working PhysX in another free engine on my machine and it's faster than whatever physics engine. (I haven't got a PhysX card.)

betajaen

23-09-2006 21:21:29

I have one.

I believe the problem when people assume that PhysX cards are slow is that because the card can simulate more things at once it takes advantage of that, but the graphics card can't keep up so it slows down. GRAW good example of this.

A better example to show is the CellFactor demo; I get 4 FPS without a PhysX card and 40 with.

<~Creatine~>

23-09-2006 23:48:01

I have one.

I believe the problem when people assume that PhysX cards are slow is that because the card can simulate more things at once it takes advantage of that, but the graphics card can't keep up so it slows down. GRAW good example of this.

A better example to show is the CellFactor demo; I get 4 FPS without a PhysX card and 40 with.

Yes, Cell factor running with Ageia card only. (But it can running without it ;)) The publiser on the chance that everybody will buy Ageia card who want to playing this game. :lol: It's a game only (and few big games UT2007 and 1-2 big games) and insecure all projects will prefer this expensive system.
MS answered:
http://members.microsoft.com/careers/search/details.aspx?JobID=6b94ac4f-0627-4851-8e6a-633186d96261

Machine:
Opteron 180, 2x1GB Mushkin XP4000 Redline, Asus A8R32-MVP, X1900XTX Crossfire, BFG PhysX PPU, X-Fi Extreme Music, 150GB Raptor, FSP 700W PSU, Lian-Li PC-777B, Dell 3007 WFP and Creative Gigaworks S750.

Options:
Game played at 1024x768, no AA (can't be forced from the ATI drivers ), 8xAF, HDR on, dynamic shadows off using a single x1900xt.

pics:
http://www.mashie.org/hardforum/cf_10.jpg
http://www.mashie.org/hardforum/cf_07.jpg
http://www.mashie.org/hardforum/cf_03.jpg

Excellent optimalization. :lol: :lol: It will running on NASA machine...

BlasterN

25-09-2006 04:06:19

I made more deep test, but with combination of tri-mesh, convex-mesh, Boxes and NxController.

I have to say that a scene with a 12k tris into a trimesh, three controlers (IA, HERO, and the camera) and a few boxes and convexs dont go down 72fps (limited because vsync) 3200 AMD64, 6600GT. plain textures onel light, stencill shadows. You will se a demo as soon I have some basic AI.

So yes, PHYSX is powerfull more than the other? I only have to say one thing Physx is easy to learn, but nxOgre is even easier.

betajaen

25-09-2006 09:24:59

That is one of the things that I feel people use PhysX, not only because it's awesome but it doesn't have a large curve in learning.

Everything is fairly easy to work out, and not only they provide in depth documentation but a series of tutorials with source and documentation. Not many Physics libraries, or many libraries do that.

Of course when you get NxOgre involved it's insanely easy.