GeForce 6800 Ultra freezing!! *resolved*

[Chippy99]

Well I don't know about the OP, but this problem is certainly NOT "resolved" for me.

I bought a BFG GF6800GT some while back, fitted a VGA Silencer and did the 1.4v bios mod. I found it was a truly excellent overclocker, reaching 453MHz stable (but with the occassional artefact) and - more sensibly - 440MHz completely rock solid and problem free. This was all in my old DFI Ultra Infinity nf2 board.

A week or so ago, I upgraded to an MSI K8N Neo2 and Athon 64 3200+. Everything works perfectly apart from this single problem: the GF6800 lockups.

I cannot run more than 400MHz core on my GT without getting these lockups. I know I shouldn't complain about "only" 400MHz from a GT. But a 40MHz drop is quite a lot to have to stomach.

I have tried *everything* to fix this.

1. 2d & 3d the same, coolbits
2. 2d & 3d the same, edited the bios.
3. FW off
4. Sideband addressing off
5. AGP aperture 128, 256 & 512
6. AGP speeds 66, 67, 68MHz
7. CPU speeds right down. (I even tried running 5x200MHz CPU, and it still happens! Or I can run 10x260MHz CPU. It makes no difference).
8. different nvidia graphics drivers

None of the above had any effect.

But here's the fascinating thing. If I run 440/1120 at try something like 3dmark01, I will get a lockup *instantly*. It will then sit there, locked, for 10 seconds and then spring into life and continue.

However, I *never* get any lockups at all - at any speed - in Doom3, Quake3, Quake2, Quake. (I have tested all of them). This problem *only* occurs in Direct3d, not in OpenGL. Has anyone else noticed this?

Also, very occasionally - its happened 3 times now - instead of the classic 10-second-lockup, I get a BSOD STOP Error. 0x000000EA.

This is the good-old-fashioned nVidia Infinite Loop error. Remember those????

I am wondering if at the bottom of all of this is a low-level chipset driver bug (or graphics driver bug)?

Chip

 

[RampantAndroid]

dunno, I really have no idea what to do...maybe some of these cores are no good?
did you up the AGP voltage?

As for OCing the 6800, my 6800U can make it to 479 MHz 3d core, and 1.27 GHz mem and be stable.....

 

[Slappy00]

I had crashes/lockups in 3dmark05 and Joint ops as well as some other games, only when running 400/1100, it ran fine at 375/1100 (as a side note if your system is stable at a lower clock speed then you might want to look at power supply/consumption, particualraly if you have a prescott (or even a northwood) overclocked). IMO stability at lower clock speeds but instablitiy at higher clock speeds can be caused only by a few things: Lack of power to stablize the GPU at higher speeds, heat buildup, non performing core. The fast-writes thing really only helps with AMD chipsets as Intel has better integrated the feature, although it doesnt hurt to try.


My solution was quite simple, even with my antec 430w truepower my 12v line was sapped due to 3 HDD (one of which was a raptor), two optical drives, a Koolance Exos, and various fans. I went and did a few things: updated my video drivers to 67.42(beta) using driver cleaner, then i removed all my prior ATI ref using driver cleaner. Then i unplugged one optical drive and a fan or two and gave teh GF 6800GT its own 12v power line. Seems to work great now at 400/1100. I have an OCZ power stream on the way from the egg that has a 33A (396W) 12v power line as well as beaing a 550W power supply.

I always recommend trying the ceapeast solutions first, but sometimes you just need to upgrade something else to get that new piece of hardware happy in your ssetup.

 

[IdahoB]

It definately seems to be connected to running on an Athlon 64 - which SUCKS. I eventually got my GT to run after buying th same Neo2 Platinum and 3200+ but I think I'm one of the lucky ones for whom fast writes disabling worked. Good luck finding a solution.

