PC Troubleshooting and Hardware Monitoring · 4:43am Aug 3rd, 2023
Warning: Not pony-related in the slightest.
This blog entry is primarily to gather second-and-so-on opinions from more tech-savvy readers of mine on my observations during hardware monitoring during the running of several demanding games.
Tonight, after an 11-hour workday, I sat down with my preferred hardware monitor (MSI Afterburner v4.6.4) to see temps and usage of both the GPU and CPU during loading, gameplay, and menus. Afterburner gives me real-time graphs of core clocks, temps, usage, and in-game framerates.
GPU: EVGA 3080 Ti FTW3
CPU: Intel i7-9700k
RAM: 32GB DDR4
Resident Evil 2 Remake (2019)
Average GPU temp: 68C
Surprisingly optimized. Butter smooth 60 fps at 4k.
Game was set at 120 fps after booting, had to manually set at 60.
Resident Evil 4 Remake (2023)
Average GPU Temp: 83C
Demanding, despite running on the same engine as RE2.
Game was set at 120 fps after booting, had to manually set at 60.
Halo Infinite (2021)
Average GPU Temp: 77C In-game, 83-84C in menus.
Game is actually more resource-intensive in the menus than in the game, this has not been patched or improved since launch.
Mostly smooth 60 fps in-game.
Dead Space Remake (2023)
Average GPU Temp: 86C
Very demanding. GPU thermal throttling did not bring it down from 86C, but did lower average framerate to high 40's.
Prolonged 100% load on CPU during the loading of the shader cache. Average CPU temp during shader load, 77C, down to mid 60's in-game.
Framerate jumped to 333 fps during shader cache loading, no spike in GPU temps or usage.
No Man's Sky (2016) (This was a game that crashed previously at 4k120.)
Average GPU Temp: 81C
A confusing game from a monitoring standpoint. With infinite procedural generation, some have said this is a very CPU-heavy game. While CPU usage spiked in several places, it was never to 100%. Lag spikes from 0.5-1 seconds long when going from space to atmosphere, vice versa, and landing on a planet. Average CPU temps during all activities ranged from high 50's to low 60's (flying in space, flying in atmosphere, walking on the surface).
GPU utilization never went above 55%. A mostly-smooth 60fps, with very slight hiccups occasionally.
Like I said, confusing. Procedural generation should be intense on the CPU, but then again, they got this damn game to run on the Switch.
American Truck Simulator (2016) (This was a game that crashed previously at 4k120.)
Average GPU Temp: 58C
This was the last game that crashed on Monday. At 4k and 120 fps, it was pushing the GPU into the high 80's. After correcting the settings, high 50's.
Absolutely minimal CPU usage in-game. CPU temps averaged mid-50's throughout.
Conclusions and thoughts:
I started today pretty sure that I was going to have to replace my 4-year old closed-loop CPU cooler. Now, I'm convinced it's working properly, and I'm more confused than I was yesterday.
On the advice of readers in the previous blog, I went into Event Viewer, where I found six "Critical" level events in the last 30 days, all the same, roughly 4-5 days apart, with the same error code and task number: Error Code 41, Task Number 63, which is basically, "system shutdown unexpectedly." I was unable to view any other non-critical events, as I believe only the criticals go back more than 24 hours, and by the time I looked, the last random restart was about 25 hours ago.
On the advice of a slightly more tech-savvy friend, I opened the case and checked the how well the CPU cooler was seated; it might as well be super glued to the motherboard; no give or wiggle room or play whatsoever.
I can only account for the games I was playing for two of the six random restarts; ATS and No Man's Sky.
If these restarts continue to happen after hours of gameplay, I'll continue making notes, but nothing out of the ordinary happened during tonight's monitoring and troubleshooting.
So for now, I don't believe I'll be replacing anything, and will start to attempt to set aside some dough for a new build, because this rig sure isn't getting any newer. Best case scenario, I think, would be things just got too toasty because I didn't realize some games were trying to pump out 120 fps at 4k, when the hardware simply isn't cut out for that level of performance.
