PC rebooting while ingame [FIXED]

[Pest Fest]

Hello developers, you're doing a good job so far and I love your game, even though it often crashes my pc.

I often can't play for 10 minutes before the game crashes my entire PC and it simply restarts. So I never get an error message.

The temperatures are all great, I monitor them constantly.
Until a few days ago I had a different setup with a Ryzen 9 5950x, I had no problems there, now I have a Ryzen 7 7800X3D and I have constant crashes.

Here's the rest of the hardware.
Case: Corsair 7000D Airflow
Mainboard: MSI MAG X670E Tomahawk WIFI with the latest Bios 1.90
GPU: RX 7900XT
RAM: Corsair Vengeance 32GB DDR5-6000, CL30-36-36-76 EXPO
CPU: Ryzen 7 7800X3D
CPU Cooler: Arctic Liquid Freezer II 420 (push pull)
PSU: Corsair Professional Series 2023 HX1000i 1000W
The rest of the hardware shouldn't be relevant to the crashes.

As you can see from the power supply, I have no problem with the power supply (voltage peaks) and the cooling is also oversized, the cpu doesn't get above 70°C

I tuned the CPU with the Curve Optimizer, but it makes no difference whether I use the tuning or not. It's the same with the RAM, it makes no difference whether I use EXPO or not. The system otherwise runs stable, prime95 stable for 24h+, furmark stable, all other games run without problems, I even have no problems with a stream that is encoded via the cpu.
The temperatures of the SSDs and the RAM and VRMs are also very good.

I used the latest graphics card drivers (24.3.1 & 24.4.1) and also the latest chipset driver 6.02.07.2300, Windows 11 Pro was also reinstalled. The game files have also been verified.

I feel like Hotfix03 made things worse.

What source of error did I miss?

I hope you can work with the information =)

[Pest Fest]

First of all: GZW is running stable
what did I do for it?
The game uses AVX2 and this can cause CPUs that are undervolted, e.g. with the curve optimizer, to crash.
I found with CoreCycler/y-Cruncher/prime95 that the CPU was not stable with AVX2, with FFT size "All" and with the y-Cruncher mode "19-ZN2 ~ Kagari" and a runtime per CPU core of 15 minutes . If you undervolted normally with SSE and the FFT size Huge, everything ran stable, but that wasn't enough for gzw.

The old undervolting settings were stable enough for all games and benchmarks over days, but not stable enough for AVX2/GZW.

TLTR: the cpu must not be undervolted so severely, the settings must be able to withstand AVX2.
I tested the whole thing for stability with CoreCycler 0.9.5.0alpha2 and haven't had any crashes since then.

Starting the game as an administrator doesn't hurt either.

If anyone would like to have the exact settings for CoreCyler, I can share them.

[TheWarden]

If the PC is restarting, the likely culprit is the PSU

[Sn4pFPS]

Set cpu fan at max speed , i bet temps going 100+ and pc reboot.

[Pest Fest]

Messaggio originale di Sn4pFPS:

Set cpu fan at max speed , i bet temps going 100+ and pc reboot.

temps are fine, cpu is never over 65°C, cooled by an artic liquid freezer II 420 AIO with 6 fans (push/pull), and the gpu temps are ok too, no thermal throttel

[Pest Fest]

Messaggio originale di TheWarden:

If the PC is restarting, the likely culprit is the PSU

is anybody reading the post? No its not the psu, as i wrote, it was the undervolting, it wasnt stable with avx2.

[CHEFJUSTICE]

Dude, you saved my Life. I had ecatly the same specs nad same problem. All benachmarks instga crash and reboot. I gotta read up that AVX2 ♥♥♥♥. Never heared of it. It solved all my problem. Change some hardware ordered a new cpu. Just for this ♥♥♥♥ option in bios you have to disable. Thanks amd

[KickAss]

i have a 8k computer state of the art 4090 etc literally best in slot every peice and it restarts randomly alot

[Chameleon_Silk]

Messaggio originale di Pest Fest:

First of all: GZW is running stable
what did I do for it?
The game uses AVX2 and this can cause CPUs that are undervolted, e.g. with the curve optimizer, to crash.
I found with CoreCycler/y-Cruncher/prime95 that the CPU was not stable with AVX2, with FFT size "All" and with the y-Cruncher mode "19-ZN2 ~ Kagari" and a runtime per CPU core of 15 minutes . If you undervolted normally with SSE and the FFT size Huge, everything ran stable, but that wasn't enough for gzw.

