PowerColor R9 270 crashing

[LeeChantrey]

Hi there,

First of all I take ESD precautions and ground myself when working with ESD sensitive equipment.

After I bought my new PowerColor R9 270 (to replace my Sapphire Radeon HD 5850) some less demanding games play fine (such as PayDay) but graphics valley benchmark and dying light cause the system to crash with a random color and lines going down the screen. This usually happens in the first 15-30 minutes. I've RMA'd the card and they said they've used it 2 days straight on The Witcher 2 and 3DMark and found no problems and have since returned it to me. I have a 500w Antec TruePower PSU and I have tried other PSU's (all 500w to 550w Antec TruePower PSUs that I know work fine) and it hasn't helped in the slightest. I've tried putting the R9 270 in my friends rig and he gets exactly the same symptoms I do. I put my Radeon HD 5850 back in my computer and did the same tests and everything is stable. I've tried installing countless different driver versions (ones marked stable and beta and nothing helps) The temperatures are well under the specified guidelines by AMD and the R9 card manufacturer guidelines. The minimum PSU requirement is being met to the best of my knowledge. The CPU temperature is fine. To the best of my knowledge I have not damaged the card with ESD.

For years I have always recommended and worked with AMD/ATI hardware as I find you get a better ratio money vs performance wise than you would do with nVidia and Intel. I've built rigs for my friends and I've had exactly the same issue with another friend with his R9 and AMD hardware. In the end I made him RMA his R9 and get a pre owned nVidia 770 because I couldn't seem to solve his problem with his R9 (he's had absolutely no problems since moving to the 770).

I'm at a loss to what the problem might be. I've literally tried swapping every component I can think of and reinstalled drivers and updated bioses. Nothing is working. Everytime I get my hands on an R9 or have recommended an R9 they seem to not work for me or the people I've recommended them for (although only 3 cases in total with 100% failure rate). All seem to have exactly the same symptoms that I've seen personally. So I have altogether stopped recommending this series of card until I know why it's not working for my friends and I.

There are a few similarities to all the computers I've built / tested / recommended the R9 on.

Windows 7 64bit SP1 Ultimate
Antec or Corsair PSUs at the minimum or higher wattage requirement specified by the card manufacturer.
Asus (PCI-E 2.0) DDR3, USB3 motherboards
AMD CPUs
Corsair or Kingston memory at a minimum of 8GB
Zalman fan or corsair liquid coolers where possible (Except on the GPU to not void warranty until card is confirmed as stable)
Zalman cases where possible
(Have been built by I)

I've also tried a 750WATT Corsair PSU Bronze (just be absolutely sure it's getting enough power).

At least one machine had the R9 as it's first graphics card (instead of an upgrade) It also had these problems.

I have an old P45 Intel motherboard somewhere with a Dual Core CPU on it - just out of curiosity - I will try the R9 on that. Though it will be DDR2 based and PCI-E 1 and many other different variables so it wouldn't prove much if it worked.

Does anyone else have any ideas on how I can get an R9 to behave?

 

[LeeChantrey]

Reducing ambient temperature seems to have made things stable (for now). It now stays below 70c at all times under load.
Before that the temperature reported under load was 76c- I've observed that it soon after crashes when in this temperature for too long (20+ minutes). Graphics Valley Benchmark now stable at 3hrs so far. Since I don't want to void my warrenty all I have to go on is pure speculation. If this turns out to not be the case for you, I apologize.

76c is (according to the spec) is a perfectly acceptable temperature but that doesn't mean the vram chips are running at safe levels (although usually the gpu heatsinks connect the vram and gpu chips, meaning all your chips should be about the same temperature) I don't have the proper equipment to confirm this but I theorize that the vram chips are gradually climbing in temperature to a point they reach critical/unstable levels. So I guessed reducing ambient temperature might help and it seems to. When I used to overclock my old Intel CPU (Which I am NOT doing to the card, but it's interesting to note), I recall having to raise/reduce vcore voltages. It could very well be that something isn't getting enough volts when it needs them under load. I don't know exactly what was happening, it's pure theory.

PS: The type of crashing looks and feels oddly familiar to the crash you would get if you were overclocking the memory on a graphics card. I want to repeat that I have never attempted to overclock the card in question.

There are other methods to reduce temperature - and that may well extend the life of the card

underclocking (may void warranty)
aftermarket fans (may void warranty)
reapplying thermal grease (may void warranty)
increasing airflow in your case
manual fan control

Lee