Far Cry mostly just PS 1.1 - even the water

[mrchimp]

All three cards were running in very high quality water mode. It turns out that Far Cry uses 1.1 shaders for rendering water, even in ultra high quality mode, which renders even more reflections on the water than the other modes:

http://www.firingsquad.com/hardware/far_cry_nvidia/

I find this very interesting, it's not just the nvidia cards that use PS 1.1 Ati's do aswell meaning this game has some serious nerv calling itself a SM 2.0 game and even more nerve to suggest SM3.0 will make a difference when it's barely 2.0 yet.

Personally I find it really hard to believe the water is just 1.1 but if it renders the same on a DX8 card as a DX9 card then it must be true, of course the DX9 cards will use there PS 2.0 engine to render the image but they will only use features that were in PS1.1.

Of course Far Cry does use PS2.0

So what does Far Cry use 2.0 shaders for? Apparently just lighting, everything else is handled by 1.1 shaders.

I don't know what they mean by lighting, it's a very large area of 3D graphics, but I can't see any difference between a GF4 and a R9xxx as far as lighting is concerned in those shots, maybe on some inside ones you can?

 

[Nostradamus]

next patch is gonna make it 3.0

 

[mrchimp]

I think you missing the point

 

[PvtRyan]

It would explain the 5950U being faster in Far Cry than the 9800. Because Nvidia has a weakness with PS2.0 and not with 1.1

 

[Nostradamus]

honesty mrchimp... i dont care if they lied, i dont care if they ate a pile of poo...what matters to me is that it looks nice...

 

[mrchimp]

when since was the 5950U faster than the 9800?? even if it is, it uses lower precision, the NV30 isn't too speedy when it comes to PS1.x compared to the r300 either.

Also I noticed the GF4600 was almost as fast as the GF FX 5700 in some of the tests.

EDIT: if only people didn't care when valve "lied" and for the record I'm not implying CryTech have lied about anything... over exagerated maybe, but thats not the point of my post, I'm just saying FarCry isn't much of a SM2.0 game, I'm not takeing a dig at CryTech or FarCrys graphics, I'm quite suprised at what they have been able to do with PS1.1.

Nvidia however are filthy,lieing,cheating,scum :cheers:

 

[Asus]

Benchmarks of 6800 using the 6800(default)DeviceID and then the R300(9800)DeviceID.
Reverse. Pictures showing the R300 using Nvidia's DeviceID vs its default (9800)DeviceID
The program chooses the path based on the DeviceID.

 

[mrchimp]

well the last link brings Nvidia down to size, I'm sure DH won't mind me doing this

I don't know what to think, maybe NV will get the performance back up without loseing IQ maybe they won't.

BTW I didn't start this thread to have a go at NV(although thats the way it's going). I was just amazed that the water in FC was PS 1.1.

 

[Death.Trap]

Personally I don't think the water is that great. I turned all my settings on high, had anti-aliasing on. The water in HL2 is waaaayy better. It's better than any game I've played before, but nothing to have a stroke about really.

 

[Asus]

Again we are at more FPS vs holding image quality.
I guess we should wait for the X800.

Although you should take the above links with a grain of salt.
Changing the code path could not allow it to use legit optimizations using commands for Nvidia hardware which can be special at times.
Although that picture of the Nvidia code path on the R3xx is terrible.

Farcry's water looks very nice for not being PS2.0.
I saw no differences between the pictures for water (except that last comparison with the error on Nvidia hardware, laff).
Those GF4 owners should be very pleased with the lack of difference in water visuals since its not using PS2.0.

 

[FictiousWill]

"wow, I can't believe anybody would even try playing with.... such short bars."

 

[crabcakes66]

"wow, I can't believe anybody would even try playing with.... such short bars."

umm....yeah i really dont want to play with my fps dipping down to 16 either.

 

[mortiz]

The GeForce 6800u still has a performance gain on the 9800XT using the R300's path, that's without AA and AF which has been proven to give 6800u's a significant advantage.

