Nvidia caught green-handed
THE INTERWEB is up in arms with stories about Nvidia cheating in Crysis, all of which seems a little 2005 to us here at l'INQ.
Those of you with memories longer than goldfishes will remember the crazy days of 3DMark 05, where it seems like even that bloke down the road with the crazy dog was caught fiddling with benchmarks. Nvidia killed image quality, ATI lorded it over the green team before being caught doing the same, and the whole thing led a bit of a revolution in benchmarking and image quality testing. No longer was 3DMark the exclusive arbiter of system performance and a good job, too.
Now the claims are back, with the green team's latest beta driver adding performance in Crysis whilst seemingly nerfing image quality.
The hype? That Nvidia's driver QA team are back to their wily ways. The real story? Come on guys - the performance increases being talked about here are barely a few frames per second, which is hardly anything to write home about. More to the point, given the increasing focus on image quality the hardware benchmarketing sites have run with, what are the chances Nvidia would shoot itself in the foot like this?
Chances are, we'd suggest, that the beta driver is just that - beta - and we'd expect a good-performing, IQ-centric official driver in time for Crysis' full release on the 16th.
Storm in a teacup? On the internet? There must be some mistake. ยต
Share this:
ShareBusiness Intelligence Software
Customer Relationship Management (CRM)
Enterprise Accounting Software
Enterprise Resource Planning (ERP)
Enterprise Systems Management
Meanwhile I hear the new Need For Speed demo gives enormous input lag and bad FPS on people who report running 8800's and intel core2's.
Whereas people with ATI cards and AMD CPU's report smooth sailing.
Well, it seems that the green team turned down the rate at which Crysis updates its water reflections, from like every 3 frames to every 10 frames, at a considerable cost in image quality.

The whole story is here by the way:

http://www.elitebastards.com/cms/index.php?option=com_content&task=view&id=487&Itemid=29
My dog is not crazy!
The problem with this is that all the testing being done uses Nvidia's Beta driver versus the AMD/ATI WHQL drivers so they do have to point out problems in the Beta driver. Catch 22.
ALERT THE INTERNET!
I don't remember any cheating in 3dMark05. You probably meant 3dMark03 and GeForce FX.
They cheated with every version of 3dmark, and will with the new one too.
And they both (or should I say all) 'cheat' with gamespecific optimizations, you can set it to aggressive, plain oroff in the driver panels but reportingly 'off' should not be taken too literal.
Of course one man's cheat is another man's optimizing, it's all in the eye of the beholder I guess.

Your dog has a prion loose.
Good ole Ferret... Scattergun revisionism. So long as we appear even handed & tag both main IHVs, eh...? az is right. Major issues with nvidia throwing it's toys out of the pram with 3DM03 & DX9, where GF FX couldn't keep up. They had no choice back then as RV350 was faster than their flagship NV30/35. Application optimizations are standard fare for all IHVs, but Nvidia has to use those two floors of mass computing power for something - and it's not just regression testing...
Kind of funny that it happened right when the 8800GT came out. Plus that it only shows it high res where most review sites will do their runs at......
i remember cause i was tryin 2 set world record on 3dmark for the 5800 series... got second place which was nice though :) how i miss that card!

(fyi. it died cause i had thermal epoxied heat sinks to it and wanted to remove them, so in my hot-headed state i thought ripping them off would be "just fine"... but it wasnt lol but instead i got to see what the GBA pin pattern was like :-)
But they had to have done something.
I have a quad-core Q6600, with 2GB of DDR2 (2 sticks of 1GB) and a 8800GTS.
I was using ForceWare 158 something when I installed the Crysis demo, and it crawled along at slideshow speed.
A friend of mine pointed out the 161 beta and, although I don't generally install beta drivers, I tried this one.
And lo! and behold, I now have a smooth 27FPS, sometimes going up to 35, and that in 1280x1024 and all in High detail.
There's simply no way that I'm going to believe that a new driver revision is going to jack up my cards performance by 400%. A nice 40% would already be stellar, but ten times that ? Never done before, can't be done now.
They have to have cut SOME corners.
Since when was the InQ the mouthpiece defenders of companies?

Interesting you forgot the original FartCry.exe example, and relied solely on 3Dmark, as if nV didn't tweak the last Crytek game. 

How convenient/inconvenient that the 'performance driver' specifically recommended for reviews happens to be the only one with issues, just prior to the games launch and the launch of a competitor's product. 

Which driver were you preping your review with prior to the promise of a replacement?

Did nV come to you prior to Hanner's article? As a member of TWIMTBP was this issue unknown to them, and everyone else?

Seems like alot of excuses are being made for something that was added to the Crysis performance drivers, and yet no one denies that the performance was increased or that the image quality was lessened. 

So is the InQ now a Beta advocate, a whole lota betas and no WHQL, it's easier to explain 'the undefendable' that way, just saying 'oh we didn't know'. 

Of course this could have nothing to do with the flurry of reviews about to hit the web, where 1-2 fps mean nothing of course, especially when used for fps/$ comparisons of 'value' card. 

Seriously, where's Charlie or Fuad's healthy anti-party line skepticism of IHVs?

It's been a while since I've seen the InQ make a headline "No story to see here, move along, we don't have enough credible evidence to publish and opinion". 

>X~P