My Test of GRAW running with and without physX

fini · 30 Apr 2006 at 19:09

[ui]ICEMAN said:
800x600 with all lowest options enabled, same story.

On

Explosion 35
Grenade 55
Gunshot 80

Off

Explosion 60
Grenade 90
Gunshot 110

That's shocking - surely the graphics card's not the bottleneck at that res - even with the extra graphics from the PPU to draw. That must mean the physics card's 'overloaded' with even in the minor effects it's having to display in GRAW :eek:

.

fini

Cyanide · 30 Apr 2006 at 19:10

[ui]ICEMAN said:
As a reviewer I always do my best to remain impartial, you can't judge the performance of this card on one game, whicht could possibly be one very badly implemented game at that.

Review the game then

[ui]ICEMAN · 30 Apr 2006 at 19:16

fini said:
That's shocking - surely the graphics card's not the bottleneck at that res - even with the extra graphics from the PPU to draw. That must mean the physics card's 'overloaded' with even in the minor effects it's having to display in GRAW .

fini

It wasn't the bottleneck at the previous resolution either. These aren't mid range systems I'm testing on here, these are absolute top of the line, what should be more than capable of handling a few extra polygons flying around. We obviously have some fundamentally flawed implementation in GRAW, whether its driver related or merely poor optimisation by the game developers is still unknown, in all honesty I don't think we'll know for sure til one of the major games like UT2K7 hits the shelves.

Impy77 · 30 Apr 2006 at 20:31

Im interested to see what game developers do with the extra physics processing power, perhaps we shall get more physics based gameplay (ala HL2) or maybe even some clever physics based puzzle type games.

People in the thread have mentioned UT2k7 as being one game they want to see how the physics are implemented, however I can't see how they will really take advantage of the add-in card.

The reason being that UT is a online multiplayer based game, and in these type of games the one thing you cannot have is discrepancies between what each client sees. So for example, if you have a physX card and the vehicle interact with the game world in some way, that has to basically show in the same way as some one who doesnt have an add-in card - otherwise you can end up with a situation where the vehicle ends up in a different position on each players screen.

Thus I reckon all you will see in 2k7 is more 'fluff' on the explosions (eg vehicle explosions or the power core is taken down in ONS), similar to the explosion fluff we are seeing in GRAW. Other than that the card might take the load of CPU during vehicle crashes etc and give a minimal increase in fps at best.

Not that I want to see the card fail, I think it will be interesting to see how the add-in card fares against physics implementations using the GPU. To be quite honest anything that can inject a bit of innovation into game development has to be good right now.

Impy

Cyber-Mav · 30 Apr 2006 at 21:01

flop goes this card.

FirebarUK · 30 Apr 2006 at 21:18

Too early to tell if you ask me. I'd say they probably haven't sorted out load stuff properly and programmed it right, but hey.

Cyber-Mav · 30 Apr 2006 at 21:54

err, no, don't you think that they would have tested this game properly using the ppu to see how it works. the game makers knew this would happen and well what can they do about it? nothing,

the ppu is underpowered and is basically useless.

all i can say is that with the rise of the ppu all it means is that game programmers can become even more lazy since they don't need to optimise physics code in games, they just offload it to the ppu in theory and be done with it. this allows game to be produced at a faster rate.

the future of pc games i getting darker all the more.

BubbySoup · 30 Apr 2006 at 22:05

Falkentyne said:
Thanks for proving my point, Iceman ,and for testing that out.
And also sorry that you had to waste $(200 300 USD?) to just be scammed like this. I will not be buying this card to be the physics version of the S3 Virge.

Because that's all it is.....i mean wow, look at the HUGE reduction in framerate when the video card is completely removed from the picture ! That's **Exactly** what the S3 virge did back then. (I should know--I had one).

I have no clue what you are talking about comparing it to a virge card but yes, let's all write off a new tech based on a huge sample of ONE game which has so obviously had the PhysX support botled on as an after thought.

