For the past 2 or so weeks two of my rigs have been having this issue. I’ve switch power supplies, ram and ram order, gpu’s, verified I’m not overheating anything, and read through windows event logs.
edit as of 8/14/2023 @ 1601 i managed to capture the issue on video two separate times. not sure how i can link it here directly thought. ive reset windows settings, uninstalled and reinstalled d4, uninstalled battle net fully, reset bios settings, etc…im just lost.
my issues could be:
im playing diablo 4 for x amount of hours and the game starts locking up, stuttering, or even freezing.
i open inventory/stash and move around in there for x amount of time and my screen goes black and the computer reboots.
And i’ve also noticed a HUGE memory leak. D4 Uses almost 100% of my ram with nothing open in the backround.
Here are my specs:
Operating System
Windows 10 Pro 64-bit
CPU
Intel Core i7 10700K @ 3.80GHz
Comet Lake 14nm Technology
RAM
G.Skill Trident Z 32GB Dual-Channel DDR4
Motherboard
ASUSTeK COMPUTER INC. PRIME H570M-PLUS (LGA1200**
GPU: Nvidia RTX 3090 Founders Edition
PSU: EVGA SUPERNOVA 1050W GOLD
DISPLAYS:
Asus VG34v (3440x1440 @165hz) < Monitor that i play diablo on.
AOC 22V2WG5 (1080x1920 @75hz)
STORAGE:
931GB Samsung SSD 860 EVO 1TB (SATA (SSD)
465GB Samsung SSD 970 EVO Plus 500GB (m.2)
Check that the power supply is connected correctly. The connection should be tight and separate cables should be used for each video card connector. It is also undesirable to use adapters as they can cause problems.
This statement needs to be verified. For example, enable logging in some utility such as gpu-z, and after a failure, look at the end of the log.
Asus anti surge can also be falsely triggered. But I think if there is something, it is not a false alarm.
Exact SAME issue, I thought D4 had fried my PSU and was considering getting a new one.
I even thought I had micro-powercuts because of the total reboot/crash of the PC. Not even a blue screen or an error message. I checked the Windows event logs and it’s the same as if I had just pulled the plug on the PC, there’s 0 error prior to the unexpected reboot.
Yesterday I tried reducing the max frame-rate in game from 140 to 120 and it didn’t crash. I’ll test today again, maybe it was luck (it was), or maybe it’s related to the performance/patch. I have an 3080 Ti as a reference.
This game is really poorly optimized for the budget and the teams that went into it. There are some serious QA issues here.
Also, just for clarification: I checked my PC vitals and I’m not getting over 85C even at 100% use of the GPU, but I still suspected it might be that. Reducing the frame rate to 120 brought to under 100% gpu usage most of the time, so maybe that’s why it reduced the crashes.
Edit: Nope, the 120 frame cap did nothing, I guess yesterday was just luck. I just crashed this morning after 3mn of Helltide play.
The motherboard VRM may be too weak for this processor. If the CPU is cooled by a good cooling system, the VRM may overheat. There are probably no sensors on the VRM.
It is not quite clear what such a cheap version of the motherboard is doing with such an expensive processor.
The game doesn’t use 100% CPU, so it’s never really going to pull max sustained wattage that would overheat the VRMs. At most, you’ll see maybe 25% usage. I see less than 20% on average on a 13600.
Some players reported that they have the game loads the CPU completely. This may be a compilation of shaders. Or even an OS update or other unplanned load.
Yeah maybe, but that would be for like seconds at a time max. Most VRMs can handle like 10-30 seconds of PL2 before dropping back to PL1. If you want to be able to run in perma-turbo-boost, you need a board with good VRMs though. So even if the VRMs were stressed and thermal throttled, all that would happen is the CPU would drop from boost to base.
Just saying though, it’s extremely rare for my CPU usage to ever go above 20% and the only time it happens is for like 1-2 seconds at a time if there are a bunch of brand new shaders that have to be compiled (like right after updating graphics drivers when it wipes/invalidates the cache). After even an hour of playing and warping around, you don’t see it again.
EDIT: I should correct this by not going based off usage, since that’s related to core/thread counts. In terms of wattage, my client never really goes above 50W on the CPU. This is a 125w base/180w turbo CPU. So pretty much even a 510/610 could handle that without ever reaching VRM limits while playing D4.
This motherboard can attempt to power a CPU at +200 watts and fail. This is a rare motherboard/processor combination that is probably poorly tested. And if the only limitation is temperature, trying to deliver those +200 watts.
510/610 chipsets don’t have the ability to overclock the CPU like that, they won’t try to do it.
Also the air in the case may already be warmed by a +300 watt video card. You don’t even need some serious load to overheat everything in the case when you have 300+200 watts of heat.
It will just supply whatever the max output of the VRMs can handle. For instance, if they tap out at like 150w, then you’ll get some CPU output somewhere between base and turbo. Anyways, the point is that it’s not likely the issue, since D4 doesn’t really load CPUs that hard vs something like rendering that will pull the full 100% sustained out of the CPU. Pretty much any board’s VRMs can handle 50-75w sustained indefinitely(they’d only be dissipating 10-20w of heat from the VRM clusters if you assume ~80% downstep efficiency). And VRMs can be rode hard, like perma-120C hard and pretty much every other capacitor on the board will go bad before the ones inside them do.
PL2 mode isn’t overclocking, it’s just the turbo mode of the CPU.
So be it. Another marketing bullshlt.
Who cares what it’s called if it gets you in trouble?
The problem is how that heat is dissipated. And it’s not dissipating very well. Passive cooling and most of it is dissipated on the board and next to a 200 watt heat source that dissipates some of its heat including through the motherboard using it as a heatsink too.