nVidia 169.04 -Crysis Rendering bug gives 7% performance boost?

I guess you haven't been around long if you don't know the lengths companies will go to in order to win benchmarks on big triple-A games. I'll say it again: we've seen this plenty of times before. Marketing is king in these situations. A few percent here, a few percent there, and it all adds up.

I'm sure Nvidia doesn't consider it "swindling", I'm sure they consider it simply good marketing and business practise. That's what special review drivers for particular games are for.

Or are you suggesting the Nvidia driver team is so incompetent they release a special driver for a particular triple-A game and it doesn't even render everything that's supposed to be on screen? And it's just a fluke that it happens to make the game go faster?

 

*cough*

That's with the drivers which were pointed to by the nzone page when I downloaded the Crysis demo. Seems to me we've gone from ROFLcopter to Black Helicopter in the space of 10 days.

 

If it gets the benchmark wins, correct visuals are secondary.

 

The 163.75's are fine in Crysis, use those.

Oh, sorry. That's not what this thread is about I guess....

I'm too old and tired to compete in pissing contests on teh wheb these days.

 

In conspiracy theory terms I'd expect Nvidia will fix whatever that code is AFTER benchmarks have been run and reviews done.

Granted a few sites might note what is going on, but they'll have some disclaimer that they contacted Nvidia and they are aware of the problem and that it will be fixed in a future driver release. But, faster is faster... Especially when most people just look at the graphs...

/conspiracy_mode off

Also, I think someone mentioned that water reflections are also adversly affected in Bioshock. Meaning it's something in the driver and not in adjustments that Crytek have made to Crysis. Can anyone else verify this is happening in Bioshock?

Regards,
SB

 

this is fixed in a driver revision ((thats only .o1 or .o2 digits higher than the 169.04 driver series. I cant reveal the actual driver number as it may change by the time it goes public)). But performance looks to be largely unchanged.

Chris

 

It will be nice to see EB's followup then.

 

What, only around 7% or so?

 

I am thinking this is just a glitch that will end up causing no real change in the numbers when its fixed. Much to do about nothing I think.....

Oh to add I could care less who blows the wistle on who these days. Both IHV are looking closely at the other and both are ready to jump when they spot something that does not add up. You could almost think of it as each doing QA for the other

 

No.

 

Great news - Thanks for sharing Chris!

I've front-paged that news, linking to this post, on Elite Bastards (not that most of you can get to it probably!) - Hope you don't mind.

 

Hehe, if it results in better quality drivers by each IHV, I hope they do continue to blow the whistle on each other.

Regards,
SB

 

I havent been able to access EB for days now.

 

No. 1 fps-->2 fps is a huge difference digi far more than 7% and it means nothing, both are worthless.

Percents mean jack if you don't include the actual numbers.

This seems a storm in a teacup, a gale, a crysis perhaps

 

Don't worry dig. There are numerous optimizations for each key title in IHV drivers that are constantly tuned with each iteration. So the slack will be made up elsewhere as part of the normal regression testing & optimization process. Indeed, performance may well improve with the subsequent - see, we told you so.

 

Nvidia seems to think they do.

 

God those graphs always make me laugh... It's like those ads for videocards where they show Videocard A as being '80%', then show their overclocked card at '110%' and go "Our card be the givens you the 30% more of performance!".

...Sure, ok

 

So you are saying Digis statement is as reliable as Nvidia PR? That is low man, real low.

 

That is what I got on my Radeon when I tried to run AA with shaders at anything less than high.