I bought a AMD Radeon 6870 GPU with 1 GB of VRAM for ~$150 (this was over 6 months ago) at an electronics store.
For some reason, when I do GPU intensive things (specifically, games that use up a lot of VRAM and such), my computer's screen can go black and hang forcing me to restart my computer.
I've noticed that the GPU can get up to ~70C and this will usually happen at around ~75C, but when I underclock the video memory with MSI Afterburner, it can reach temeperatures up to ~80C without causing artifacting and the usual problems associated with high temps on GPU's, so I doubt it's directly connected to temperature.
Is there a way for me to fix this short of buying a new GPU?
I had a 5750 and that was the driver that causing issues like this, rollback to previous or newest and see if that corrects the problem. Run your computer with the side panel off to see if temps goes lower (maybe you have a bad ventilation in there so heat build up around the card... The last option would be to RMA it under warranty...
Post your system specs. The issue may be attributed to many things - drivers, temperature, PSU to name a few.
As mentioned, update the video card driver and make sure the video card has ample cooling. Removing the side panel helps or add a side fan to intake cool air onto the video card. If auto control isn't keeping the temp down, you can also manually adjust the video card fan to max when gaming. This can be done through the catalyst control center.
Also, if you're overclocking the video card, back it off a bit. Likewise, if your computer system if overclocked that could be causing it.
It could be something else in the system; so as mentioned, please list your specs.
First of all, thank you to everyone who responded, especially Rodney. That's the first time anyone has responded with a video on youtube, especially with so many subscribers and whatnot, to me. Awesome.
Anyway, on to the topic. I'll do my best to give as much information as possible that's relevant.
AMD Phenom II X4 Black Edition (was X2 but core unlocked) B60 Processor OC'ed to 3.6 GHz (Turning off OCing/Core unlocking doesn't help)
MSI Military Class II 970A-G46 Motherboard
8 GB of Kingston DDR3 Hyper X Blue RAM
Two 1 TB Western Digital Hard Drives (one's storage, other is back up)
VisionTek (<-brand) Radeon HD 6870 with 1 GB of RAM
NZXT 750 watt Hale82 Series Power Supply
All inside an NZXT Gamma case that has 4 intake fans and 4 outtake fans for air.
Notes: I live in the US Midwest region, so it gets very hot here during the summer.
Using MSI Kombustor, My GPU can't reach 90C like Rodney mentioned in his video. It artifacts, then goes black at around 70-75C. That's low from what I've read for a GPU to cause a restart temperature wise.
My current AMD drivers are the latest. I've had them for a while (I don't think there's been an update recently).
I got the 6870 for $150. ... That kinda says to me something might be wrong with it from the get go.
Using Speedfan (great program), I've seen that my temps usually hit a maximum of 71-75C in game (on times when it hasn't black screened, that's sorta random), so there's that.
Anyway, I did some research, and it seems some 6870's that were made in France (I think) had some bad VRAM in them. Is it possible that lowering my Memory Clock speed from the default 1050 MHz to something lower might help this? If that is the problem of course...
Once again, thank you for helping :).
Edit: Forgot to put what monitor I have.
I use my Sony Bravia 42' TV as the monitor, though I have a secondary normal computer screen. The TV is connected with a HDMI and the monitor is DVI-I.
If that start from sudden, no driver update/overclock changes, i suggest to see if there's dust in the fan or VRM cooler cause it's a real problem on most cards, the VRM overheat easily, also you could try to set the fan to 100% to see if it's correct the problem or not (if yes, then you card really seems too hot and maybe only 1 component is not stable and crash the rest of the card... Maybe the Memory chips are not cooled by a heatsink like the gpu so they goes hotter than the temp you see in CCC and the VRM don't have temp sensors so that's an option to check)
PS: if you can't find the problem, i really suggest you to RMA the card before the warranty passed out...
I tried setting the Memory clock to lower than the default, and now it can get up to 90C without crashing (using MSI Kombustor, and I stop it once it reaches that temp). I guess it was bad VRAM then.
I recently installed an aftermarket cooler on the card so it was recently cleaned. That, and I bought it at an electronic store and not a website, so it basically has no warranty. Oh well.
Thanks for the help guys. The problems been (mostly) fixed.
Your memory is surely overheating, does the cooler comes with ram cooler ??? if not, you could try these coolers... they will reduce the temp of the memory a lot...
If it comes with a large heatsink that cover memory chips, do you use the sticky "TIM" with it ???
Also, does the Voltage Regulator have a heatsink on it ?? This is the surely the hottest part in there for sure so you can check for it and it's responsible to manage Voltage input/output in the card so if the core or the mem doesn't receive enough voltage then this one is the guilty...
PS: What cooler do you have on this, it will help me to see the most plausible cause of that problem...
I still suggest you to RMA the card if it can't run at the "stock speed" after testing all options cause it's surely defective then...
The aftermarket cooler I installed is the Arctic Cooling Twin Turbo II. It came with lots of fins with sticky thermal tape stuff on them to put on all the RAM and Voltage Regulator. It was kinda difficult placing them all and still getting the heatsink to fit because my GPU has a different layout than most 6870's thanks to it's brand.
Like I said though, the problem is mostly fixed. I wouldn't want the card getting over 90C either way as that's entering the danger zone for electronics. That and I can't RMA. It was purchased at a store. A physical building. Not a website.
Thanks for the help though :).
You can request a RMA from the manufacture even if it was bought at a retail store vs an online one.
Yeah, contact Visiontek and ask for the warranty/RMA policy or ask if their card ever had problem like that, maybe they had a new bios/Driver on that card that corrects the problem...
Copyright 2013 © Godem Online Inc. | Web and server solutions by NewTech Solutions.