Frequent PC crashes on Apex Legends

shoegaze_ibex

New Member
Hi everyone,

I'm trying to solve some fairly frequent crashes I've been having. I play only Apex and Valheim. It never crashes on Valheim but it crashes fairly often on Apex. When it crashes the screen freezes, the sound drops out, and the PC will hang there until I hold the power button. My airflow is not great and the CPU temps stay around 80 during Apex, but I think from what I've read with an i7 4790k this shouldn't be crashing the PC like this? It isn't overclocked by the way. Nothing is jumping out at me in Event Viewer leading up to the crashes but trying to solve this is by far the most in depth troubleshooting I have done so a lot of this is very new to me.

I've downloaded Generic Log Viewer and recorded a log leading up to a crash but making sense of it is definitely over my head. I've attached the log file below. I'm just not sure where to go from here, please help me haha.

Gryphon Z87 mobo
GTX 1080
i7 4790k
2x 8GB DDR3 1600MHz
 

Attachments

  • 10.26.2021.zip
    138 KB · Views: 4
Last edited:

Dalai

Well-Known Member
The log doesn't show anything out of the ordinary - temps, voltages, wattages are all fine. If you suspect bad airflow as the reason, then open the case and let some big fan blow at the system, but somehow I doubt that will help (unless some temperature that's not being monitored is too high).

The only thing I see you can do is to try to find other games or programs that lead to similar issues. I recommend Prime95 and FurMark for this, although I suggest to be careful with the latter as it's kind of a power virus, i.e. don't run it for too long. Run each of them on its own, then run them both together and see if any crash occurs.

Did you try different versions of the nVidia graphics drivers? Did you use DDU (in Windows Safe Mode) to remove it completely from the system?

Regards
Dalai
 

shoegaze_ibex

New Member
The log doesn't show anything out of the ordinary - temps, voltages, wattages are all fine. If you suspect bad airflow as the reason, then open the case and let some big fan blow at the system, but somehow I doubt that will help (unless some temperature that's not being monitored is too high).

The only thing I see you can do is to try to find other games or programs that lead to similar issues. I recommend Prime95 and FurMark for this, although I suggest to be careful with the latter as it's kind of a power virus, i.e. don't run it for too long. Run each of them on its own, then run them both together and see if any crash occurs.

Did you try different versions of the nVidia graphics drivers? Did you use DDU (in Windows Safe Mode) to remove it completely from the system?

Regards
Dalai
Thanks so much Dalai. I've tried using DDU in Safe Mode to completely remove the nVidia drivers and then installed the latest one and the crashes persist. I downloaded WhoCrashed and it's reporting that the crashes are caused by video TDR errors but this information hasn't led me to a fix yet. I'll download Prime95 and FurMark as you recommend. I recorded a few more logs leading up to crashes, does everything still look fine on these?

In any case, I truly appreciate your feedback and advice and thanks again!
 

Attachments

  • hwinfo logs 2.zip
    1.9 MB · Views: 2

Dalai

Well-Known Member
I recorded a few more logs leading up to crashes, does everything still look fine on these?
Yes, they still look fine - after removing two additional lines from one of them which caused Generic Log Viewer to show a linear line to match the data points.

In any case, I truly appreciate your feedback and advice and thanks again!
You're welcome.

A couple more questions:
  • Since you have a GTX 1080, I guess it uses two PCIe power connectors. Do you use separate cables for them or did you daisy-chain them? The latter can be problematic, especially for higher-power cards.
  • Which PSU do you have? Do you have a spare one, so you could swap them temporarily?
  • When did the crashes occur first? Any change in hardware or software around that point in time which could have triggered this?
 

TomWoB

Well-Known Member
Hi shoegaze_ibex,

in your first logfile you have many "Performance Limit" = YES entries (e.g. "Performance Limit - Reliability Voltage"). I don't know exactly what this means, but it doesn't sound really good (value "1" = "YES")

PerformanceLimit.png

Tip: play Apex with "low graphic settings", to reduce power consumption, to check if there is a "power problem" ...

Regards
Tom
 

Dalai

Well-Known Member
@TomWoB:
That's alright. There's always some performance limiting factor, either voltage, power, temperature, utilization etc. for GPUs and CPUs, or any chip for that matter. Reliability voltage means that the chip can't go any higher with its (core) voltage without risking stability. If the voltage could go higher without any limit, the chip would damage itself.

I agree on the test with different graphics settings, though I don't really think that it's going to reduce the power consumption, at least not that of the whole system. Lower graphics settings makes the CPU work harder and the GPU work less, so the GPU will (probably) consume less power but the CPU might consume more power. Still, it might reveal an issue with the PSU or the PCIe power connectors.

Regards
Dalai
 

shoegaze_ibex

New Member
Yes, they still look fine - after removing two additional lines from one of them which caused Generic Log Viewer to show a linear line to match the data points.

You're welcome.

A couple more questions:
  • Since you have a GTX 1080, I guess it uses two PCIe power connectors. Do you use separate cables for them or did you daisy-chain them? The latter can be problematic, especially for higher-power cards.
  • Which PSU do you have? Do you have a spare one, so you could swap them temporarily?
  • When did the crashes occur first? Any change in hardware or software around that point in time which could have triggered this?

