Having trouble resolving an issue with my GPU constantly crashing while gaming. I had a R9 290x and in December, all my monitors randomly went dark while playing PUBG. It ended up being that it killed the GPU and it could no longer POST. For the next few weeks I was able to play very res games using my integrated graphics. I then received a brand new GPU as the 290x had a lifetime warranty, so I just got a new R9 390x a week ago. Same problem, instantly crashed out when I tried to join a PUBG lobby.
I started playing some less-intensive games like SMITE, Portal 2, LoL, etc.. and these actually did okay. SMITE crashed once, but usually I could game for several hours fine. I then tried Civ 6, which was good for an hour then crashed. By crash, all 3 monitors go dead but I still hear sound in my headphones.
Up till now I have had a stock cooler and thought maybe my CPU was overheating (was about 75-80C in SMITE, and 85-95C in Civ 6), installed the new cooler 2 days ago and got SMITE down to 65-68C, but the issue still occurs, and now it is occurring on SMITE often. I also have noticed that if I have remove my side case and stick a large fan nearby, this seems to help with some of the crashing, however, my GPU and CPU temps seem fine, if not lower, so I am not sure what else would/could be heating up?
My current setup is:
CPU: INTEL I7 4790K 4.0G 8M R
With brand new Corsair H75 cooler.
RAM: Ripjaws X Series 16GB (2 x 8GB) 240-Pin DD
Motherboard: GIGABYTE GA-Z97X-SLI
PSU: EVGA 850W 110-B2
GPU: R9 390x-8256 (Driver: tried both the latest V17 and V18 one)
Things I have tried so far:
The only new new change to my setup over the past ~4 years (when I got all of my hardware) was that a week before the first crash happened in December (that killed my original GPU), I went from 1 to 3 monitors. I had a nice 2560x1440 which is still in the middle, and then I bought two 1920x1080 curved monitors to go on each of its sides. I presume 3 monitors is fine for a r9 390x though? I only game on the main monitor and use the other two for spotify/discord/web/etc.. Everything was running great for a week (high intensity games and such) until it randomly crashed one night and killed the GPU (no POST).
Anyways, I believe I fixed all the potential problems, and tried all the other things I can think of. I just am not sure what to test or try next.. I thought maybe it was PSU but the voltages looked good, and I am not sure how to test my motherboard or CPU to ensure they are "good". Could there be any issue with my outlet not giving enough power with 2 more monitors? Could something still be overheating despite the lower/better temps?
Just pulling my hair out at this point trying to figure out what could be going wrong.
Please let me know any information you want/need, I will get it to you ASAP (unless I am at work, then I will do it first thing when I am home).
I'm having the exact same problem, only i'm using a XFX Radeon HD 7970. I've also tried almost all of the things you have, to no result. My temperatures are stable at 75celcius, and my CPU doesn't go over 50.
The game that pushes it over the edge for me are Total War games, and if i put benchmark programs to the max.
In WattMan set the power limit slider all the way to the right, as far positive as it will go (+20, +50, +75 could be your cards max power limit). Then give it a try. Sounds like it could be your not getting enough power for the card. Common issue with drivers since last March.
Sorry it didn't help. Have you tried the latest drivers, just came out a couple days ago, 18.2.2. Maybe they will be better? Suggest you do a clean install from the custom settings, not an express upgrade.
Maybe someone else with more eyefinity experience will offer help.
Which psu, 850 or 450 ?, have you been troubleshooting with just one monitor connected ?
What OS, build No./vers. ?
450w PSU with a single 12volt rail isn't going to cut it. With your set up I'd be no less than 650 watt with (2) 12 volt rails.
At idle my GPU (7870DD) sits at 32-34c and my CPU will hover at 35-40c with stock fan and no overclocking.
YOU ARE A LIFE SAVER
I have been having so many issues with my OC'd Vega 64, with it crashing during stress-tests and benchmarks, I almost lost hope and thought my card was a dud. It wouldn't crash in games, unless I put the graphics to maximum, but even UserBenchmark and Furmark 720p preset would immediately crash it.
I put my Wattman profile to manual, as you suggested, and increased the Power Limit to the Maximum 50% and restarted and Furmark 720p didn't crash the system as it previously did.
EDIT: Although FurMark 720p now works, the 1080p preset still crashes the system. I've now read that because I have Vega 64, I should use two different power cables., rather than a daisy chained one, even though I have an HX1200i from Corsair.
EDIT 2: According to my UserBenchmarks, I have actually previously benchmarked this card using my old PSU and Driver 18.9.1, so either my new, much better PSU is funky, or the latest drivers are causing something weird to happen.
Glad it's better! Absolutely fix the cable issue. They need to be on different connectors. Then if you still have any issues, throttling is likely the remaining issue. That will entail a custom fan/temp curve to help keep temps down. If you are fine with the other changes as-is you won't need to do that. So glad you are making progress at least.
I reset all my settings within Catalyst Control Centre/Radeon Settings and everything is a lot smoother. I can run UserBenchmark and FurMark 1080 preset and it doesn't crash now.
However, my FPS in games has taken a small, but noticeable hit and running anything over the 1080 preset on Furmark still causes the crashes.
You think this is the cable issue?
My temps are damn good, as even running 1080p preset on FurMark for a few minutes only gets my card to low 60 degrees C.