This one benchmark only proves nVidia have put out a bad driver set (or Crytek have some dodgy code), it doesn't prove the NV40's are bad cards.

 

[jonbob]

Wouldn't changing the path be an unfair comparison? Like making a lefty write with his/her right hand so you can compare to a non-lefty.

If you force the card not to use whatever legal optimisations there are (irrespective of any illegal ones), and compare it to another card using all the optimisations of its native code path then I fail to see the point of the excercise.

I seem to recall saying that Crytek still use the NV3x path for the NV40 cards, which, apparently is going to be fixed in a patch.

 

[mortiz]

Wouldn't changing the path be an unfair comparison? Like making a lefty write with his/her right hand so you can compare to a non-lefty.

If you force the card not to use whatever legal optimisations there are (irrespective of any illegal ones), and compare it to another card using all the optimisations of its native code path then I fail to see the point of the excercise.

I seem to recall saying that Crytek still use the NV3x path for the NV40 cards, which, apparently is going to be fixed in a patch.

technically since the R300 DriverID makes the game use full PS 2.0 effects, by changing the NV40 DriverID to R300 the game is using full PS 2.0 on the GeForce card where if it was running the nVidia DriverID the game apparantly uses a mix of PS 2.0 and PS 1.1 meaning some people think that Crytek have tweaked the NV40/30 settings in order to give a performance boost at the expense of IQ. The game does the same thing using the nVidia NV3x DriverID's as well.

Simply put all the changing of the DriverID does is make the game use a pure DX9 path with PS 2.0 (which obviously eliminates nVidia optimized features as well)

 

[mrchimp]

Actually the NV profile uses all the PS2 effects, just not with the same precision, also the whole point of this is to point out that FC really doesn't use that many PS2.0 effects.

Personally I'm not sure if it's fair to run a NV card in another cards code path, I know it would be unfair in a OGL program useing vendor specific extentions but then you would expect it not to work atall. In this case it works and the IQ is better, the performance drop isn't exceptionally high either so what you might be seeing is the 6800's real performance, I think it will get higher when CryTech release the patch though.

Maybe they should run the 9800 in the NV code path and see how it performs

 

[Caminante]

http://www.firingsquad.com/hardware/far_cry_nvidia/

I find this very interesting, it's not just the nvidia cards that use PS 1.1 Ati's do aswell meaning this game has some serious nerv calling itself a SM 2.0 game and even more nerve to suggest SM3.0 will make a difference when it's barely 2.0 yet.

Personally I find it really hard to believe the water is just 1.1 but if it renders the same on a DX8 card as a DX9 card then it must be true, of course the DX9 cards will use there PS 2.0 engine to render the image but they will only use features that were in PS1.1.

Of course Far Cry does use PS2.0



I don't know what they mean by lighting, it's a very large area of 3D graphics, but I can't see any difference between a GF4 and a R9xxx as far as lighting is concerned in those shots, maybe on some inside ones you can?

What a bunch of inept idiots are this people at firingsquad.com for misleading people. It turns out this morons used a MP maps for their test, instead of using a single player one. Anyone, with an once and half of common sence would realise that, the LOD has been slightly turn down in all MP maps to improve performance. I'm really looking forthward for this morons making a bigger jackass of themselves.

 

[mrchimp]

Are you saying FC uses PS2.0 for the water in SP? maybe somone with both DX8 & 9 GPU's could take screenshots to prove this, I'v played FC and I don't think the water is any different in MP than it is in SP, but unfortunatley I don't own FC so i can't do this myself.

Also I seriously doubt the GPU has any impact on network performance. There's no reason why the water should be toned down for MP or anything else that relys on the GFX card for that matter. Things that effect the CPU will be toned down to make it fairer for low computer spec players and easier for the server, but people with highend GPU's won't gain any dis/advantage from prettier water and if they want to speed the game up they can do it without CryTechs help.