Hi again,

  • I checked and my PNY GTX 1080 is a version that uses only one 8-pin PCIe connector.
  • I have a this 650W XFX PSU. I don't have another one, but I suppose I could Amazon one to test.
  • I recently upgraded from an i5 4670 to my current i7 4790k. The same sort of crashes were occurring before the CPU change but have actually become more frequent since- that is until about two weeks ago when I tried changing the Power Management Mode in nVidia Control Panel to "Prefer maximum performance" under the Global Settings tab. This has greatly decreased the frequency of the crashes, usually a single crash every few days instead of 2-3 a session.

It's also turned out the crashes aren't limited only to Apex. In the few weeks since I last posted I've had the same type of crash on two other games(Grounded, after playing for a bit, and Valheim, as soon as the world loaded- if that matters) and once immediately upon opening Chrome.

I've run Memtest86 with the default 4 passes with 0 errors. I also downloaded Prime95 and Furmark as you recommended- I've run them multiple times and they have not produced a crash. I've included logs for each test as well as a log of running them simultaneously. I haven't captured a log leading up to a crash since changing the power setting in nVidia Control Panel as I initially thought that had solved it, but I'll be sure to run HWiNFO next time I have time to play anything.

Thanks and all the best.
 

Attachments

  • shoegaze_ibex furmark and prime95 logs.zip
    404.5 KB · Views: 2
Last edited:

Dalai

Well-Known Member
Hi.

Thanks for the logs. I just checked them, and it looks like the CPU cooling is insufficient. After a few minutes running Prime95 the CPU temperature reached Tjmax (somewhere between 90 and 100°C) and it began to throttle. Admittedly, during gaming the CPU isn't under as much load as under Prime95, but it's still an indication that the cooling is not good. A CPU shouldn't throttle regardless of load.

As for the "Prefer maximum performance" setting, I'm not sure if this setting reduces the frequency with which the GPU (core) switches to a lower power state. If so, it could mean that the GPU also has some issues.

Which BIOS version do you have installed? The 4790K requires at least 2003 according to ASUS, and newer BIOSes improve stability (that's what they claim anyway).

Regards
Dalai
 

shoegaze_ibex

New Member
From what I understand, that does sound like what the maximum performance setting does. I've read it prevents the GPU from dropping to idle clock speeds.

My BIOS is 2103, however I updated it years ago to correct a problem I was having with Windows properly syncing the date and time and I did so using the ASUS BIOS Updater utility. I checked earlier today and noticed that in System Info the date of the BIOS version reads 8/15/2014 instead of 9/12/2014 as it does on the website. This seemed weird and I've also read that using BIOS flashback and a USB is encouraged over the Updater utility due to reduced risk of errors. So I just (re)flashed BIOS 2103 using the USB method. The date of the version still shows 8/15/2014 but I guess all I can do is cross my fingers that using the Flashback method rather than the updater program will improve stability.

I don't think I can do any better in the CPU fan/heatsink department due to microATX space constraints, but I have two pretty weak case fans right next to the CPU that I could try upgrading. I've also watched a couple videos of people delidding their 4790k's and getting around a 10° decrease.

By the way, should I be running Prime95 at default test settings or is there something I should change?

Thanks again
 

Dalai

Well-Known Member
The date of the version still shows 8/15/2014 [...]
That's normal. A BIOS is compiled or built at some point in time and might get published days or even weeks after that. Manufacturers usually test their software, which takes time. Although some software doesn't seem to be tested at all sometimes...

I don't think I can do any better in the CPU fan/heatsink department due to microATX space constraints [...]
It's less the size of the mainboard limiting the choice of CPU coolers but more the case size and layout. The thing with downdraft coolers is that they might cool the CPU quite well (and also the VRMs) but the hot air is not exhausted out of the case as well as with a tower cooler. This can result in higher CPU temperatures.

A couple of years ago I had an older platform and used a Noctua NH-L12 on it - which is of a similar design as the Scythe Big Shuriken 3. The VRMs were cooled quite nicely, but the inside of the case got pretty hot, causing all the fans to ramp and become louder and it also resulted in high CPU temperatures. Because of the CPU cooler's design and its airflow the rear case fan exhausted primarily cold or mildly warm air - which is not the best it could do. I also tested different fan orientations and configurations on the cooler but in the end I went back to the Noctua NH-U9B and mounted its two fans as low as I possibly could to direct some air over the VRM heatsinks. Overall it was the better experience.

Which case do you have? Is it large enough to fit a small tower cooler with a 92mm fan on it? If cooler height is a concern, some of these tower coolers have a height of just 125-135mm, like the bequiet Pure Rock Slim 2, or the Noctua NH-U9S (or U9B SE2), or the Cooler Master Hyper TX3 Evo. However, I'm not sure if the TX3 would be able to handle the heat load at an acceptable noise level.

I've also watched a couple videos of people delidding their 4790k's and getting around a 10° decrease.
Although I don't doubt these claims, you should keep in mind that this is a dangerous procedure which could result in the CPU's death in the worst case. Even if it's successful, the CPU is still going to run at 90°C instead of 100°C. But that might not be the result because due to the somewhat larger (theoretical) temperature headroom the CPU might throttle less, which, in turn, results in higher clocks and higher temperatures in turn. You'd be back to where you're now, more or less.

I don't think delidding is the first thing to consider when dealing with these temperatures. Also keep in mind that high temperatures inside the case causes everything to heat up. Well, except the drives if they get cold air, which is most often the case.

By the way, should I be running Prime95 at default test settings or is there something I should change?
Usually I use the Torture test with small FFTs and let it go until the temperatures are not climbing anymore.

Regards
Dalai
 
Top