PDA

View Full Version : 6800 Ultra 106 Error Causes Crash


SDJASON
02-16-05, 04:18 PM
Ok... so my 6800 is cool and all, but it crashed a lot...

I enver really knew what it was, however recently Nvidia added an error tracker thing to the bios...

My computer crashes usually 2-3 times a day while gaming, very annoying....

The probelm is the crashes are completely random, they can happen at the beginning of a game, after 4 hours, and NOTHING causes them (no combo keystrokes, certain sounds, movements, actions, ect)

The thing is ive realized the error log has an ERROR 106 for every time my computer locks up...

This error is Stress Test Transition L2->L1 or something

Nvidia has no other info on it, nor does my manufacturerer (PNY)

Obviously something is fubaring with my card to cause these lockups, and even slight underclocking/stock speeds doesent phase it....

So im wondering what exactly this error is, and how to fix it... Please dont say cooling, ive already got a Delta Fan (190CFM) Blowing on it, ****ing thing sounds like a hairdryer...


Thank you in advance

~Jason

saturnotaku
02-16-05, 04:32 PM
Need your system specs, especially the brand and wattage of your power supply.

SDJASON
02-16-05, 04:46 PM
INtel P4 3.2E @206 FSB
2x1024 COrsair 3200 Ram @ 3-2-3-7
Abit A17 Mobo, latest Bios
Xp-90 Heatsink w/ AS5
Antec 500W psu 20A on the 12V


Is it related to a drop in the 12V rail, because i have noticed my mbm logs the 12V rail drops below ~11.8, as low as 11.76 as the comp locks up too?

saturnotaku
02-16-05, 04:59 PM
11.76 is still within normal bounds. Have you tried running without your system overclocked? Looks like you have 2 GB of RAM. Try backing off on your timings by setting your CAS latency to 2.5. 1 GB sticks tend to be far more sensitive to frequency and latency changes than 512.

Subtestube
02-16-05, 05:28 PM
Matter of interest, are your 2d and 3d clocks different? It seems to be crashing on a mode shift, and I wonder if it might be the shift from 2d to 3d or vice versa.

SDJASON
02-16-05, 07:12 PM
Happens Stock or OC'ed

Ive even underclocked both the cpu (and hence the ram) and the vidcard by a couple mhz

Still locks up and gives me the 106 error


I really just wanna know what this Error is...

Stress test Transit L2->L1
Ive right clicked on it for help and gotten nothing

Subtestube
02-16-05, 07:24 PM
Nope, I'm not talking about whether you're OCed/stock/otherwise - what I mean is, are your 2d and 3d clocks different? I'd post pics of the panel in Forceware/Rivatuner, but my desktop is currently elsewhere. So... are your 2d and 3d clocks the same? It's just I recall people having a lot of problems with early run PNYs (or possibly some other company), and being able to fix them, purely by setting their 2d clocks to be the same as the 3d ones. As you're doing a Level 2 to Level 1 transition (unintentionally), that sounds like the video card is shifting modes. I'd guess that it might be that the clocks are changing, and that that's somehow causing instability.

Of course, if the clocks already are the same, you can disregard everything I've said.

SDJASON
02-16-05, 08:48 PM
I leave it at no overclocking, however upon switching it to manual it defaults to 400/1100 for 2d and 425/100 for 3d, leading me to assume they are different even with "no overclocking"

I set both to 450/1.2 manual overclock and reboot..... still happens... So im assuming it has nothing to do with that...

ALso the full error message is

Error code 106: Silent Running: Stress Test Transition L2>L1

Silent Running??

If its relevant, Oc'ed or not i have the box for fan always running checked....

saturnotaku
02-16-05, 08:50 PM
Can you test the card in another computer? Did you disable AGP fastwrites in your BIOS or if that adjustment isn't there, with RivaTuner? Otherwise, I think you may have yourself a bum card.

OWA
02-16-05, 09:09 PM
I had to set my 2d/3d clocks the same through a bios mod instead of just setting them the same in the control panel. While setting them the same in the control panel lessened the issue, it didn't go away completely until modifying the bios. I also have a PNY 6800U.

Also, if you're trying to troubleshoot lockup problems, it's probably not a good idea to use overclocked settings when testing (i.e. 450/1.2). Try setting them the same at stock settings (425/1.1) instead.

YOGIx213
02-16-05, 09:38 PM
Had the same problem with my PNY 6800 Ultra. The card came 400 in 2D, 425 in 3D. So, whatever clock I set it at, I make sure both 2D and 3D are the same. Example, right now, I clocked my 3D to 440, so I set the 2D to 440 as well. I don't get the game crashes at all, now that they're set the same.

Subtestube
02-16-05, 09:46 PM
Just a quick note: Did a bit of googling on your behalf, and I turned up these two forum threads:

http://www.aoaforums.com/forum/showthread.php?t=29296&page=1&pp=20
http://3dgpu.com/forums/lofiversion/index.php/t2716.html

Quoting the replies of relevance:

"Its nothing to worry about. The thing is your using funny drivers (not extremely good ones i might add) and these drivers are unlocking developer and debugging tools that are otherwise hidden. The 6800 does a numbers of things in the background, stress testing itself, purging the frame buffer, speeding fan up and slowing fan down on order to make sure the thermal probe is working. This is just another one of the procedures the card follows on a daily basis. What troubles be is that you have water cooling on that card. This isnt a great idea on 6800s as the GPU isnt the only part of the card needing cooling. When the stock cooler is on, air flows through the heatsink, and over the fins on the PWM area of the 6800. When watercooling with a card like the 6800, or even the 5900, you MUST cool the rest of the card to some degree, probably a very low rpm 80mm fan blowing over the heatsinks on the rear of the card (i mean near the external power connector), failure to do so will result in an unstable card over time, and ultimately a dead card. make sure you take this into account."


"I have seen a few posts on here and other forums concerning the error in the error log that says "Silent Running: Stress test transition: L2 -> L1" or something similar, I have this erorr in my log a few times, but like others, I had no knowledge of any action caused by this error (clack screen etc...)

The reason for this error is whenever you overclock you Nvidia GFX card and the test fails to do a desired clockspeed in the driver overclocker or rivatuner, it puts this error in the event log, so basically , there is nothing to worry about, now if you don't overclock and have this error, don't worry about it unless you get problems too.

Thanx, nice first post eh?"

I don't expect they'll help, but it suggests that if you're encountering really serious probs, you might want to RMA the card, as your problems may not relate directly to this error.