 

[Chippy99]

Thanks guys,

I forgot to say, yes I tried AGP voltage. I run 1.6v anyway, but I did try 1.7v. Didn't try 1.5v - didn't see the point. I ran 1.6v old my old board. I have always run 1.6v AGP.

With respect to the power issue, well I have a OCZ Powerstream 520w PSU with the card on one of the dedicated lines. 12v line is at 12.24 (tested with a multimeter) and it never budges. I know that's a little high, but I can be bothered to adjust it and anyway, its within limits. BIOS reports 12.02v.

Also, I have a Winchester chip that uses much less juice than my old XP 2500 at 2400MHz. And my ram (PQI Turbo ram) is running at 2.65v and is luke warm, so I don't think that's a big power drain either. All other components are the same as in my old rig, and that worked fine with the same PSU. So all in all, I don't believe its got anything to do with the power supply.

I agree with IdahoB: This is definately a Athlon-64-centric problem. And it seems to be worst with A64's and nf3 chipsets. I don't think the K8T880's are quite so badly affected.

But its still a mystery why it doesn't happen in OpenGL. Hmmmmmmmm.

Chip

 

[Slappy00]

"But its still a mystery why it doesn't happen in OpenGL. Hmmmmmmmm. "

maybe its the way DX9 handles requests to the memory and video card?

Also chippy how much amprage does the ps put out on the 12v line, when comapred to how much you have loaded on it? Although I didnt use a multimeter (all molex connectors are used in my setup :/) I used the much less accurate Abit EQ monitor and it seemed a bit low 11.8v, but still IMO well within the acceptable range. The lockups still happened but like you mentioned not with Doom3 or with rthdribl (google it its a good DX9.0 stress test) , only 3dmark05 and a few other games.

If your card is stable at a lower clock speed then I would tend to think it is a hardwrae / heat issue as software incompatibilites appear at any speed. Of course my first suspicion would be the power supply, but since in your case that does not seem to be the problem I woudl go and start tweaking my BIOS.

Start with the proven "turn off fast writes" , its really an Intel thing and it realyl doesn't help much in AMD setups.
Thats the easy solution if that doesnt work turn everythign off that you can without making your computer not boot to windows (and go batty). Then slowly start turning everythign back on or up (one at a time). and see what option causes this to occur (assuming that turning everything off solves your problem in the first place). Yes it sounds tedious but in the end if it is a BIOS feature/ mobo issue you will save your self some headaches dealing with the video card RMA dept.

 

[BlackEvo]

Originally posted by: randomx
I have an MSI Neo 2 Platinum and recently got a PNY Geforce 6800. I was having problems with the game hanging in all DirectX 9 games... UT2004, Doom 3, Far Cry, etc. Nothing drastic, just pausing for 10-15 secs. every 30-45 secs. Frustrating. After updating all drivers, reading this thread, and checking Nvidias recommended bios settings, I found the problem. In the bios, I turned off the Fast AGP write, set the AGP speed from its default of "Auto" to "4X8X", and set the AGP aperature to 512MB. (Nvidia says to make it half of the system RAM, but there was no option to set it to 1024MB)

Result: everyting so far runs extra smooth.... Try it out, its a quick fix.

Ok, I just downloaded Powerstrip, so I'll try that. I have the SAME mobo as you with a 6800 Limited Edition from Newegg. This thing runs like crap when I use 3DMark05, and restarts my computer after the first test. I have an Antec 550W PSU, which I beleive is good enough. There are seperate power connecters going into my 6800.

I'm going to try that, but somehow I think maybe my OCZ Gold Rev 2 PC4000 RAM might be an issue if this doesn't work. I bought the RAM from someone here so they will hear from me if that's the case :|

BlackEvo

 

[BlackEvo]

Originally posted by: HITechJeff
RampantAndroid:

My power supply is listed in my sig, it's the best there is.

SOLUTION TO PROBLEM FOUND!

