One of my 7990s has got a problem...

zYx · 14 Jun 2014 at 23:50

It's hard and it would be long to explain all this, but one of the cards, clearly is not 100%...

Despite the fact the CrossFireX won't function and I always had some weird problems with drivers etc.

The card has an LED and it should be on if the card's ON. I won't go into details, but the rest of the hardware is fine, one of my 7990s is causing that my PC freezes during booting up, and it happens at any stage, randomly.

For example, right now, Windows is running, one of the cards is "ON" and the other one is not. The device manager shows three 7990s, one of which is 'disabled'. I enabled the third 7990, the screen went black and came back on. The device manager shows three 7990 enabled. However one of the cards (LED) is still off.

PC rebooting.

The PC froze with BIOS boot logo... I can only reset it.

It's booting up again and it's on. One LED is on and one LED is off. There's three 7990s visible in the device manager.

That's what happened just now. The situartion will be different tomorrow. The PC will freeze when the windows logo is pulsing during booting up.

Sometimes I get USB devices being unrecognised errors, or some weird drivers are being installed. It happens after I had to reset the PC.

Is there anything I can do? VBIOS flash or something?

I only have 6 days of warranty left though... (Apparently I have one more year if warranty left)

edit: after another reboot, there's now four 7990s visible in the device manager, one of which is disabled. I'm going to enable it now. It's possible that I get 0xB3 BSOD error. This time the device cannot start, and there's a problem. When I scanned for a new hardware the problematic 7990 disappeared from the list and there's only three 7990s left in the device manager.

Obviously there's a problem. I contacted OcUK and Sapphire already.

AlamoX · 15 Jun 2014 at 04:00

beh i got confused with your post, i thought you had 3x7990 cards running, like 6 gpus.
about the led, during desktop phase 2nd card goes to sleep i think, untill it's needed in a game or bench or whatever require more compute.
the 2nd card is disabled during benchs too ?
start by uninstalling drivers, and install the 14.4 catalyst.
at some point i had weird freezes with my pc, i opened and unpluged all psu cables, and pluged them again, and voila no more freezes, i remember this probleme drove me nuts couldn't figure out the issue.
also try running them at base clock, if they are overclocked, and run some gpu stress test like furemark or valley.

LtMatt · 15 Jun 2014 at 08:54

1. Uninstall drivers
2. Download, install and run DDU.
3. Tick delete AMD C: and restart and clean in safe mode
4. Install 14.6 Beta, restart.
5. Install afterburner, restart.
6. Disable ULPS restart.
7. Has that made any difference?

zYx · 15 Jun 2014 at 09:53

LtMatt said:
1. Uninstall drivers
2. Download, install and run DDU.
3. Tick delete AMD C: and restart and clean in safe mode
4. Install 14.6 Beta, restart.
5. Install afterburner, restart.
6. Disable ULPS restart.
7. Has that made any difference?

I've tried this apart from #6. I currently have 14.4 installed. I tried 13.12, 14.6 and couple of other versions.

I mentioned that the PC freezes during random stages at booting up, before windows even starts. Could drivers have anything to do with it?

And regarding the LEDs, both LEDs are meant to be on all the time.

UTmaniac · 15 Jun 2014 at 10:09

Isn't this ULPS doing its job, turning off all but one graphics card until there is a need for it?

zYx · 15 Jun 2014 at 10:12

As far as I remember, both LEDs were always on before I noticed a problem.

Also, Xfire settings are missing from CCC.

Reinstalling drivers may or may not help, but if it does, it's only until another (re)boot when the problem starts again.

LtMatt · 15 Jun 2014 at 10:33

zYx said:
I've tried this apart from #6. I currently have 14.4 installed. I tried 13.12, 14.6 and couple of other versions.

I mentioned that the PC freezes during random stages at booting up, before windows even starts. Could drivers have anything to do with it?

And regarding the LEDs, both LEDs are meant to be on all the time.

Ok, do all what i said, but try my fix here first.

http://forums.overclockers.co.uk/showpost.php?p=25247146&postcount=5

mmj_uk · 15 Jun 2014 at 11:33

Have you tried removing your CPU overclock? did you stability test your system at all?

The USB unrecognised errors sounds like some sort of corruption that is happening and freezing at the Windows screen is usually hardware stability.

iB CoMpL3XiTy · 15 Jun 2014 at 11:54

Have you tried both of the cards separately (one at a time)?

Also have you tried swapping the cards around, it may be a problem with one of the cards or maybe one of the PCI-E lanes.

zYx · 15 Jun 2014 at 13:13

LtMatt said:
Ok, do all what i said, but try my fix here first.

http://forums.overclockers.co.uk/showpost.php?p=25247146&postcount=5

Ok I'll do that next.

