I have an insane problem, the quick version is: I've noticed my computer has started to "randomly" crash, or crash overnight. I've reduced it down to something very simple... When I switch on my 12W LED BEDSIDE LAMP (!), my computer freezes. The lamp has nothing to do with the computer, other than being plugged into the same extension lead. And through process of elimination, the crash seems to be the fault of my graphics card (SAPPHIRE RADEON RX VEGA 64). More accurately: My computer is running fine, playing a YouTube video for example. I turn on, or plug in some device into another electrical socket. Something to draw more power. I'm not talking about plugging anything into the computer. Not every time, but very frequently, the computer will instantly crash: the display will go black and lose connection to the monitor. Music that was playing will keep playing. The hard drive activity light continues to flash. The computer is not fully frozen, but the display never recovers. Pressing NUM LOCK on the keyboard doesn't work. When I restart my PC, and look in Event Viewer, I'll see: "Warning: Display driver amdkmdag stopped responding and has successfully recovered.". It did not recover... I've made a video showing the problem: Literally all I do is plug in a lamp, and my computer crashes... Some details: I actually have TWO "SAPPHIRE RADEON RX VEGA 64" cards, and both of them repro the problem! Both cards were in my computer since I purchased them from new, a few years ago. I'm doing testing with only card one at a time now. I have done EXTENSIVE testing between multiple computers and many different components: I have the computer shown in the video above: i7 8700k, ASUS ROG MAXIMUS X HERO, etc. I have another computer with an AMD Ryzen processor/mobo etc. It also repros here, with absolutely no shared components except the graphics card. I've tested with three high-end power supplies: EVGA SuperNOVA 850 P2, Corsair HX1000i, and a second EVGA SuperNOVA 850 P2. Repros with all of them. I've tested with different PSU cables (making sure not to mix brands), especially different PCIe power cables, still repros. I've physically moved my computer to another house incase "dirty electricity" was the problem. It repros at another house. Turning on the lamp with its own switch, with the switch on the extension, or physically plugging in the lamp plug all repro the problem. It's not just my lamp, plugging in my iPhone charger (!!!) also repros the problem. I've tested with the computer plugged directly into a wall socket, and also with three different extension leads. I've tried two different monitors, plugged in with HDMI, DisplayPort. I've tested with multiple different AMD driver versions, including the latest 20.10.1, 19.4.3, other random ones. The problem does NOT repro with other graphics cards including: "Asus Radeon RX VEGA 56 ROG Strix", "ASUS NVIDIA GeForce GTX 1070 ROG STRIX OC", my random old "EVGA GTX 680". There's no overclocking anywhere. I have also tried underclocking my GPU, it still repros. I have tried every debugging/reduction step I can think of, and the single common factor is the GPU. The problem will also repro when in the BIOS config. So it's not Windows/software related. Sapphire GPUs only have a 2 year warranty, and unfortunately my cards are almost 3 years old now. I've asked Sapphire about it a couple of weeks ago, but they haven't replied yet. It might take them a while... I could really use some help with this. Do you agree that it must be the GPU(s), or am I missing anything? Assuming that it is the GPU(s), what could be wrong with them? What is happening here? Any questions or ideas?
I've had a similar problem with an R9 390X and an RX580, switching on my 11w Fluorescent desk lamp. Display driver crashes and never recovers the link to the monitor, eventually changed from HDMI cable to DVI-I. Problem gone, it must somehow upset the HDCP and black out the picture.
Perhaps set it up then disconnect everything from the computer, leaving it powered on and the monitor powered on, cycle the light a few times and see if it actually is a cable picking up the spike or if its an actual device, or going through the power cord. I was in a situation once where particularly staticy colleague would touch a metal frame on my desk and my monitors would power cycle.
Thanks for the ideas. In general, the problem has become harder to reproduce. Good in general, but annoying for debugging it. I have a feeling my problem repro's less with the Corsair HX1000i I'm currently using, compared to the EVGA SuperNOVA 850 P2 I was using before. I tried disabling HDCP on both my monitors in AMD settings, and I was able to repro one crash in quite a lot of attempts. I don't think it's the cables, because I've repro'd this on three different singly-connected monitors, each using their own cables. One of the monitors was tried when I took my computer to a different house. I tried a bit of monitor cable swapping at some point, but it was not fruitful. The lamp already uses an LED bulb, a GLS E27 12W. I don't think the bulb is relevant though, since I have also repro'd this by plugging in the most boring of items: my standard iPhone charger. It wasn't as consistent to repro with the iPhone charger, though. This is a possibility. Googling for "static graphics card crash" found this thread: https://www.reddit.com/r/techsupport/comments/f2j9r6/gpu_crashing_issue_static_electricity/ It's not identical, but it does feel similar to my problem, and it's also a Vega 64. That guy says he grounded everything and it still happens though, with no resolution. I will try your suggestion, and I will also try setting up my computer out of the case on wood, see if I can repro it there. This sounds possibly similar: https://community.amd.com/t5/graphi...thout-cause-pc-techs-are-clueless/td-p/111801 And this sounds like the exact same problem: https://forums.tomshardware.com/thr...pped-almost-everything-out-no-change.3438358/ It's nice to see I'm not the only one with this problem!
See if you can borrow or acquire a UPS to regulate the quality of power to the computer? Even a cheaper UPS with a short battery life would do as the purpose isn't to use it as a UPS, just to clean up the power delivery to the PC.
I can no longer reproduce this problem. I didn't do anything, it just gradually happened less often and then never. That's annoying for debugging the problem, but good enough for now since I'll soon upgrade to a new GPU... when I can ever order one. Sapphire completely ignored my support ticket, and even worse "deleted" it - I assume they automatically delete inactive tickets after a certain amount of time. I will avoid Sapphire in the future, it's not the first time I've had problems with them. Bad warranty, bad customer service. My next step would probably have been to try a UPS with AVR.