I have found and 100% FIXED the problem I was getting with the video and audio freezing. Please note that I would HIGHLY advise anyone BEFORE flashing their BIOS on any video card to try this first. For me and some others, the solution IS NOT the drivers, video card problem, etc. Here's the solution:

Go into main system BIOS.

Turn "Fast AGP Write" to OFF/Disabled

Make sure "AGP Speed" reads "4X8X" and NOT "Auto" (seems to be issue with "Auto" detect here)

Turn "AGP Apature" up as high as possible. Mine was set at 256MB, but changed it to "512MB" the highest setting possible.

Save and exit BIOS.

Well, that did it for me. Now, ALL games run SILK SMOOTH on MAXED out settings and 4XAA! Weeeee!!! I have frame rates up to 71.8 on Far Cry at 1600X1200. Half Life 2 ROCKS also with MAX settings. Everything now is great so it seems like it was just some BIOS tweaks that was needed. Too bad tech support doesn't go over them in more detail or else I could've solved this long ago. At least I know what to check now when this happens and maybe it can help others too. I also think it would be helpful if people would start posting their detail BIOS settings AND their system specs together. Otherwise the solution may be right under their nose (like me) and just not realize it.

Thanks for the replies,

Jeff

Hmm, that's what the last guy I replied to said. I'll try that too...

BlackEvo

 

[chinkee]

Originally posted by: BlackEvo

Originally posted by: HITechJeff
RampantAndroid:

My power supply is listed in my sig, it's the best there is.

SOLUTION TO PROBLEM FOUND!

I have found and 100% FIXED the problem I was getting with the video and audio freezing. Please note that I would HIGHLY advise anyone BEFORE flashing their BIOS on any video card to try this first. For me and some others, the solution IS NOT the drivers, video card problem, etc. Here's the solution:

Go into main system BIOS.

Turn "Fast AGP Write" to OFF/Disabled

Make sure "AGP Speed" reads "4X8X" and NOT "Auto" (seems to be issue with "Auto" detect here)

Turn "AGP Apature" up as high as possible. Mine was set at 256MB, but changed it to "512MB" the highest setting possible.

Save and exit BIOS.

Well, that did it for me. Now, ALL games run SILK SMOOTH on MAXED out settings and 4XAA! Weeeee!!! I have frame rates up to 71.8 on Far Cry at 1600X1200. Half Life 2 ROCKS also with MAX settings. Everything now is great so it seems like it was just some BIOS tweaks that was needed. Too bad tech support doesn't go over them in more detail or else I could've solved this long ago. At least I know what to check now when this happens and maybe it can help others too. I also think it would be helpful if people would start posting their detail BIOS settings AND their system specs together. Otherwise the solution may be right under their nose (like me) and just not realize it.

Thanks for the replies,

Jeff

Hmm, that's what the last guy I replied to said. I'll try that too...

BlackEvo

hm...does this apply to the 128mb geforce 6800 OC as well? or is it just for the 256mb 6800 ultra? my system is unstable even when trying to run 3dmark. its a mobile 2400+ @ 2.2Ghz w/ 1 gig corsair xms ram. i turned fast writes off, agp speed to 8x, and the agp apature to 128mb. however, my powersupply is a antec smartpower 350w. what is most likely the source of my problem?

 

[BlackEvo]

Well, that helped, but it didn't fix the problem. I just talked to eVGA and I am RMAing this headache of a card and using the step up program for a REAL 6800 Ultra. I get a copper heatsink fan and $50 eVGA.com bucks. Plus free shipping of the card to me.

Not a bad deal. I do have to pay the difference from the invoice cost of my LE compared to the Ultra ($150) and shipping to them (like $5). All in all it's a nice little package, but only because of the copper heatsink/$50 eVGA bucks.

BlackEvo

 

[Chippy99]

Originally posted by: Slappy00
"But its still a mystery why it doesn't happen in OpenGL. Hmmmmmmmm. "

maybe its the way DX9 handles requests to the memory and video card?

