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

Chris, i wonder if you could answer this for me.

I recently received my Asus 8800GT (the one with CoH:OF), and as such installed the drivers provided with it (not sure which driver version it was). I loaded up the Crysis demo and tried it at 1440x900 & Ultra High, and the performance was great, easily playable.

I then installed the very latest beta release, 169.04, and *zap* it was very unplayable, mouse lag was present, and i had to drop to High (performance on the previous drivers was AWESOME on High).

Could you explain why ? Were certain effects turned off in previous drivers ? Is there big performance improvements coming in future releases ?

Thanks in advance .

 

I know it's been done before, but usually it would result in a SIGNIFICANT performance difference, and the card series in question was behind its competition.

Neither is the case, and with such a minor bug (you really don't notice it in-game), I'd hardly call the driver team incompetent because of it. You get bugs in all software (especially ones called "beta") at some point.

You're kidding, right? You think websites wouldn't notice the glitches nutball posted? I think you're a little off your rocker.

I have tested Bioshock DX9 with 169.04 drivers, and saw no problems. The low-framerate animated textures the poster was probably referring to were always there.

I'd say that puts a lid on this debate.

 

Looks like nvidia has released a new

http://forums.guru3d.com/showthread.php?t=243026

Will EB follow up?

 

NOT UNTIL AFTER THEY BLOODY SWITCH OVER SERVERS ON FRIDAY!!

 

163.75? Aren't they old?

 

I'm sure we all wish you well with that!

 

They're older than the new beta and newer than the old WHQL. So they've fixed the new problem with a newer older set than the old new set that caused the new problem.

Got it?

 

Everyone just keep thinking happy thoughts at Quinn and sending him positive mental energy...I think we're really gonna do it this time.

Question; did nVidia change their preferred Crysis demo drivers then?

 

This driver is actually older than the latest beta driver. (163.75 WHQL vs 169.04 BETA)

I'm not sure why the Guru3D guy posted this. The above is neither found in the "Release Highlights" section on Nvidia's site nor in the release note for the ForceWare 163.75 WHQL. This is actually the "Release Highlights" for the latest beta driver, ForceWare 169.04.

 

You're kidding me, right?

 

Nope. The guy has apparently messed up with the highlight notes.

P.S: Nothing really funny in that though. It's just a mistake.

 

Are we sure about that?

The release notes for 163.75 say:

Tho I admit it would be very interesting if they fixed the issue that fast with an older WHQL'ed set. The obvious implication would be a configuration file somewhere rather than whatever really gets WHQLed.

 

Or is their "fix" just remove the optimization? Someone should test if this set has slower performance than the 169.04.

 

Are these differences understood?

GeForce
http://www.pcgameshardware.de/?menu=browser&article_id=619942&image_id=727870&show=original
http://www.pcgameshardware.de/?menu=browser&article_id=619942&image_id=727872&show=original
Radeon
http://sylphys.ddo.jp/upld2nd/pc2/src/1194353301762.jpg

GeForce
http://www.pcgameshardware.de/?menu=browser&article_id=619942&image_id=727871&show=original
http://www.pcgameshardware.de/?menu=browser&article_id=619942&image_id=727873&show=original
Radeon
http://sylphys.ddo.jp/upld2nd/pc2/src/1194353372350.jpg

 

Well, that's a separate but related question. WHQL is a process that takes some time. There is no way to my knowledge they could have submitted these to WHQL after the EB article and gotten them approved that quickly. So the fix was either in some weeks ago before they submitted them to WHQL and before EB reported on the issue. . . or the fix was added at the last moment after the EB report but does not impact WHQL status for this set.

Edit: Or the 163.75 release notes reference to Crysis is about something else entirely!

 

And what kind of Cryis/Geforce characteristics could be responsible for that, given that the GT has a lot more texturing ability than a GTS?

Or rather some problem with the test itself, given the stunning amount of other crysis benchmarks on the net that support your thesis?

 

The pictures are from a different spot though. The radeon looks terrible in the first set.

The second set is harder to tell it appears the trees look better on the driverbug in the second one, but the lighting is different.

 

>> But this doesn't make sense because the speed differential is within a few frames per second.

the amount aint important but the fact that it has occurred, true whilst this may just be a bug on nvidias part, personally it smells more of cheating.
ive been a fan of nvidia over ati (due to better nvidia's better devsupport/ quality drivers etc but this leaves a bad taste in the mouth, which will last a while, i ask them is the sake of a benchmark for a topical game worth it?)