Is this the first hardware accelerated PhysX game?

[MrNasty]

NEWS: This IS H/W accelerated - get it here: http://www.stokedrider.com/game.htm

Can't find any other comparisons of PhysX vs no PhysX scenarios, so I guess this could be the first PPU benchmark hehehe - 2 firsts in a day for [H]

Finally - Any requests on what to do next?

OK - my BFG card arrived today

So I've downloaded half of the internet trying to find a game/program that uses the hardware accelerated PhysX ver 2.4.0.

The only program I've found so far is Stoked Rider featuring Tommy Brunner.

The PhysXCore.dll that comes with the demo reads version 2.4.0 - so the number's right!

Unfortunately there isn't any option to enable or disable HW physics in the game, so I'm going to try FRAPSing some 800x600 action with the BFG card enabled and disabled - I'll plonk some screenies and results up later.

UPDATE:
The STDout.txt file the game generates when starting up outputs the following:

/////////////////////////////////////
///// PHYSX SCENE INITIALIZED ///////
/////////////////////////////////////
Ich bin ein hack der was chaeckt...128 / 128
RiderEntity::RiderEntity()
RiderEntity::~RiderEntity()
RiderEntity::RiderEntity()
RiderEntity::~RiderEntity()
HardwarePhysic::~HardwarePhysic()

Even after disabling the PPU this section stayed the same, so I guess it's not definitive proof

UPDATE 2:

FRAPS:
With PhysX card enabled

Frames, Time (ms), Min, Max, Avg
1579, 23016, 47, 191, 68.604

Without PhysX card enabled

Frames, Time (ms), Min, Max, Avg
1655, 25219, 46, 193, 65.625

This was a pretty rushed comparison - not to mention pretty crude - and tbh I don't see the card getting an awful lot of work in this game...

UPDATE 3:
Seeing as 3Dmark06 uses the PhysX engine I figured I'd try a little benching based around that.

Unfortunately the core.dll seems to be much older than the 2.3.1 version (the oldest .dll supported by hardware).

I benched all the same and found no differences between the mark with the PPU enabled or disabled - unsurprisingly...

I then gave switching the .dll's around a try on the hunch that 3DMark06 isn't a physics showcase and so would only use basic features likely to be the same between the old version and new version. This didn't fare very well - 3DMark crashed upon starting the CPU test, with an error message indicating the NovodeX engine refused to load.

Scavenging the loader.dll from Stoked Rider only yielded an error message on startup detailing an illegal instruction.

Rig:
3800+ venice s939
2gb Corsair XMS 3200
256mb 7800GTX (stock)
Res: 1024x768 - all other options at default

 

[AaronP]

Where are you buying these cards?

 

[MrNasty]

Where are you buying these cards?

It's mentioned in one of the other threads - check them out

Unfortunately you can't get them anymore - AFAIK - and I don't want to get them into any more trouble at the moment...

 

[MrNasty]

OK - after more testing it appears that there is a real difference in terms of ragdoll movement detail between hardware enabled and hardware disabled.

When the PPU's enabled, it looks more fluid, so I'm going to try FRAPSing at a point when the ragdoll model comes into play - will post the results.

UPDATE:

New tests done at 800x600 with every option turned off (LODs still 50)

PhysX off test runs

Frames, Time (ms), Min, Max, Avg
1721, 13269, 89, 195, 129.701
1418, 10361, 91, 207, 136.859

PhysX on test runs:

Frames, Time (ms), Min, Max, Avg
1176, 8399, 100, 188, 140.017
1104, 8889, 98, 199, 144.198

These are both from the same drop point - it seems that without PhysX the section I'm FRAPSing is different. I've recorded demos (vids I mean) of these and it seems there is quite a bit of ragdoll detail difference... Looks like it does use it

 

[jebo_4jc]

In about 2 weeks you will get a real test:

http://www.xbitlabs.com/news/multimedia/display/20060420060459.html

I think if GR-AW is cooler with the PhysX, I will probably uncontrollably splooge and buy it just because I'm that excited about the possibilities of the new tech.