TR Forums

I'm not a complete expert on this computering stuff so any help or advice anyone could give would be greatly appreciated.

Story: I purchased parts for my build this January / February so this is quite a new build. Everything installed wonderfully and I've used it every day since then. My GPU is a MSI GTX 760. I've been playing Bioshock Infinite, Injustice, and Far Cry 3 most of those days. I got hooked on Far Cry 3 (yes, I know I'm a few years behind on my games) and was playing a marathon session about a week ago. After about 3 hours, all of the sudden the graphics went wonky and the computer froze. I assumed I had overheated the GPU and reset the computer. I checked the CPU temp and it was in normal range and then checked the GPU temp. It was a little high but nothing outlandish.

So I took this as a sign that I needed to spend more time with the kids and took a few days off of gaming. Yesterday I decided to give it another go. Booted up the computer and then updated the gpu drivers (just in case). Then I loaded up Far Cry 3 again. Played for about 2 minutes before the game crapped out and the computer froze again. I certainly hadn't been taxing the pc this time.

I tried to reset the pc but kept getting errors @ the Windows boot screen. Mostly "video_tdr_failure" but occasionally "memory_management." I attempted to get into Windows about 30 times without success. The Bios showed cpu temp was very low. I couldn't get to the GPU temp because my Bios doesn't show that and I couldn't get into Windows. I opened the case and checked that all the fans were spinning (case, gpu, cpu, and psu). Touching the gpu told me that it was still cold. I made sure the gpu was inserted all the way. Tried to boot again and failure again.

I took out the gpu and hooked the monitor to the integrated video. Everything booted right up and worked perfectly. So I thought "perhaps those new drivers suck...". I reverted back to the previous drivers and prayed. Put the GPU back in and booted up. Windows errors again. Went through Windows troubleshooting for a while with no success. Finally, after resetting about 8 more times, Windows booted right up. WTF?

Now that I was finally in Windows with the GPU installed again I opened MSI Afterburner and looked at the temp. Everything was normal. I opened up FurMark to test the GPU and watched the temps. Temp went up of course but it performed admirably and didn't have any errors. So I thought I better boot up Far Cry 3 and give that the test. Played for about 30 minutes while watching temps using both Afterburner and GPU-Z. They never got above 70 C and nothing died.

Any ideas what might be up?

CPU: Intel Core i5-4670K 3.4GHz Quad-Core Processor
CPU Cooler: Cooler Master Hyper 212 EVO 82.9 CFM Sleeve Bearing CPU Cooler
Motherboard: MSI Z87-G43 Gaming ATX LGA1150 Motherboard
Memory: Kingston Black Series 8GB (2 x 4GB) DDR3-1600 Memory
Storage: Crucial M500 240GB 2.5" Solid State Disk
Storage: Seagate Barracuda 2TB 3.5" 7200RPM Internal Hard Drive
Video Card: MSI GeForce GTX 760 2GB Video Card
Case: Corsair Graphite Series 230T Black ATX Mid Tower Case
Power Supply: Corsair CX 600W 80+ Bronze Certified Semi-Modular ATX Power Supply
Optical Drive: Asus DRW-24B1ST/BLK/B/AS DVD/CD Writer
Operating System: Microsoft Windows 8.1 Pro (OEM) (64-bit)

Things I've tried:
Windows Memory test: all was fine
Check hardware for spinning fans
Press and make sure everything is properly seated
Newest Nvidia drivers

It certainly sounds like a videocard problem, but I always start by testing the memory just for process of elimination. Run Memtest86 overnight or for a good 4+ hours. Even if there's nothing physically wrong with the RAM, it could be a timing in the BIOS.

In the meantime, you could remove the card and test it in another system to see if the same issue occurs.

Memtest86 for sure.

I also wonder if the GPU VRMs or Memory might be overheating. This might explain why you get problems after long gaming sessions.

Well, TDR failure indicates a device not responding... my best guess would be that you overheated your video card and now it's wonky somehow... stick it in another machine to confirm, and then hope it's under warranty.