PS. I have Afterburner 3.0.1 installed.

mmj_uk said:
Have you tried removing your CPU overclock? did you stability test your system at all?

The USB unrecognised errors sounds like some sort of corruption that is happening and freezing at the Windows screen is usually hardware stability.

About 30 minutes ago, I switched to older BIOS with default settings, no overclock and both cards seem to be ON. A driver installation took place during the first boot and the second card turned on.

I rebooted the PC once and both cards are ON. Performance tab in CCC reappeared, and I can even enable CrossFireX, but last time I did this, I got 0xB3 BSOD, and when I get that, all sort of weird problems start again.

iB CoMpL3XiTy said:
Have you tried both of the cards separately (one at a time)?

Also have you tried swapping the cards around, it may be a problem with one of the cards or maybe one of the PCI-E lanes.

I have swapped them and the only improvement I can see is that Windows does not freeze any more. The PC only stops during BIOS logo or Windows logo.

My mothoerboard has got 4 PCI-e switches. It seems like one of the cards is not right. My system is watercooled, so I can't keep removing the cards.

Anyways i'm still on 14.4 and with CrossFireX disabled. Before I enable it.

mmj_uk · 15 Jun 2014 at 13:47

So you are saying the second card functions correctly with your CPU at stock?

If so I wouldn't jump to any conclusions about your GPU, your CPU may have just degraded/been damaged due to high voltages. iirc 1.4V was considered pushing it on Sandy Bridge.

edit: plus you could have a lot of corrupt files causing issues? if it's not too much trouble I would try a fresh install with CPU etc at stock.

zYx · 15 Jun 2014 at 15:14

mmj_uk said:
So you are saying the second card functions correctly with your CPU at stock?

If so I wouldn't jump to any conclusions about your GPU, your CPU may have just degraded/been damaged due to high voltages. iirc 1.4V was considered pushing it on Sandy Bridge.

edit: plus you could have a lot of corrupt files causing issues? if it's not too much trouble I would try a fresh install with CPU etc at stock.

I'm pretty confident that files are ok, but who knows.

I haven't enabled the CrossFireX yet as I'm afraid of BSOD…

I'll test it like this with no overclock for a few days. We shall see.

If it's not the card, why are they both behaving differently?

zYx · 16 Jun 2014 at 22:34

Unfortunately, today one of the cards is OFF again. The PC froze when I enabled my USB speakers. Weird stuff is happening.

That is with default BIOS settings. No overclocks...

postmanfw · 16 Jun 2014 at 22:54

Silly question but have you tested your ram with memtest?

http://www.memtest.org/download/5.01/memtest86+-5.01.usb.installer.zip

Put cpu back to stock to rule that out of the equation...as said 1.48v 24/7 is high...i killed a 2600K using less...degraded to point it wasnt stable at stock even

oops just read last post and overclocks off already...my bad.

Have you testing pc for stabilty at stock using prime95 or OCCT etc?

zYx · 16 Jun 2014 at 23:05

postmanfw said:
Silly question but have you tested your ram with memtest?

I did. 7 times in a row :> No problems.

http://www.memtest.org/download/5.01/memtest86+-5.01.usb.installer.zip

Put cpu back to stock to rule that out of the equation...as said 1.48v 24/7 is high...i killed a 2600K using less...degraded to point it wasnt stable at stock even

oops just read last post and overclocks off already...my bad.

postmanfw said:
Have you testing pc for stabilty at stock using prime95 or OCCT etc?

I did that too.

All other hardware appears to be fine. Even with my 4.8 overclock :>

zYx · 18 Jun 2014 at 17:30

Yesterday both cards were ON (both LEDs were ON) and today only one is ON. The device manager is showing this:

gareth170 · 18 Jun 2014 at 17:45

maybe a PSU issue..

only way to be sure it's not the cards is to run one card for 2-3days then the other for 2-3days

mmj_uk · 18 Jun 2014 at 18:37

Do you have another spare system to test the graphics card on? it's the easiest thing to do really it'll tell you whether it's the graphics card or something else.

The trouble nowadays is that everything is on CPU, memory controller, PCI-E controller, so push silly volts through it for long periods and funny stuff can start to happen elsewhere in the system. It might be the graphics card but who knows? you just have to try to narrow it down (use another system if possible).

JamesM · 19 Jun 2014 at 08:21

I thought the LEDs on the 7990 were to indicate that a GPU was 'disabled' under energy saving mode or something

Out of interest could the Revo Drive be effecting the cards as it's also PCIe based?