Those results look kinda weird to me
, but do neatly support a "GPU done got too hot" conclusion (esp. since the issues started hitting during peak summer).
Couple of questions (if you have them in your data, don't bother re-running and enjoy your rig if it has stopped crashing):
I'm not yet ready to pin all the blame on an overworked GPU
, and am very suspicious of your PSU being overloaded/overwhelmed 
As for a new rig it will depend on what level of performance you expect/demand. You probably still got ~2 years with proper good performance out of your current one yet (modulo streaming
) unless game makers surprise me even more[1] 
or you want to jump to the new shiny resolution/refresh rate. And you can afford to (and, in this rando internet person with little to no awareness of your current context, should 
) wait until either Intel gets all its ducks in a neat row and/or AMD puts out a second-gen DDR-5 platform (Then again, my last rig lasted me ~8 years with very small upgrades inbetween 
).
If you were to upgrade now, you'd have to change at minimum CPU, Motherboard, and, if jumping to AMD, RAM (I believe 13th-gen Intel still supports DDR-4, complete with motherboards supporting both -- either up to 2 DDR-5 DIMMS, (x)or up to 2 DDR-4 DIMMS). You should also consider changing your AIO (even if you can find the adapter plates for the CPU you end going to) and storage (if it's original to your build for reliability reasons[2] and/or not yet PCI-E Gen 4 for performance reasons). Don't bother with GPU (Nvidia 4xxx isn't worth it for you), and the PSU is a toss-up (wait until you need/want extra power or port compatibility)
[1]: Offer void for storage capacity/speed
[2]: If it's purely/mostly a gaming rig, and unless you rotate your cast of installed games (as opposed to install and forget) it shouldn't have too high a number of writes, but I just can't bring myself to put that much trust in SSDs
5740519
My PSU is a EVGA 1200W, so unless something is seriously wrong with it, I don't think it's power delivery.
5740519
Yeah, I'm not looking to break the bank with a new build, I was planning on sticking with Intel, and bringing over RAM, GPU, and perhaps the PSU from this build, cutting down costs considerably.
I just want a HUGE case this time so I can do a Noctua DH15 air cooler, and that thing is a monster.
In my experience vague issues like this something turn out to be either a defect of some kind or something really fucking stupid.
5740519
As to your other questions, I was honestly pretty focused on usage and temps this time. If I do this again, I'll look at power usage and individual cores.
I am going to suggest also using CPU-Z Hardware Monitor for a check of more than Afterburner covers.
Watch your package temp, and double check against Afterburner.
What cooler is mounted? If you don't know the name post a pic and we'll probably be able to tell you. I ask because the Intel stock cooler is legendary at not performing well.
It does sound, like others have said, that this is a CPU issue. However, there could be other causes and ruling out the easiest.
5740544
Aftermarket closed loop. NZXT Kraken with a 240mm radiator.
5740547
Well, there goes my theory.
Sheesh those GPU temps are high. I would look to mitigate that before moving any further into troubleshooting.
5740624
Case is too small for a two- or three-fan radiator. My case was too small, otherwise that is what I would have done.
Also considering stepping back from 4k gaming, going back to 2k or Full HD if I decide on triple screens.
5740629
My current case
They're pretty cheap on the used market if you're thinking about upgrading, lol.
Currently have 8 Noctuas keeping everything cool, never have issues with temps. Airflow is king.
5740650
Yup, definitely going for a super tower for my next build; either Corsair or the Lian Li V3000 that caught my eye at Microcenter last weekend.
And yeah, I want something that big so that I can fit a Noctua D15 in there, as well as have every single fan in there being Noctua. I've heard too many amazing things about their fans to prioritize RGB over them next time.
5740653
Oh man that Lian Li is gorgeous!
I can vouch for the D15, it's a fantastic cooler.
5740654
My current case is a NZXT H510.
The D15 would just barely not fit, sadly.