Also chippy how much amprage does the ps put out on the 12v line, when comapred to how much you have loaded on it?

34 Amps enough for you?

Although I didnt use a multimeter (all molex connectors are used in my setup :/) I used the much less accurate Abit EQ monitor and it seemed a bit low 11.8v, but still IMO well within the acceptable range. The lockups still happened but like you mentioned not with Doom3 or with rthdribl (google it its a good DX9.0 stress test) , only 3dmark05 and a few other games.

I do get them in rthdribl. Perhaps not as often, not sure.

If your card is stable at a lower clock speed then I would tend to think it is a hardwrae / heat issue as software incompatibilites appear at any speed. Of course my first suspicion would be the power supply, but since in your case that does not seem to be the problem I woudl go and start tweaking my BIOS.

I think we can safely rule out heat as the cause. Here's why: With my old rig - and my XP2500 at 2500MHz chucking out a load of heat, my GF6800GT idled at around 50c and peaked at high 60's. Not bad. But in my new rig with my cooler Winchester CPU at 1.45v only, the GT idles at 46C and peaks mid 60's. Yet it ran 440MHz before and won't do 420 now. Conclusion: Heat isn't the problem.

Start with the proven "turn off fast writes" , its really an Intel thing and it realyl doesn't help much in AMD setups.
Thats the easy solution if that doesnt work turn everythign off that you can without making your computer not boot to windows (and go batty). Then slowly start turning everythign back on or up (one at a time). and see what option causes this to occur (assuming that turning everything off solves your problem in the first place). Yes it sounds tedious but in the end if it is a BIOS feature/ mobo issue you will save your self some headaches dealing with the video card RMA dept.

Done all that. Nothing fixes it.

I can't RMA it. Its only supposed to run 370MHz. I can't RMA it because now it will only do 415 when it would do 440 before. I think BFG would (rightly) tell me to get lost. And - apart from anything else - there's a VGA Silencer on it now, which kindof stuffs the warranty.

But none of that matters. I am simply trying to find out what the "problem" is. I have no desire to RMA the card anyway: generally I am very pleased with it.

Chip.

 

[Laza]

Well I wouldn't go blaming AMD for making incompatible CPUs or motherboard makers for making bad nForce2/3 boards. There's definitely something fishy about the 6800 series cards, and I'm amazed that nobody has officially taken this up. Too many people experience the same problems with a wide variety of systems. Probably it won't be easy for nVidia to fix this, it probably would require new silicon. But companies have pulled back products before, I think we should be entitled to new gfx cards as well.

 

I am having the infinite loop error. But my case is very strange.

Specs:
Pentium 4 2.4 GHZ Processor
512 MB Ram
80 GB Harddrive
Nvidia GeForce 4 TI4200 with 8XAGP
ASUS P4PE Motherboard
Windows XP Home with SP2

Here's the story:
My computer is about a year and a half old. Ever since day 1 my video card has been running everygame I played like a charm. Now, I am in a computer class (only in 10th grade) and during the past couple of days we took apart the computers we were using in class and had to rebuild them. I rebuilt mine sucessfully, and I decided to take a look at my home computer. I popped the case off and too a look inside of my own PC for the very first time. I noticed only 1 RAM stick, so I decided to find out what kind of RAM it was incase I ever wanted some more in the future. Anyways, the NVidia card was blocking the one bracket on the ramstick, so to remove the ramstick I had to remove the video card. I made sure that I let all of the static electricity off of my body by touching the case, and then removed the video card and ram stick. I then put the ram stick back in and the video card back in without any problems and turned back on my PC.

Here's the problem:
My computer went to the Loading Windows screen and the bar was moving fine. After the loading sequence, the screen went black for about 30 seconds and the Windows restarted. I went into Safe Mode and turned "Automatically Restart on System Failure" off and then restarted my PC. It did the same thing, but this time after the blank screen for 30 seconds an error came up about the INFINITE LOOP and some other things about the nv4_display drivers and all that. So anyways, I tried downgrading to older drivers and the same problem was occuring. I finally decided to just uninstall the drivers altogether and now I can get into Windows fine, but I am using the graphics built into the motherboard :\. So basically, I can't play any games now or nothing.

