Follow along with the video below to see how to install our site as a web app on your home screen.
Note: This feature may not be available in some browsers.
Moloch said:6600GT is slightly faster, features= no.
Ati can do dynamic brancing and geometry instancing, better than nvidia apparently.
But if things don't change, the 6600GT is a tad faster, but I think there's some nice gains to had with catalisy A.I, not to say nv wont do the same, however.
Hmm..rancor said:No they can't dynamic branching, it is not just for lights, and thats all ATi can do, its not true dynamic branching since ATi can do 512 instructions per pixel shader they just did a combanition shader of 3 lights in one pass, Sm 3.0 complient hardware can do 4 lights and more if special measures are taken.
GI is capable by ATi but can be done by any card sm 1.0 and up if programmers to find this one out but it can be done. In Ogl it is much easier to use with Vertex Bound Objects.
Moloch said:but I think there's some nice gains to had with catalisy A.I, not to say nv wont do the same, however.
No shit stupidR1ckCa1n said:NV has been doing it all along, my friend. Take a looks a NV's driver ini and see upwards of 20 games it does shader replacements/tweaks.
Moloch said:No shit stupid![]()
R1ckCa1n said:NV has been doing it all along, my friend. Take a looks a NV's driver ini and see upwards of 20 games it does shader replacements/tweaks.
You realize when one firm starts optimizing for a game, the other firm does the same right?R1ckCa1n said:errrr.... here's what you typed, son "not to say nv wont do the same, however"
You are implying NV doesn't do it now.
You dont need proof.. its fucking nvidia for christ sake.rancor said:Give us proof before you say that![]()
Moloch said:You dont need proof.. its fucking nvidia for christ sake.
You dont need proof when it's generally accepted.
rancor said:Give us proof before you say that![]()
Generally accecpted by those you realize nvidia have been doing this kind of shit for ages.rancor said:Its general accepted by whom? does futuremark come out and say there is shader replacement goin on, does Epic say this, does, Blizzard say this, does Id say this, I want to see the words from the people that working with nV say this. and backed up with proof of differences with screenshots and shader analysis, Take doom 3's shaders and run them through a shader analyser, and show me this.
R1ckCa1n said:Interesting stuff... so my 61.77 driver has the following profiles courtesy of Nvidia Corp....
Medal of Honor
Alice
Icewind Dale
Quake3
Diablo II
4x4 Evo 2
Splinter Cell
Halo
Tomb Raider
Wolfenstein Enemy Territory
Jedi Knight 2
RTCW
RTCW MultiPlayer
Soldier of Fortune 2
Soldier of Fortune 2 MultiPlayer
Jedi Academy
Jedi Academy MultiPlayer
Call of Duty
Call of Duty MultiPlayer
X29
FarCry
Splinter Cell - Pandora Tomorrow
Splinter Cell - Pandora Tomorrow MultiPlayer
The Westerner
PS: are these all "TWIWMTBP" (whatever) games ?
Im sure nvidia isn't doing any kind of optimizing for those games when you dont have them onrancor said:You have to turn thos on manually!
Moloch said:Im sure nvidia isn't doing any kind of optimizing for those games when you dont have them on![]()
rancor said:That is your believe until I hear proof from a notable source they aren't and thats my belief
R1ckCa1n said:True, nV has never done anything like that before.![]()
ati did that in 3dmark 03 way back, no one complained about the iq, just that they were replaceing shaders.rancor said:They would have been caught its not that easy to change shaders and get exactly the same results. Just do a pixel by pixel analysis, it will show up somewhere.
Moloch said:ati did that in 3dmark 03 way back, no one complained about the iq, just that they were replaceing shaders.
They do it doom3"In Doom 3, there's actually a bit of shader replacement going on. For example, the game uses texture lookups to determine the values of specular lighting fall-off rates, essentially reading a precomputed value by fetching that value from a texture. It just so happens that ATI's hardware can compute the algorithm faster than it can fetch the answer from a texture, so that's what it does: It replaces texture lookups with the math it approximates. The result should be identical, but with a significant speedup."
Looks better in some parts.