Is it possible that you encounter problems only when playing graphic intensive games, and only after sufficient period of time to heat up the components?
My concern is that the GPU is shutting down due to overheating, which can lead to a failed hardware problem (like flipping a coin, it may restart fine once or a hundred times before it just finally gives out). The use of a system on low processing programs may bring the heat withing the chassis to a higher level, but nothing to worry about. And then when increased by intense play the GPU overheats, and the problem occurs (and this also making it harder to track down, as if you start plaing intensly immediately it takes longer to go from room temp to overheated then if it was already warmed up by running low intensity apps).
(This route is what I would choose, as it addresses the problems in order of "least cost to replace" on the average system...Additionally, it is recommended that you follow the instructions of dicke, your original helper on the issue. I refuse to interrupt his process and I am just providing an opinion based on the facts at hand)
If this seems like a possibility I would first check for and replace any and all bad components for the cooling system. The remaining components in the chassis needs dusted out (with canned air designed for computers ONLY) to ensure the coating of dust does not add to the heat problem. Focus on the heat sinks too. Re-arrange the cabling if possible to get them out of the way of the air flow and get the maximum possible cooling. Then you test your system again, if you get another failure you need to test on a known good (EMPHASIS....KNOWN good) GPU, and replace if the problem is not replicated on the good GPU. If the good GPU fails as well you may need to look into upgrading your cooling system (if the chassis still gets blazing hot) or looking into another component that may be bad or affected by poor heat dissapation (IE: is the CPU properly connected to the heat sink? Do you have a bad or loose RAM stick?)
@dicke - I would agree with you on one problem at a time as well as the approach you took, but the heat listed after your post concerns me and adjusts my opinion on how to troubleshoot this particular system since the errors may be a chain reaction started by heat problems. I apologize if it seems I am just walking all over your process.
Edited by hamluis, 11 June 2014 - 05:38 PM.
Removed unnecessary quotebox - Hamluis.
-- Windows 7 Ultimate on custom built system, Windows 10 on under powered laptop. Sophos UTM 9, Ubuntu Server and Windows Server 2008 R2. HyperV Virtualization --
"The hottest places in hell are reserved for those who in a period of moral crisis maintain their neutrality," John F. Kennedy