TR Forums

Hi everyone,

I took the risk of purchasing a used HIS Radeon R9 290X from someone on eBay who had used it (along with a number of identical cards) for Bitcoin mining. I got it yesterday, but have run into some trouble. I took out my other card (a Diamond Radeon HD 7970), put in the 290X, and fired the system up. I was worried from the start because when the system booted, it had the Windows 8 logo on the screen, but didn't have the animated circles indicating it was booting- it just sat there. I eventually reset it and then it booted to the desktop, but trouble continued. When I tried to play a 3D game, it failed, throwing an error within the game saying that the video card or driver was unstable. I then tried installing the newest beta drivers (i had the latest non-beta drivers, 13.12, installed before), rebooted, and tried again. This time, the screen went red; the system had hung, and I had to reboot. Before that, I'd tried to adjust the video playback settings in the Catalyst application, and the system crashed in the middle of that too. I uninstalled all AMD drivers and then installed the 13.12 drivers anew, but that didn't seem to help. At first, I figured the crashes just occurred when I tried to play a 3D game or when I was changing settings in the Catalyst Control Center, but on two other occasions, the system crashed when I was typing (coincidentally, typing up a post like this in these forums) in an all-2D desktop session. These weren't hard resets, but Windows crashes where 8.1 gives me the big and says a problem has occurred, then it saves whatever debug data it has to disk and reboots. Eventually, I took the 290X out and put the 7970 back in, and things were back to normal.

Now, obviously I have to wonder if this card is defective. I was pretty sure I had it seated in the slot correctly and in as far as it would go, and I don't think the PSU is an issue (I have a good Seasonic 750W Gold PSU in here); certainly the PSU was able to handle two 7970s in CrossFire not long ago, so I'd think it could handle a single 290X. Also, if it were a problem with a PSU that couldn't supply enough power, I'd think it would be a hard system reset when it crashed, since that's the behavior I experienced (admittedly a number of years back) when I put a new video card into a system that had a PSU that wasn't quite adequate for it. I also don't think cooling is likely to be the issue, but I suppose it's a possibility- again, the same system ran fine with two 7970s, and heat hasn't been an issue.

Has anybody seen the kind of behavior I'm experiencing with the 290X before, and is it a sign of a defective card or something else? The PCIe power plugs were both definitely all the way in and supplying power, and the GPU fan was running fine. In case it matters, I'm running Windows 8.1 Pro 64-bit on an ASUS P6T7 WS SuperComputer motherboard. Any help would be appreciated!

I would say the card is bad as wel;, either that or the person took off the heatsink fan combo and didn't put it on back correctly.

You've got plenty of PSU, the GPU fans are running and there are no hairballs in the way. Given the needs of miners, I can only wonder if your seller modded the card's BIOS and it's no longer playing well with others?

The other option is that your seller just ran it like a galley slave until it failed and that's why he put it on fleaBay.

Pretty clearly defective. You might be able to get a warranty replacement (depending on the terms of the warranty) so I would look into that as quickly as possible. If not, contact the Ebay seller.

Gah, surely the Bitcoin mining was a dead giveaway?!

For sale:
Card that has been run 24/7 since it was purchased in a poorly-grounded plastic box, overclocked to within an inch of its life and jammed in tight with plenty of other cards to make sure it was competing for airflow.

On the offchance it isn't wrecked, check what BIOS it's running as per Captain Ned's suggestion.

I know, perhaps it was foolish to buy a card that had been used for mining. I figured that since I've been running Folding@Home on my card for a while and it's still working fine, it might be the same with someone else's card.

I tried putting the 290X in again a few minutes ago, and at first things looked promising: I ran the FurMark burn-in test for the full 15 minutes without crashing (GPU temperature got up to 95), but then when I loaded MechWarrior Online, it was just two or three seconds into the 3D part before it crashed to the red screen again and I had to do a hard reboot. Then, when I rebooted and was starting to type this very post, I got another Windows crash, so I'm giving up on it. Thanks for your input, everyone.

Also, for what it's worth, the seller said that the BIOS and voltage settings for the card were not modified.

Evan wrote:

Also, for what it's worth, the seller said that the BIOS and voltage settings for the card were not modified.

Which could be not true...

Have you tried flipping the BIOS switch on the card and running the second BIOS? This is what the reviews talk about as Uber and Quiet mode but it is just a pair of different BIOS images.

What temps do you see in GPUZ when it is idling? Load temps aren't necessarily meaningful because the GPU will not allow itself to go past 95C even if the cooler isn't operating properly. What do VRM temps look like?

And also remember that in Windows nothing is "2D" anymore really. The card always has a 3D load on it. Once in Windows with the driver loaded, it does very frequently adjustments of clock speed + voltage as load changes, and this seems to cause some cards problems. It can be PSU related but yeah your PSU seems very solid.

I have a feeling it is defective unless the BIOS is messed up or the cooler isn't correctly attached.

did you check if the 2 graphics card are crossfire compatible? as far as I know, to crossfire cards, they must be of the same architecture. and the 7970 and 290x are not of the same architecture. the former is Tahiti architecture and the latter is Hawaii. so they're not compatible. did u try running just the 290x on its own?

You play MWO? nice.

Pug or are you part of a group?

Regarding your card, doublecheck temps, voltage, try the auto-overclock feature of the catalyst control center and see if that triggers a crash. That could be an indicator that the card is on its way out.
If possible, try undervolting it and see if that helps the card's stability. Maybe something in the card is abused and just needs a lower heat threshold.

As said in the thread, check if the cooler is properly attached. Maybe even try a repaste if you are confortable doing something like that. Of course all that in the case you can't return the card anymore.

i suggest u undeclock the card on both core and mem and see if it works out it will still be faster than ur previous card mostly so its worth a shot