Moving on to some constructive views, we can see from this thread that ;

- The cards are over-priced.
- It would not be worth buying one yet, even if they were in the ideal price range of £50-£100.
- Adding support for PhysX as an after-thought does not seem to be a good idea.
- There is something very wrong with the way GRAW handles the PhysX PPU and/or the PhysX drivers.
- We have yet to see a game built from the ground up with PhysX in mind.

[ui]ICEMAN · 30 Apr 2006 at 22:05

Cyber-Mav said:
err, no, don't you think that they would have tested this game properly using the ppu to see how it works. the game makers knew this would happen and well what can they do about it? nothing,

the ppu is underpowered and is basically useless.

all i can say is that with the rise of the ppu all it means is that game programmers can become even more lazy since they don't need to optimise physics code in games, they just offload it to the ppu in theory and be done with it. this allows game to be produced at a faster rate.

the future of pc games i getting darker all the more.

Since when has this ever applied in the world of game development, I've lost count of the hundreds of times features have been rush coded, last minute additions that would have been better off being left out, it happens all the time with graphics cards, it is still far too early to pass judgement based on ONE software title from a publisher known very well for putting pressure on the companies developing the games to get them out earlier than they should be.

BubbySoup · 30 Apr 2006 at 22:12

[ui]ICEMAN said:
Since when has this ever applied in the world of game development, I've lost count of the hundreds of times features have been rush coded, last minute additions that would have been better off being left out, it happens all the time with graphics cards, it is still far too early to pass judgement based on ONE software title from a publisher known very well for putting pressure on the companies developing the games to get them out earlier than they should be.

Agreed

Take the just released Dreamfall for instance. The installer falls over by checking to see if your C drive has room to install Dreamfall. If it doesn't, it exits without letting you change drive letters!!!

To get around it you have to edit your registry...

So yes, I can see GRAW being released with PPU support practically untested/unoptimised.

Falkentyne · 30 Apr 2006 at 22:31

Impy77 said:
Im interested to see what game developers do with the extra physics processing power, perhaps we shall get more physics based gameplay (ala HL2) or maybe even some clever physics based puzzle type games.

People in the thread have mentioned UT2k7 as being one game they want to see how the physics are implemented, however I can't see how they will really take advantage of the add-in card.

The reason being that UT is a online multiplayer based game, and in these type of games the one thing you cannot have is discrepancies between what each client sees. So for example, if you have a physX card and the vehicle interact with the game world in some way, that has to basically show in the same way as some one who doesnt have an add-in card - otherwise you can end up with a situation where the vehicle ends up in a different position on each players screen.

Thus I reckon all you will see in 2k7 is more 'fluff' on the explosions (eg vehicle explosions or the power core is taken down in ONS), similar to the explosion fluff we are seeing in GRAW. Other than that the card might take the load of CPU during vehicle crashes etc and give a minimal increase in fps at best.

Not that I want to see the card fail, I think it will be interesting to see how the add-in card fares against physics implementations using the GPU. To be quite honest anything that can inject a bit of innovation into game development has to be good right now.

Impy

You make a very, very good point.
All clients should be able to see all ragdoll physics and other effects, but again----someone with a physics card should get **HIGHER** FPS than someone without one. If this is not the case in 2k7 multiplayer, then the card was a complete scam.

An AI accelerator would be more useful than this thing......

Impy77 · 30 Apr 2006 at 22:38

Cyber-Mav said:
all i can say is that with the rise of the ppu all it means is that game programmers can become even more lazy since they don't need to optimise physics code in games, they just offload it to the ppu in theory and be done with it. this allows game to be produced at a faster rate.

the future of pc games i getting darker all the more.

I understand your concerns here, but part of programming is in effect to unload part of your program to something else and you dont need to worry about the details. In fact this should benefit programmers as long as the API is properly optimised.

