Lately, I’ve been having a problem with my computer randomly black screening/restarting. The only thing I can think of that might be doing this is either A) PSU failure (not likely considering the model I have and it’s well-within-warranty age), or B) my STRIX RX480 crashing the system due to heat issues.
The STRIX RX480 has always been a hot card. Path of Exile easily pushes it above 80 if you don’t crank up the fan curve. Guild Wars 2 keeps it above 70 at all times. It idles at 50c with a 30% fan speed for ****'s sake. Every time the computer crashes, I have observed it has been in situations where it’s very possible that a sudden spike of GPU usage could have been the case… making a turn into a particularly busy spot in an MMO. Entering/exiting the Heaven Benchmark (Seriously, wtf is up with Heaven? Opening and closing that program SPIKES GPU usage/heat and causing the coils to whine like a baby in a blender. My old 760 did this in Heaven and so is this 480). So, with that in mind, I checked the internets and found people who had issues with their 480s crashing from heat a lot.
Soooo… I’m 90% sure that’s the issue. The thing is, it didn’t used to do this so damn often. Granted, it is summer but… It happened just now, prompting me to make this thread, and it’s cooler now than some mid-day gaming sessions where I’m borderline breaking a sweat myself and the computer did NOT crash then. Granted, it could have been just the specific gaming circumstances that did or did not cause a crash at those different times of day and heat.
My case needs way better airflow (cheap fans), I know, and the fact that I have an old FX-6300 putting of 50c of its own heat from a massive air cooler less than 2 inches away isn’t helping. Sooo… I have a few questions.
Is the 480 programmed to register as being hotter than it actually is? If yes, then my custom fan curve is actually hurting it, not helping. The fans won’t spin up in time (the entire point of a false positive reading). Is there some kind of software issue that I could approach? And finally, should I just be looking to get more airflow? I’ve been thinking of 3D printing a custom side panel so I can strap a massive fan to the side to force a strong, positive case pressure (and, I’ll be honest, create a cool effect with a fan spinning over my… excessive… amounts of LEDs)? Lastly… would it be better to just take off the obviously inadequate 3-fan stock cooler and use a cold plate for water cooling? I was going to use my old water rig for my CPU upgrade, but… meh.
Simply starting Guild Wars 2 caused a system crash this evening. Upon restart, I decided to fire up Heaven benchmark without any background applications going on so I can see what would happen.
It crashed 30ish seconds into the benchmark… at only 78C, which is within manufacturer’s safe range.
As I now think this might be something else other than the GPU, here’s my build:
CPU: AMD FX-6300 (It’s old and has spent a lot of time OCed. Might be wearing out?)
Mobo: Gigabyte FXA-990UD3 rev 3.0
PSU: 750W Thermaltake Toughpower Gold
Other than Prime95, which I’ve ran in the past when I returned the build to stock clocking, what can I use to diagnose what is causing this?
I’m now convinced it’s not the GPU’s heat. I found online that others were having trouble with the Strix RX480s heat so I did what a few others did and replaced the thermal paste with Arctic Silver. Idle temps went down immediately.
I booted up Heaven benchmark and it immediately crashed/restarted PC… with the GPU not getting above 55C, as I was watching it to make sure. I have no idea what could possibly be causing a problem like this. It is similar to the problems I’ve had with a previous, cheap PSU, but this PSU is so much better that it’s either not it or it’s dying on me early and I need to get it replaced.
bsck your cpu frequency off a lil bit i started having issues with my old 9590 as well and backed it down to 5 ghz and havnt had any issues yet since
Why would CPU frequency be an issue here? It’s a stock 3.5GHz.
I did just notice something though. MSI afterburner says the GPU core clock was 1330 while the website says its close is supposed to be 1266. I don’t remember changing it, but I’m going to manually put it back to 1266 and see what happens.
Also, MSI afterburning is saying that the GPU’s memory clock is 2000… website says it’s supposed to be 8000. WTF?
EDIT: just checked. 1330 is it’s peak, self OC, so that’s normal.
Clicking “go” on Furmark caused an instant restart.
I’m almost certain the problem is either the PSU or GPU, but I can’t figure out what
do you have a spare pu you can toss in to see if it continues to do it
No, but I have a friend that can loan me one tomorrow probably. Been thinking of doing that.
The computer works perfectly fine unless I put it under load… which is exactly what happened with my last PSU when it was failing; so, I’m fairly sure that’s what it is. It doesn’t crash if I play ICEY (a lightweight, 2D sidescroller) but if I do any benchmark software or Guild Wars 2/Path of Exile, it crashes.
Two PSU failures in less than 4 years… fml.
Here’s what I find odd about the crashing: it doesn’t seem to crash when it’s just the CPU stressed. Prime95 didn’t crash it, even with 100% load. But high CPU AND GPU usage crashes it every time. I have no tools to rule out it’s just the GPU usage that’s crashing it, though. I know of no software that stress tests ONLY the GPU and I know that Path of Exile/Guild Wars 2/Heaven benchmark use both GPU and CPU. So, that being said, I think that does narrow it down to PSU failure under load.
maybe run prime95 and then open furmark with some small thing, like 1m particle low res, while monitoring temps and power consumption in hwmonitor or aida64