The old undervolting settings were stable enough for all games and benchmarks over days, but not stable enough for AVX2/GZW.

TLTR: the cpu must not be undervolted so severely, the settings must be able to withstand AVX2.
I tested the whole thing for stability with CoreCycler 0.9.5.0alpha2 and haven't had any crashes since then.

Starting the game as an administrator doesn't hurt either.

If anyone would like to have the exact settings for CoreCyler, I can share them.

yeah I tried that curve optimizer stuff and all it did was make my system unstable every so often especially at the lower clocks (which is counter intuitive) normally systems are unstable at high clocks but what ends up happening with the optimizer is you get less power at low clocks as well, theres ways around that but after running some tests and actually getting less performance with the optimizer versus no optimizer....

theres also the fact that I'm gpu limited in 4K and practicially 0 difference if im 3.7GHz or 4.6GHz.

I think undervolting and lowering TDP is actually the new field of overclocking, which is funny if you think about it because its the polar opposite.

I much prefer running 48-52c 1v under load than 78-90c 1.45v in the game.

glad you got it sorted.

[Chameleon_Silk]

Messaggio originale di KickAss:

i have a 8k computer state of the art 4090 etc literally best in slot every peice and it restarts randomly alot

that is not normal behavior, something is running to hot - your PSU is not up to the task for the 4090 - or your RAM is mismatched pairs, overpopulated slots and/or incorrect voltage or timings. I'm betting you also have overclocked RAM out the box and isn't playing nice.

[Deadly Chicken XoP']

Messaggio originale di KickAss:

i have a 8k computer state of the art 4090 etc literally best in slot every peice and it restarts randomly alot

Generally or just in this game ?

Check thermals, could be overheating. What cooler do you have ? What CPU other than "the best in slot"

[Sergiu]

FOR EVERY INTEL CPU OWNER....the solution is extremely simple for cpu's overheating regarding how good your AIO. etc is ....JUST disable HYPER THREADING in BIOS and problem solved. There is absolutely no performance difference

[Blue]

Messaggio originale di Silent:

Me and my friends that have intel and ryzen 5800x, 5700x, hand 0 crashes and issues with this game so far.
Stop concentrate on your pc, if other works fine, this is unreal engine 5, anything can happen
Ofc the game need optimization and servers.
I use asus x670E-F, 7800x3d and 4080, and i never had one bug or issue that break my game.

You sound really lame. It can entirely be the game that doesnt work well with a system. Instead of a system not working well with a game.

[Silent]

Messaggio originale di Blue:

Messaggio originale di Silent:

Me and my friends that have intel and ryzen 5800x, 5700x, hand 0 crashes and issues with this game so far.
Stop concentrate on your pc, if other works fine, this is unreal engine 5, anything can happen
Ofc the game need optimization and servers.
I use asus x670E-F, 7800x3d and 4080, and i never had one bug or issue that break my game.

You sound really lame. It can entirely be the game that doesnt work well with a system. Instead of a system not working well with a game.

That post is old, sir.

Your profile picture doesn't appear mature, so I will make an effort to be nicer than you.

Are you attempting to assist someone, sharing about your gaming experience, or simply trying to come across as a philosophical arrogant?

[Krasnoyask]

I was having constant shut downs with an 1500W PSU, turns out it was a PSU issue. After reading the PSU manuel, it mentioned PCI-E "A" connection was only for a GPU with one or two eight pin connectors while PCI-E "B" was for a GPU with three or more 8 pin connectors. I read this post trying to fix my issue, having plugged all GPU eight pin connectors to "B" connections no more issue.
Played twenty eight hours since no issue, couldn't even start a mission before.
Be-Quiet Dark Power Pro, maybe more manufacturers do this.

[Warmech]

Messaggio originale di Krasnoyask:

I was having constant shut downs with an 1500W PSU, turns out it was a PSU issue. After reading the PSU manuel, it mentioned PCI-E "A" connection was only for a GPU with one or two eight pin connectors while PCI-E "B" was for a GPU with three or more 8 pin connectors. I read this post trying to fix my issue, having plugged all GPU eight pin connectors to "B" connections no more issue.
Played twenty eight hours since no issue, couldn't even start a mission before.
Be-Quiet Dark Power Pro, maybe more manufacturers do this.

Yea this sounds like a common multi-rail PSU setup where "A" and "B" refer to different power rails that can output a limited amount of amps on specific connectors. Typically you want to either separate out your CPU/Mobo and GPU connectors onto different rails, or split evenly across rails (depends on the PSU) to prevent overcurrent protection from tripping.