Open a command prompt as administrator and type sfc /scannow let it check for and repair corrupted system files ...
Here's what I did.
1. I added time to the registry editor - now at 16
2. I ran furmark for a half hour and the temperature remained about 69 degrees C the whole time, even while under full load.
3. I reset the page file.
4. I ran a system file check.
5. I have the driver verifier on, and will turn it off tomorrow.
6 I turned off hardware acceleration in my chrome and firefox browsers.
6. I did not do a memtest or hdd test.
I no longer get the blue screen, the "video.tdr.failure" (atikmpag.sys) message when I plug in the HDMI cable. I get partial displays on both the television and monitor. It cycles through, trying to resolve the same thing on both screens, doesn't, freezes, and then 10 or 15 seconds later, tries to make both displays work again - it starts over.
After I pulled out the HDMI cable, the boxed message, "Display AMD Driver stopped responding and has successfully recovered. This message came on and off a whole bunch of times before I right clicked display settings on the desktop and clicked on "detect".
Any other thoughts?
I have been dealing with this same issue myself, and here is what I found was the fix, in some cases your software tweek may work but. 99.99% of the issue is not software. it is hardware. Now this is for laptops. I have not seen the design on desktop pcbs to determine if its the same method of manufacturing.
First AMD shame on you for not having the gpu installed in a socket. The problem is how the chip is soldered to the board. The way they baked the board leaves the solder cold and when things get hot it creates open joints or even shorts and the gpu gives you the errors. that vid chip gets so hot that it is in fact shutting down. be careful setting the tdr too high. most software reset timers are set to two or three times the manufactures specs so after .75 or 1 second the chip should be resetting or finishing its task and moving on its way. I see some as high as 16. Having it this high will guarantee you will melt the chip. So what is the possible fix. Try and reflow the chip. There are plenty of vids on it. do not do the hairdryer. it may work but get serious about your device and use the proper tools and methods. but if you do not know what you are doing take it to a reliable shop and let them reflow the chip. I had mine reflowed last week and this error is gone. video is crisp solid and no more heat or errors. I went through all the motions in this thread software hardware checks reset to factory etc. for me and many its all how the gpu was attached to the pcb. I have noticed also that this is not isolated to just amd. all the gpu chip sets have been done like this. Sony, HP, Fujistsu, Acer, etc.. they all have gone this route and should be ashamed to have built things in this manner. Good luck and if you have gone as far back on software to factory reset and still getting the error. it is more then likely the soldering of the gpu. IMHO
I'm just a average Joe, but I don't think AMD does anything except sell the graphics chip. The vendor does with it as they will. Some are better than others.
I have been experiencing the same issue several times losing a lot of work while working on Adobe Illustrator CC 2015 because the application crashes with the driver but does not recover afterwards. It does that especially while scalling or moving objects and at very random moments. I'm running an Intel i7 4ghz, 16gb ram, AMD R9 200 on Windows 10 64bit so no idea what can cause this problem. My drivers are up to date. Unfortunately I don't have the TDR key in my registry, so what should I do?