This is what RAD (Rapid application development) has always been about. In fact directx is all about this (and will be even more so with DX10 when so strict rules of how stuff is rendered are finally laid down - no more one card does DirectX one way and another does it different). DirectX has allowed game programmers to say 'render this' and DirectX goes away and takes care of the details. The next gen physics programming will need such an API as well. At the moment we have APIs from Havok, PhysX and other physics engines as well.
At some point there will be a popular physics API laid down in the same way DirectX has become the norm for games to use rather than OpenGL, and I have no doubt the most popular one will be what ever Microsoft chooses, whether it be in house or brought in 3rd party. A physics API is going become part of DirectX just like sound and graphics are now, at least thats my theory anyway.

Things change I've changed · 1 May 2006 at 01:55

Falkentyne said:
You make a very, very good point.
All clients should be able to see all ragdoll physics and other effects, but again----someone with a physics card should get **HIGHER** FPS than someone without one. If this is not the case in 2k7 multiplayer, then the card was a complete scam.

tbh thats the way I thought it was going to be implemented and i'm hoping thats the way it will be done in future games releases.

Soul Rider · 1 May 2006 at 11:00

Impy77 said:
At some point there will be a popular physics API laid down in the same way DirectX has become the norm for games to use rather than OpenGL, and I have no doubt the most popular one will be what ever Microsoft chooses, whether it be in house or brought in 3rd party. A physics API is going become part of DirectX just like sound and graphics are now, at least thats my theory anyway.

Quoted for truth. Microsoft WILL bring physics implementation into DirectX or WGF as it is now known. Then everything will be so much simpler, if not always the best quality.

Although there is a fightback going on against DirectX with OpenAL, but I have a feeling this will just go the same way as OpenGL.

ihatelag · 1 May 2006 at 11:01

lowrider007 said:
tbh thats the way I thought it was going to be implemented and i'm hoping thats the way it will be done in future games releases.

After watching the "real time footage demo" from the AGEIA website, I noticed something which I didn't notice before...when the explosion happens (right hand screen) with PhysX on, the explosion looks nice, but you can visibly see the FPS get hammered down...if you look closer you can even see the debris slow down like it's in the matrix.

If you keep switching between PPU off and PPU on screen, over and over again, you can definately notice that, when the PPU is off it shows a faster FPS...pretty funny that I didn't notice then when looking at it the first time...I was too much in awe with the nice explosions to even pay attention to it

http://physx.ageia.com/footage.html

Psypher5 · 1 May 2006 at 11:13

wow, it really kinda does too!

Well, maybe it doesnt look like frame rate drop totally, as when he lookes to the diff object its smooth, more of an effect? (but from playing oblivion I know things can still "look smooth" when your busy watching pretty-ness at low fps, like 20-30)

Also the cellfactor video still makes me dribble

What kind of beastly machines do these developers have

Cyber-Mav · 1 May 2006 at 11:42

cell factor's gfx are nothing impressive. reminds me of a game called MDK. all it has added is lots of boxes moving around which can easily be done with the cpu alone.

play a game called red faction. its the first game i ever played that allows the terrain to be deformed with rockets/grenades etc. and no ppu is needed for that.

G.O.A.T · 1 May 2006 at 11:55

Cyber-Mav said:
cell factor's gfx are nothing impressive.

I'd hate to see what it takes to impress you. :eek:

turtle3 · 1 May 2006 at 12:06

Why does the GRAW installation insist on you installing the Agea Phisx card drivers even if you dont have one :confused:

Is it to tempt you into getting one just to remove the disheartening ''Agea PPU not installed'' Icon it puts in your taskbar..lol

BubbySoup · 1 May 2006 at 12:09

Cyber-Mav said:
all it has added is lots of boxes moving around which can easily be done with the cpu alone.

play a game called red faction. its the first game i ever played that allows the terrain to be deformed with rockets/grenades etc. and no ppu is needed for that.

Making sweeping statements like that is very easy when we have no means of testing it currently, but you clearly are not fully realising what is happening on the screen. There is absolutely NO WAY a CPU could calculate what is going on there. It's not simply boxes moving about. Every single box's path is being determined by the collisions going on around it, including all the other boxes.

Also, terrain deformation is play school physics compared to moving body interaction.