Completely clueless, probably a crossfire issue, need help

crashadog

Distinguished
Nov 12, 2011
11
0
18,510
I built this desktop computer over the course of a few months, and it's been running for 3 months now. My problem has been persistent during this entire time that I've been using it, and I think it's finally time to ask for help in figuring this out. First off, here are my specs (nothing is overclocked):

ASUS M4A89TD Pro/USB3
Phenom II x4 975
Gigabyte 6870 HD (x2 for crossfire)
Emprex 21.5" LCD 1680x1050 60hz (Borrowing this until I get my own monitor)
G.SKILL 2x2GB DDR3 1600mhz (reads as 1333mhz in the bios, but it's not my main concern right now)
Antec HCG 750w
AZZA Spartan 102E

The problem is this: Whenever I try to utilize crossfire for gaming, within a (short) matter of time, everything goes haywire. The sound will go crazy (skips the last bit that got through over and over and over) while my monitor goes black, and then I'm told that there is no signal going to it. Then the sound cuts out entirely. I have no indication of whether or not I still have any input into the computer through my mouse and keyboard, and all I can do is hit the switch to power off the depressing way. I am able to boot right back up afterwards if I want to.

I've done a lot of experimenting and observing already, and here's what I know. I hope it's helpful.

-This is only a problem when I'm trying to use crossfire. If I disable crossfire, this NEVER happens. EVER.

-This issue only occurs under heavy load on the graphics cards. As in, either gaming or running a gpu benchmark (furmark).

-It doesn't matter whether the game I'm running supports crossfire or not.

-I have the latest drivers, the latest AMD Vision Engine Control Center, and the latest Application Profiles.

-When the issue occurs, my computer APPEARS to still be running. The CPU and PSU and case fans are all still spinning, the lights are still on... But I've noticed that the TOP GPU seems to be shut off. While the bottom GPU's fans continue to spin, the top's do not.

-The PSU is bottom-mounted, and is just beneath the lower GPU. I've tried turning the PSU upside down, to suck air in from the vent beneath where it's mounted. I thought perhaps the hot air from the bottom GPU was overheating the PSU. But nothing changed from this experiment.

-I've never seen the GPUs surpass 80C. Even during a furmark burn-in test, the issue pops up and I lose signal just as the top card is nearing 80C, which usually takes approximately 2-3 minutes.

-I recently updated my mobo bios to the latest version. I learned that my GPUs apparently have an official bios update available for them (right from gigabyte's site), but I have no idea how I would go about flashing them. I can't figure it out and I couldn't find a good guide to follow.

-I've tried swapping the cards in their top and bottom spots, and have even swapped out the crossfire cable wire thingy too. No changes.

I really hope someone can help me figure this out. Thanks in advance to anyone who spends some time with me on this.
 

clutchc

Titan
Ambassador
If, after swapping the cards around, the same card always fails... I'd say it's a bad card. If after swapping the cards, the opposite card fails, I'd say the slot/board is the issue. I presume you uninstalled the previous driver before installing the new one. And have the Xfire bridge(s) in place. (I can't remember if the HD 6870 takes 1 or 2 bridges) The PSU below the cards should not be an issue. And, I wouldn't try the bios flash, either.
Be sure you have the motherboard's chipset drivers installed, too. Either from the disk or download. Don't rely on Windows installing the correct generic driver.
 

crashadog

Distinguished
Nov 12, 2011
11
0
18,510
Thanks so much for replying. I actually posted my issue to a few other message boards, and we were fairly quick to determine that my motherboard was defective. I have an advance RMA processing with ASUS right now, and hopefully I will see this problem vanish as soon as the replacement motherboard is installed.

Since it's the same model, will I need to re-install the chipset drivers? Should I uninstall the ones I currently have before installing the replacement motherboard? Or would that be unnecessary?
 

clutchc

Titan
Ambassador
Good question. If you could be sure the board is an exact match to the previous board component-for-component, the same drivers should all work.
Btw, I wonder if swapping 'identical' boards like that would cause an OEM version of Win7 to assume the board had been changed, and therefore halt the activation? I would think not. But...
 

clutchc

Titan
Ambassador

You're good to go then.
Just for what it's worth... awhile back I swapped out boards and re-installed my original OEM version of Win7 that was with the old board. Figuring it would be a fail, I continued just to see what would happen. Surprise! I was able to use it simply by calling a toll-free number and answering a few simple robo-questions followed by typing in code numbers per the robo-voice. I was given a new license free of charge. Some months later, I ended up selling it and the machine it was on and haven't heard any complaints yet...
 

crashadog

Distinguished
Nov 12, 2011
11
0
18,510
I just received my replacement motherboard and spent all morning carefully installing it. I booted it up and tried to go into the bios first, and there was a strange "fatal error" after stating that I installed a new CPU. Said something about a chassis intrusion and that it was halting processes. It stopped there and I had to reboot. This time I didn't try to go into bios, and everything started up perfectly. Curious, I rebooted once more and tried to go into bios again. Everything was just fine this time. I disabled the "spread spectrum" as recommended elsewhere and saved the changes. Then I went into windows and opened up FurMark to try another burn-in test, wanting to see if replacing the motherboard had finally fixed my problem. It didn't; the exact same issue came up.