My Question:
Do you think that somehow when I removed my video card I sent some static electricity to it and fried it?
Perhaps it's a problem with the ram?
Do you suggest that I pay 39 dollars for tech support to take a look at it and then additional fees for whatever they have to repair?

Like I said, I'm only a 10th grader with intermediate computer skills, and I am very low on cash. Any help would be appreciated.

 

[chaonatic]

You didn't fry your card, don't worry. You prolly should just reseat your RAM and card and make sure they are all the way into their sockets.

You didn't touch anything else?

 

Nope didn't touch a single thing other than the video card and RAM. By reseat, do you mean take them out and put them back in? I have tried that and the problem still wasn't fixed.

Another thing I noticed is that inside my PC there is ALOT of dust. Perhaps some of the dust got into the ports and are causing problems? What do you recommend me doing to clean out my case?

 

[AssyrianKing]

Darlord,

First of all your chances of damaging hardware due to static electricity are EXREMELY low and RARE, Ive been working with computers/notebooks for 6 years taking them apart and building them without using static matts or and form of de-static procedures i dont even bother touching the case or anything. You dont need to take these precautions unless you work under a factory where large generators,belts,machines,etc cause alot of static electricity in the area.

So now you know about static electricity and its just exagerated but dont forget it altogether. Touching your PowerSupply before anything is good to do on cold/dry days.

The problem your having with windows blanking out could be due to your system/windows being corrupt from pulling out your ram and puting it back in, i dont know why this is, but ive had it before where i pulled ram out and thats all and put it back in its place and my windowsxp wouldnt boot saying its corrupted, i had to reinstall. (maybe it was due to my system being overclocked.)

The only option i can think of is reinstalling windows, also dust wont really cause problems unless you got that much dust built up on your heatsink of your cpu or card that its stopping it from cooling properly, it wont hurt to clean it but it wont fix your problem.

So reinstall windows or grab another hardrive if you have around and see if that boots into windows.

 

AssyrianKing,

I took your advice and reformatted my PC. It fixed my video card right up! Thank you for your help.

 

[Chippy99]

Originally posted by: SirDude
This is a stupid question, but what does "fast writes" do?

Its not a stupid question and I bet half the people here don't know ;-)

Enabling it allows your CPU to write directly to your video card's memory. If it is disabled, your CPU needs to write data to main memory first, and then it can be copied into the card's memory from there.

Everyone keeps saying you don't need it, disable it etc etc.

This is potentially duff advice. It depends. Enabling fast writes makes video playback considerably better and uses noticeably less CPU when FW is enabled. Especially for high definition WMV and DIVX material, that is not accelerated by 6800's right now and for which your CPU is doing all the work.

Also, any applications that do lots of AGP transfers will see significant benefit with fast writes enabled. Fortunately there aren't (yet) any games that fall into that category, so if all you want to do is game, then yes you can disable it.

Chip

 

[SiMbOt3K]

Yo I took some advice from a few posts on here and used the latest Riva Tuner program to adjust the 2d clock speed to equal the 3d clock speed in my ASUS v9999 Ultra Deluxe 6800 Ultra which i've had troubles with since day one. Finally this card is working i just gotta allow Riva Tuner to load settings each time i boot windows and this is some soft modding thats worked sweet and hopefully doesnt mess my warranties. Thanks guys.

 

[Navid]

Originally posted by: AssyrianKing
First of all your chances of damaging hardware due to static electricity are EXREMELY low and RARE, Ive been working with computers/notebooks for 6 years taking them apart and building them without using static matts or and form of de-static procedures i dont even bother touching the case or anything. You dont need to take these precautions unless you work under a factory where large generators,belts,machines,etc cause alot of static electricity in the area.