As others stated it sounds like overheating.

cphite wrote:

Well, TDR failure indicates a device not responding... my best guess would be that you overheated your video card and now it's wonky somehow... stick it in another machine to confirm, and then hope it's under warranty.

I believe MSI's warranty is two or three years. Either way the 7XX series hasn't been out long enough for the warranty to be expired.

Thank you for your replies. I'll try memtest and see what I can find and then try the card in a different system. *Crosses fingers

I had a similar problem with 2 seperate cards a BFG 9800, and a MSI 6950. Both had great cooling on the processor, but nearly no cooling on the memory..beyond a piece of aluminum bolted on to the main heatsink. I started having boot issues with both along with lockups on the 9800 and wonky colors and driver resets on the 6950. Eventually the 9800 just plain refused to post and was dead. The 6950 soldiered on a bit longer with crazy colors and loosing 256MB of the memory on the card before it died. Just because the GPU is running fine, doesn't mean the memory hasn't cooked itself to death...esp on heatsinks with poor cooling or bad airflow over them.

Heheh, the problems I had with a BFG 9800 are the stuff of legend. Didn't surprise me at all when they went out of business. I was able to take screenshots of the problems with the 9800 as well as one or two other cards, I'm going to be going through some old data this weekend so I'll keep an eye out for them. It was some good stuff.

In my case it actually was the GPU, not the memory, though it wouldn't surprise me at all if the memory got cooked too.

Just as an update, I ran Memtest86 over the weekend. I've never used it before so I wasn't totally sure what I was doing. I ran whatever the default test was and it ran for just over 3.5 hours and returned zero errors. I realize now that I could have ran it for longer but this is leading me to believe that the memory is probably fine.

I got into Windows on Saturday morning and played Far Cry for about 5 minutes before weird colors and rendering errors starting popping up. It didn't freeze Windows so I was able to Alt-tab out and look at GPU-Z. Temp looked fine and so did everything else. I was able to get out of the game and run Furmark. It ran just fine for over 20 minutes at full resolution (about 70 fps) and 80 degrees Celsius at full load. Weird.

I reset the computer and went back to Far Cry. I was now able to play for 2 straight hours without a hiccup. Weird. I don't know what to try at this point.

It's almost certainly the card. As others have said, it doesn't necessarily have to be the main GPU; it could be the power circuitry or one of the RAM chips overheating. The next step would be to test the card in another PC and try to reproduce the same problem. If you don't have another PC available, just RMA the card.

definetly the card, time to RMA

i had similar happen with a EVGA GTX 460, RMA'd without problems.

GL

Forgive my ignorance, but should I be looking to RMA with the manufacturer (MSI) or vendor (Newegg)?

You should go with MSI as I doubt newegg would cover something after this much time. Just expect the card to be gone for a while, Ive done RMAs with MSI before one took a month to get back, the other took a month and a half.

when i RMA'd my EVGA it took 11 days, just saying.

but GL

Just got the card back from MSI. It took exactly 7 days. Not bad! To be clear, once they got my card in their system they just sent me out a different card that I assume had already been refurbished from someone elses RMA. There was even a paper inside the shipment stating what the barcode on the one I sent in was and what he barcode of the one they sent back to me was. They were different. I assume that is standard protocol?

Anyways the new one seems to be working good so far...I'll see how it goes after a couple longer playing sessions.

That is pretty quick, I would have liked mine to be that fast lol. The card you get back is a refurb, and it may be good or it may fail again...really depends on what they fixed on it. The 6950 I got after my first one failed as a RMA was a chinese card and was clogged with dust and dirty as heck. After some cleaning and blowing bunnies out it worked for 5 months before dying. When I tired to RMA that one, they refused to do it as they said it wasn't a valid card. I ended up taking to a person from MSI and they finally had it go through for the RMA, and the card they sent me as the RMA was in much better shape and looked like it was new.

The wait for my card may have just been that they didn't have any to send me at the time, and had me wait until they had some available. Still was a really long wait though, maybe things are better now as it was around 2 years ago I did my RMA.