I am very upset that this issue won't go away. It shouldn't be the power supply, since I was able to test out another one and nothing changed (though the other was the same model and all, but they're crossfire-certified!), the GPUs both work flawlessly by themselves when the other isn't plugged in, with only the top one shutting off while both are plugged in and working (or even with both plugged in and only the top working if running FurMark), and now apparently there's nothing wrong with the motherboard either, which is ALSO built for crossfire.

What do I do? This is really getting to me. If everything is supposedly compatible and nothing appears to be defective or broken, what am I able to do? My drivers are all updated too...
 

crashadog

Distinguished
Nov 12, 2011
11
0
18,510


I have the 2001 bios right now. It was already installed on this replacement motherboard. I had the same bios version on the previous one, though for a good long time I was using 1101.

Also, spread spectrum had been on "auto" for a decent amount of time before I read the recommendation to disable it. You think I should try enabling it then?

My monitor is connected through DVI, and I don't have speakers, but instead have headphones connected through USB with their own drivers installed.

I admit when the issue re-occurred I didn't think to double-check and make sure that only the top GPU's fans had shut off. I assumed that the same thing had happened, and that was my mistake. I was pretty upset and wasn't thinking. I'll go enable spread spectrum in the bios. Do you think my monitor or headphones may be related to this problem somehow?
 

clutchc

Titan
Ambassador
I'm afraid I'm as lost as your are. I'm just thinking of things I would try if I had those issues. Yes, might as well uninstall the USB audio drivers temporarily. Eliminate as much as possible. I presume you have the DVI cable plugged into the top card. (Normally there's no signal present on the ports of the bottom card when CF is working properly. So, if there was, that would be a clue)
 

crashadog

Distinguished
Nov 12, 2011
11
0
18,510
The bottom GPU's ports send no signal. Both of the top GPU's DVI ports are sending signal. I wasn't expecting the second (lower) of the top card's DVI ports to send signal. Should I temporarily leave the DVI cable plugged in to the lower port of the top card?

I'll leave it there for now, and uninstall the headphone drivers. Then I'll try FurMark again I guess, and see if there are any changes...
 

crashadog

Distinguished
Nov 12, 2011
11
0
18,510
No changes, issue came up same as always, and as expected only the top GPU appeared to shut off (its fans weren't spinning).

Maybe the next time I try this test, perhaps I should try plugging the monitor into the bottom card before I give up and reboot? I have a feeling that this won't accomplish anything, especially since if there were sound playing when the problem came up, it'd skip and then cut out (lending to believe that it's a full-on lockup), but I guess all I can do is experiment now huh?

This really sucks =( Maybe I should just reformat everything and start with a fresh slate, see if the problem occurs with nothing but the necessary drivers...
 

clutchc

Titan
Ambassador

It shouldn't matter which of the 2 DVI ports on the top card you use.
 

clutchc

Titan
Ambassador

It still bothers me that the top card loses its fan. The fan should never shut off; it should maybe slow down, but it should always always run.
I know you've proven to yourself the cards are both good, but something is disabling the top card. Did you verify that after swapping the cards between slots, the top SLOT card still shuts down when you stress the system in CF?
 

crashadog

Distinguished
Nov 12, 2011
11
0
18,510
I'll be as detailed as I can here. My motherboard is an ASUS M4A89TD PRO/USB3. It has 2 PCI-E x16 slots. I have tested:

Each card by itself in the top slot, bottom slot left empty.
-Both cards can run FurMark without any apparent problems (burn-in test for over 15 minutes, no lockup issue)

Each card in the top slot with the other card in the bottom slot, with crossfire disabled.
-Games run fine like this, but FurMark (After around 5-6 minutes) causes the issue where the top card shuts off, even though the bottom card isn't working. I can see the bottom card's statistics in FurMark, and it idles around 35-38C with zero load, fans kept slow and quiet.

Each card in the top slot, in crossfire.
-Games in full screen run for some time (usually short), then the lock-up occurs. FurMark tests don't make it to 5 minutes before the lockup happens.
 

crashadog

Distinguished
Nov 12, 2011
11
0
18,510


No, I'm sorry I didn't make it clear. The instant this issue occurs, I lose signal to my monitor, the top GPU's fans all stop working, and if any sound was playing, it starts skipping in a very annoying fashion for a few seconds and then sound cuts out entirely.
 

crashadog

Distinguished
Nov 12, 2011
11
0
18,510
In case I neglected to mention, this issue only ever happens when gaming in fullscreen with crossfire, or running a FurMark test (I haven't tried any other GPU benchmarks). I can surf the web, chat online, listen to music, watch videos, play games in a window etc etc just fine with crossfire enabled.
 

clutchc

Titan
Ambassador
I assume you did the usual driver/CCC uninstall, Driver Sweeper from Safe Mode, re-boot, and re-install the drivers. If so...
Unless you (we) find a solution, I would be tempted to just start over. Save the CCC/driver pkg, Furmark, and any other pgms you need for testing to a flash drive. I would reset the BIOS to it's default settings; maybe even go so far as to pull the CMOS battery and jumper the 'clear cmos' pins. Install both cards. Re-install Windows 7, but don't install any drivers from the board's driver disk (unless absolutely necessary). Install the CCC/driver package. Activate CF. Install and run the Furmark test again.
That's about as clean a test as I can think of.
 

crashadog

Distinguished
Nov 12, 2011
11
0
18,510


I'll have to save this for another day when I have the time to do this work (and recover from it). I'll reply again with developments.

Whether or not we ever figure this out, thanks so much for all of your help. I really, REALLY appreciate everything you're doing to try to help me. You're a really good person =)