Cannot Diagnose An Error - Experts Please Read!

Simba1

Distinguished
Jun 20, 2010
63
0
18,630
Hey everyone,

I was having issues with while playing games with my Diamond HD5850. Randomly lines would appear on the screen and a loud buzzing would occur, and I was forced to shut down my computer.

I ran so many tests, SO MANY TESTS, updated drivers, all that jazz.

Finally I sent the card back and got a replacment. Well, that didn't do it, and now I'm losing my god damn mind trying to figure out what the problem is.

For some games, like Source-Engine based (L4D2) or WoW full settings, it never really happens anymore - but it still does rarely.

But for Dead Space 2, it will happen, COMPLETELY RANDOMLY, ALL THE TIME. Or other demanding games. I may have been playing for 2 hours and all of a sudden it happens. There is no pattern.

I pulled my comp apart, cleaned everything with duster etc.

I ran prime95, checked H/W Monitor, all that stuff.

I got nothing.

It makes me think that its not my card, because some games run fine...I guess.

Does anyone have any suggestions?

Btw I was going to update stuff for mobo from http://www.biostar-usa.com/app/en-us/mb/driver.php?S_ID=482 but there are like 39073029 exe's in each folder, and I have no idea which to use as there are no instructions. Sh***y Taiwanese company.


Thanks.


AMD Phenom II x4 965 BE
Biostar TA890FXE Mobo
8GB Corsair 1600MHZ DDR3
Diamond HD 5850
Ultra 750W PSU
 
Power without a doubt, that buzzing noise is the chokes in either your psu or the card its self. Coil squeal is nothing to fool with unless properly addressed now I want you to do is check for the source of the noise and report back.

PSU or Card
 

Simba1

Distinguished
Jun 20, 2010
63
0
18,630
Hey guys,

Thanks for the response. Nforce, what do you mean the chokes? And what I meant about the buzzing was thru my headphones when the screen becomes a bunch of solid vertical lines, not the actual hardware itself.

As for another psu, I don't have another one to test it with, but I may just buy another one if that would fix it! :(
 


Sounds like there was some confusion as to what the problem is. Artifacts are the card but the buzzing through the headphones is a symptom of another issue that could be related. Do you have a older card to test your machine with? No need to buy a new psu yet.
 

Simba1

Distinguished
Jun 20, 2010
63
0
18,630
No I don't have another card, this was a replacement for one that did the same exact thing.

And when it crashes, like I said it buzzes thru the speakers while the screen becomes a bunch of solid vertical lines, which crashes the system as ctrl alt delete and nothing else works, so I'm forced to hold the power button.

I just don't think it could be another faulty card...what are the odds? Maybe I'll ask Diamond for another one, but I just don't think it is because it works most of the time, the error is just COMPLETELY random.
 
I have owned a few diamond cards, they have not made a custom card in what a decade now unlike the 90s and early 2000s ware they made both nvidia and ati as well maybe a 3dfx card or two. They are reference so they can't be total junk but I suggest that you bum a card from someone and give it a few trial runs. If it occurs with the known to be working card then it is your system. The buzzing noises are likely a symptom of another problem.
 

Simba1

Distinguished
Jun 20, 2010
63
0
18,630
Well the screen just did it again after like 2 hours of gameplay. This time it went tan and I was listening to an audio log, so what happened was:

Was playing, listening to audio
Screen went tan with the vertical lines
Audio cut off and buzzed as like repeating the last word of the audio log but super fast so it resembled buzzing


ahhhhhhhhhhhh what else could it be
 
The vertical lines on the 5000 series card is usually a symptom of the VRM's overheating. They are not monitored and run much hotter than the GPU itself. The VRM's that overheat are regulating the memory voltage. Try underclocking the memory, that usually stops the problem.
 

Simba1

Distinguished
Jun 20, 2010
63
0
18,630
bystander, can you give me more on that? is that mean voltage regulator module, on the card? So you're saying its the card itself?

and by under clocking do you mean my RAM? or my card?
 
I have a similar problem with my PII X4 940/ Sapphire 5850. I've only seen it in Mafia 2 thus far. I haven't had much luck either diagnosing. I'll monitor temps on my card next time, however this card worked great for a long time and only started having issues when I moved to the PII computer, so in my case I don't think it is the card.

Here's one way to see if it is a similar problem. Can you ctrl-alt-delete and exit to crashed program? And if so, if you try to start another game, does it crash again immediately? Or do the random lines and buzz completely lock up your computer?
 


I'm referring to the video cards VRM and VRAM.

There are 3 VRM's controlling each VRAM chip, and they get hot, very hot. They aren't monitored by the software so you can't easily see if they are overheating. However, I've found that whenever I have this problem, backing down the cards memory clock has removed the issue for my 5870's.
 
Are they in the main power circuitry (red circle) or are they next to each ram module? I know the mains are cooled, albeit indirectly with thermal pads. Other might not be cooled at all.

http://imageshack.us/photo/my-images/90/gpufullassemblyrender6w.png/
 
Well if it is an overheating vrm phase for the vram remove the cooler and apply an extra but very thin and gooey thermal pads with the existing pads there or thicker pads to replace what is already there for better cooling. Why you would ask, well you want more pressure to be exerted onto the mosfets so that heat can more quickly move between the mosfets in the vrm phase to the mid plate thus lowering temps. Also manually set your fan speed to a higher level or create more aggressive profiles for better cooling.
 

Simba1

Distinguished
Jun 20, 2010
63
0
18,630
I monitored my card heat last session in a less demanding game, got up to 65C, 149F.

That seems pretty hot, but I have tried manually adjusting fan speed to, like full, and that didn't work either.

That model is of a 5850, which is shorter than the 5870.

EXT - No I cannot ctrl alt delete.

Once the screen has turned and the buzzing from the speakers starts, nothing is functional and the entire system is crashed and I'm forced to restart the comp completely.

Nforce - I don't know if I'm that good to take it apart and add stuff like that haha

Maybe I will try backing down the memory clock
 

Simba1

Distinguished
Jun 20, 2010
63
0
18,630
Also guys,

In catalyst under AMD overdrive, this is my current clocks:

High Performance GPU clock settings: 865MHz (out of 875)

High Performance Memory clock settings: 1095MHz (out of 1225)


Should I just bring down the memory, and to how much? And will it affect graphics and performance?
 
65C is actually pretty darn cool for a 5850, but the VRM's are not being monitored, so it's possible the thermal tape used on them is in bad shape. When I investigated my issue with a 5870, I found the thermal tape over my VRM's had dried up and shriveled up and provided no cooling. I'm not saying that's the only cause. A faulty PSU would likely cause a similar issue.

If you wanted to try out underclocking the memory, try 1050. At least for mine, a 50hz underclock worked out the problem.