SKATAN
Power Member
Mas só no 3dmark2003
[sarcasmo] é só o que interessa não é ?[/sarcasmo]
Cheaters
http://www.extremetech.com/article2/0,3973,1188167,00.asp
A ati tira as "cheates" do 3dmark2003 e a Nvidia cada vez dá lhe mais forte ... lol
como será que se portam com o antidetector script do unwinder ?
podem sacar aqui ftp://evga.com/4467_XP.exe
se calhar são a mesma coisa que os 44.65
[sarcasmo] é só o que interessa não é ?[/sarcasmo]
Cheaters
New Detonators boost 3DMark 2003, nothing else
IN A MOVE that could easily re-ignite the still-simmering 3DMark 2003 controversy, the latest leaked Detonator drivers (44.67) dramatically boost performance over 44.03, but without containing any of the same "features" that boosted performance in that particular Detonator series.
In fact, according to ExtremeTech, the result boost is impressive, with a full 26% performance jump overall and a nearly 100% leap in Game 4. Since no evidence of driver tampering has been found you'd think it was all milk and honey for Nvidia — but unfortunately that's not the case.
Despite the mammoth performance jump these new drivers deliver in 3DMark03, we see no such corresponding increases in any so-called "real-world" tests. The term "real-world" itself is something of a misnomer when applied to benchmarks, as it implies that benchmarks run using "real" game engines somehow function differently then benchmarks run using "synthetic" tests. For the record, tests like Sandra (that perform an invisible set of calculations and spit out a calculated score) are purely synthetic, while others like Sysmark or Winstone are a blend, providing a calculated score, but one based on performance in real-world applications. In this case, the biggest difference between 3DMark 2003 and a "real-world" game is that the engine 3D2K3 uses isn't used in any shipping game.
So, we're right back to the question of how much this matters and whether or not anyone should care. Is it alright for manufacturers to highly optimize for one specific test and claim performance improvements, even if those performance improvements aren't born out anywhere else? It probably is — the worst thing that'll come of this, after all, is that ATI will whip its own new optimizations out — but NVIDIA would look much better if they could provide some significant speed boosts to games as well.
3DMark 2003, despite not being entirely synthetic, is not a game. There's no "reason" to optimize for it, save to attempt to make your product look better in a review. At least when optimizing for Quake or Unreal Tournament 2003, Nvidia can claim that they aren't just boosting how their product looks, they're improving the experience of their card users.
If there's one thing this battle has demonstrated it's just how arbitrary supposedly reliable benchmarks can be. Want to increase performance? Toss in a tweak there, a tweak here, a nudge there, a clipping issue here—and voila, card performance appears to have doubled while real world speed gains might not have budged one damn millimeter. So how about it, Nvidia? If you can double performance in certain 3DMark 2003 demos and offer a 26% improvement across the entire benchmark, can't you raise standard game performance at least a little? And if you can't, doesn't that imply that your 3DMark 2003 performance boost is ultimately worthless, since no other product can back it up? µ
http://www.extremetech.com/article2/0,3973,1188167,00.asp
A ati tira as "cheates" do 3dmark2003 e a Nvidia cada vez dá lhe mais forte ... lol
como será que se portam com o antidetector script do unwinder ?
podem sacar aqui ftp://evga.com/4467_XP.exe
se calhar são a mesma coisa que os 44.65