PDA

View Full Version : 8800 series driver bug....


therapture
12-15-06, 01:58 PM
...add to the list :D

http://www.hardforum.com/showthread.php?t=1133120

http://pages.suddenlink.net/therapture/pics/riva.jpg

skoprowski
12-15-06, 02:25 PM
Man am I disappointed...:( Talk about a big step backward for Nvidia in driver quality. Lets hope this gets corrected soon.

tacos4me
12-15-06, 03:13 PM
This could explain performance going down the toilet shortly after installing drivers. Should probably test to see if the domains are set correctly just after installing a driver..

billos
12-15-06, 03:58 PM
It's just incredible to let so much big mistakes remaining in "official' drivers.

Cm'on : i'm an engineer working on IC's hardware design and idon't understand that a big company such as Nvidia could let so important bugs happening in their high-range product...



... hmm, oh yes : i finally understand why. This is because the only remaining competitor (R600) is not out yet. So why bothering for stupid customers who spent their 650 bucks while u can enjoy ur Xmas, yes why tell me !!! :thumbdwn:

Shocky
12-15-06, 04:27 PM
This has been reported on many forums but I've actually seen no evidence to show its effecting the memory speeds other then screenshots from riviatuner which could just be incorrect readings. You would think if 2/3's of your memory were clocked at only 400MHz this would show in games and benckmarks but it doesnt.

badboy64
12-15-06, 04:52 PM
Atitool shows all 3 of memory speeds running at 1003mhz as I set them up in Rivatuner. I am using the 97.44' drivers too.
http://img315.imageshack.us/img315/5307/untitledqt0.jpg

agentkay
12-15-06, 08:14 PM
badboy64, the atitool reports three different type of clocks: 2D, 3D lower power clock, 3D clock. These have nothing to do with the clock generators that the OP was talking about.

lowdog
12-16-06, 12:24 AM
Might be more to this than is being revealed.

Like for instance the actuall mem controler on the cards may not be able to work correctly if all 3 mem controlers are set to 800mhz and that may explain why after long sessions of gaming the cards 3D preformance would become crippled = the slow downs people were experiencing + graphical anomalies.

Perhaps Nvidia purposely slowed down the mem controlers so the card would at least prefom semi normal :( , how lame if this is a possibility.

therapture
12-16-06, 02:32 AM
badboy64, the atitool reports three different type of clocks: 2D, 3D lower power clock, 3D clock. These have nothing to do with the clock generators that the OP was talking about.


...nor does ATITool check the ram speeds like Rivatuner does in real time. MY own ATITool shows the correct speed...only it is lying.

Shocky
12-16-06, 07:16 AM
Well until I see evidence other then some dodgy readings I just ignore this bug because going from 97.02 to 97.44 i've notice no difference at all and i'm playing allot of games with 16xAA@1920*1200. :p

therapture
12-16-06, 07:22 AM
Well until I see evidence other then some dodgy readings I just ignore this bug because going from 97.02 to 97.44 i've notice no difference at all and i'm playing allot of games with 16xAA@1920*1200. :p

My readings dodgy? Man I bet in real life you get made fun of alot...'cuz you're a prick.

You obviously can't read a whole thread, or you would see that I can replicate the issue on MY system easily, and others have too. I get SLOWER BENCHES AND GAME SPEED using the 97.28+ drivers....it is repeatable and consistent. Maybe YOU don't have the issue and that is GREAT, but don't go calling other people's findings "dodgy" when you don't have a clue....

Shocky
12-16-06, 07:48 AM
My readings dodgy? Man I bet in real life you get made fun of alot...'cuz you're a prick.

You obviously can't read a whole thread, or you would see that I can replicate the issue on MY system easily, and others have too. I get SLOWER BENCHES AND GAME SPEED using the 97.28+ drivers....it is repeatable and consistent. Maybe YOU don't have the issue and that is GREAT, but don't go calling other people's findings "dodgy" when you don't have a clue....

Touché ! :p

Then show us the results from games and benchmarks... Thats really all I and some others need to see and i'll roll back, untill then they could be dodgy readings. :D

badboy64
12-16-06, 08:26 AM
Touché ! :p

Then show us the results from games and benchmarks... Thats really all I and some others need to see and i'll roll back, untill then they could be dodgy readings. :D
Exactly. I want to see some proven results. I did a switch back to the 97.02's and the performance was the same as the 97.44's. Now I will just go back to the 97.44's.

Scaramonga
12-16-06, 12:54 PM
This is just after installing the 97.44 drivers for the first time.......

http://img.photobucket.com/albums/v60/scaramonga/th_9744bug.png (http://img.photobucket.com/albums/v60/scaramonga/9744bug.png)

Was OK on 96.89's :(

Cyant
12-16-06, 01:31 PM
Odd that the speed is 396Mhz and the "good" reading tells you 792...

396x2 =792

I still think it's just a frequency read error of the software because if the memory would be that much slower I wouldn't get good framerates in 8xQ/16X AA 1600x1200...

If i do the rotating fuzzy cube in ATI Tool and change the memory frequency
360Mhz memory speed I get 550FPS
900Mhz memory speed I get 1100FPS
1027Mhz memory speed I get 1250FPS

I think rivatuner is just broken when it tries to monitor the memory speed with those drivers... I am using 97.44

therapture
12-16-06, 01:53 PM
Working on some setups now....fixing to do one more driver cleaning and reinstall....I think it just might be a driver bug caused by a bad install or conflict....I have the 96.89's working well and have some benches saved, after I test the 97.44's I will report back....

Scaramonga
12-16-06, 02:44 PM
Well......I just ran 3DMark06 on the 97.44's and gained 500 marks!! from the 96.89's :)

I think this is more a Riva problem than Nvidia.

aweigh0101
12-16-06, 05:56 PM
Just confirmed this bug. Thanks for the heads up, my benchies went up by quite a bit!

Now if only nVidia fixes the broken HQ mode for the next update.

therapture
12-17-06, 03:56 AM
Some users are seeing incorrect mem readings but benches are the same, I and some others, are seeing incorrect mem readings AND less performance...it's a very finicky bug that is not affecting all users and seems very hit and miss.

I have gone back to the 96.89 drivers for the moment and everything except Q3A works great. Q3A locks the video at the loading screen, displaying a black bare field, and requires a ctrl+alt+del to kill the process from task manager, but leaves the video gamma highly ****ed up, requiring a restart to get it back to normal. But I can wait on that till Forceware X is here, I am tired of trying to fix the issue with the newer drivers. I'll hold here till we have something better.