So now you know about static electricity and its just exagerated but dont forget it altogether. Touching your PowerSupply before anything is good to do on cold/dry days.

The static energy that you can collect while walking on a carpet, specially in a dry day (winter), is enough to damage a semiconductor device. You don't need large generators, belts, machines ......

One of the main problems with ESD, in manufacturing, is that it is not always easy to detect a device that is damaged by ESD. ESD damage is not always catastrophic. It can just degrade performance while functional pass/fail tests still pass!
That is why manufacturers take it so seriously and all the operators strictly follow protection against ESD procedures and practices, in order to avoid shipping products that fail in the field and return and damage the reputation of the company.

It is always a good practice to take ESD seriously. I would not tell anyone that ESD is just an exageration.

 

[RampantAndroid]

I must concur with Navid. Hospitals even have to protect from ESD...they used to use metal wheels on the beds and such so that they would always be grounded...of course, dust destroyed this clever little plan...I ALWAYS use a static wrist bad, and I ALWAYS use a static mat and I ALWAYS use static bags...this is a $500 video card we are talking about here (or perhaps my $800 FX-53...now 700 something prob b/c of the FX-55, but still) is it worth the risk? buy a $5 wrist strap and save yourself $500. Pretty good deal I'd say!

 

[AssyrianKing]

well i dont need antistatic matts etc, i kno how to handle my hardware and have never and will never damage my hardware due to ESD, but if i was working in a carpeted area i always take high precauitions, and yes its exagerated so people will take it seriously but generally if you know what your doing you dont need to do every little thing like puting on your little wrist band and your lil matt and your lil baggies for all your lil parts n chips. touching something grounded is more than enough.

-ESD topic is now finished enough talking bout it! back to 6800's fu*in coz there made so shit.

 

[RampantAndroid]

I beg to differ....you're only closing it b/c you cant win....ESD isnt something to take lightly, you're more dangerous than someone who knows nothing of it...they know better than to go findering the stuff without protection...think of it this way, with near 500 million transitors in an area less than a 1/2 square, think of how small each trace is...doesnt take much to destroy just one of those little traces...

when you are working with large scale ICs, like TO-220s and LM555 etc, no protection is fine, but in small level, you NEED IT. Trust me, as somoene who has worked with computers for years, and is currently working on both electrical and computer engineering degrees, I should know....you can never be too careful.

As far as the 6800Us go, well, dont trust PNY tech.....they'll corrupt your system by making you install lots of worthless programs....just play around with BIOS settings, power supplies, hardware set-ups, drivers and, at worst, clock settings...thats all there is to be said....that cover pretty much all feilds, short of RMAing the thing, which I am going to do soon with PNY.

 

[VirtualLarry]

Originally posted by: AssyrianKing
First of all your chances of damaging hardware due to static electricity are EXREMELY low and RARE, Ive been working with computers/notebooks for 6 years taking them apart and building them without using static matts or and form of de-static procedures i dont even bother touching the case or anything.

Originally posted by: AssyrianKing
The problem your having with windows blanking out could be due to your system/windows being corrupt from pulling out your ram and puting it back in, i dont know why this is, but ive had it before where i pulled ram out and thats all and put it back in its place and my windowsxp wouldnt boot saying its corrupted, i had to reinstall.

I find it rather humorous that you don't see a connection between those two issues. ESD is a very serious issue with computer parts. There is no reason otherwise why re-seating RAM (with the power totally cut to the system), that it would suddenly, magically, cause the data stored on the HD to be corrupt.

 

[RampantAndroid]

I find it rather humorous that you don't see a connection between those two issues. ESD is a very serious issue with computer parts. There is no reason otherwise why re-seating RAM (with the power totally cut to the system), that it would suddenly, magically, cause the data stored on the HD to be corrupt.

THis IS a good point, unless it was seated wrong, its hard to do RAM wrong.and you dont always notice ESD damage