A real good one for you all - PC restart absolutely randomly.

[QuantumSingularity]

Ok guys i have a real good one for ya. A rig i've built for a buddy of mine is acting really weird. Sometimes it restarts itself (not shutting down) the moment a 3D application is launched aka the GPU starts drawing power, sometimes it works for 10-15 minutes and then restarts and sometimes it works fine for tenths of hours under max stress test w/o a single issue. 2 days ago i left it running 12 hours of Furmark 2 + OCCT CPU stress and it was absolutely stable. Then in the night i left in running with Cyberpunk 2077 and the character sitting in the most intense part of Dog Town throughout the whole night and again not a single issue. Then i restarted it and literally 3 minutes after that 24+ hour torture test, the moment i decided to launch any game, doesn't matter which one, it restarts itself right away. It's basically the same rig as mine - 5800X3D, X570 Steel Legend, 1000W Focus GX PSU, with the difference being 3600MHz g.skill memory and RX 7900XT. All the usual suspects have been tested - overheating CPU/GPU, mobo temps, SSDs, memory test. All the latest drivers and firmware updated and reinstalled a couple of times. I tried swapping components from my system to the other one by one and all of them work fine when mixed, but when i put his system back together it starts acting weird whenever it decides. Thought it was some driver bug/issue and tried completely fresh windows install as well - still the same thing. I've been busting my head over this one for quite a few days now to no success. I've tried swapping PSUs, GPUs, CPUs, motherboards, memory, coolers. Everything you can think of in all possible permutation and combination. I really hope some of you can think of something i've missed, because my head already hurts from all the thinking.

[rwdfsawe]

It could be power issue,overheating,something broken,malware or just the operating system problem.

[podkall]

quoting figure 8 from PSU-tierlist for Seasonic PSU (A-tier), could be unrelated though:

 

Quote

Seasonic PRIME based units experience shutdowns with RTX3080/3090 (and possibly RX6900 XT) GPUs. The cause is not the OCP tripping but a PSU design flaw as evident by the PSU not latching off on shutdown and 1000W+ models being affected too. Doesn’t manifest in 100% cases as it’s also dependent on motherboard model and GPU OC. Seasonic provides a replacement 24-pin ATX cable to fix this via support. Appears to be fixed post 2021 although there’s no official confirmation from Seasonic.

 

[filpo]

10 minutes ago, QuantumSingularity said:

memory

was this RAM at 3600 too? could be the RAM 

 

11 minutes ago, QuantumSingularity said:

X570 Steel Legend

I have to ask. Bios updated? and if it is are you on a beta bios?

 

12 minutes ago, QuantumSingularity said:

the moment a 3D application

so doesn't happen any other time? Sounds GPU related

 

12 minutes ago, QuantumSingularity said:

I've tried swapping PSUs, GPUs, CPUs, motherboards, memory, coolers.

and nothing worked?

 

1 minute ago, podkall said:

quoting figure 8 from PSU-tierlist for Seasonic PSU (A-tier), could be unrelated though:

 

 

Unfortunately it's not a PRIME unit, it's a FOCUS

13 minutes ago, QuantumSingularity said:

1000W Focus GX PSU

[podkall]

If slower RAM XMP/off doesn't fix it, try running the GPU on lower voltage just to see if that somehow helps

[QuantumSingularity]

15 minutes ago, podkall said:

If slower RAM XMP/off doesn't fix it, try running the GPU on lower voltage just to see if that somehow helps

Slower ram doesn't fix the issue, neither does lowering power target and manually undervolting and underclocking the GPU.

26 minutes ago, filpo said:

was this RAM at 3600 too? could be the RAM 

 

I have to ask. Bios updated? and if it is are you on a beta bios?

 

so doesn't happen any other time? Sounds GPU related

 

and nothing worked?

 

Unfortunately it's not a PRIME unit, it's a FOCUS

BIOS for mobo, VBIOS on the GPUs, firmware for the drives - everything is the latest there is. It was the first thing i did when he got the system to me.

 

28 minutes ago, podkall said:

quoting figure 8 from PSU-tierlist for Seasonic PSU (A-tier), could be unrelated though:

 

 

Although the unit isn't the Prime series, It's Focus GX, my gut feeling was also pointing towards PSU for some reason. It's the only thing i can't accurately stress test, because i don't have a test bench with custom loads and an oscilloscope. The thing that puts me off is that when i put it inside my system there are zero issues and it works perfectly fine. It also works absolutely fine with his GPU overclocked and draws over 340W, but sometimes acts up when the card is downclocked and doesn't exceed 280W. Also as mentioned - often the system would restart itself the moment i press Run on the Furmark test or the loading screen for any game appears.

[filpo]

1 minute ago, QuantumSingularity said:

but sometimes acts up when the card is downclocked with and doesn't exceed 280W.

might be that that's a problem

 

1 minute ago, QuantumSingularity said:

BIOS for mobo, VBIOS on the GPUs, firmware for the drives - everything is the latest there is. It was the first thing i did when he got the system to me.

have you tried the ram at 3200? or other ram at 3200?

[podkall]

43 minutes ago, QuantumSingularity said:

Although the unit isn't the Prime series, It's Focus GX, my gut feeling was also pointing towards PSU for some reason. It's the only thing i can't accurately stress test, because i don't have a test bench with custom loads and an oscilloscope. The thing that puts me off is that when i put it inside my system there are zero issues and it works perfectly fine. It also works absolutely fine with his GPU overclocked and draws over 340W, but sometimes acts up when the card is downclocked and doesn't exceed 280W. Also as mentioned - often the system would restart itself the moment i press Run on the Furmark test or the loading screen for any game appears.

have you tried testing something that really cracks the fps? because not just Graphical demand can stress GPU but trying to do as many frames as possible too

[QuantumSingularity]

Tried a few games with AFMF. If it works, it works no matter of the settings or the game. But when it doesn't it simply refuses to work at all. The instant reboot after launching any 3D application is really interesting. It's not CPU related because no matter how much i stress the CPU no issues are present. It's not memory. In my mind it's narrowing down either to GPU or PSU, with as mentioned my gut feeling pointing towards PSU. Anyone with any idea how to stress the PSU specifically?

As for the GPU although OC-ed it goes beyond 340W, it never exceeds temp of 77°C with Hotspot of 85°C. Well within the specs for the Nitro+ Vapor-X RX 7900XT, which btw is advertised as 368W Total Board Power card on Sapphire's site.

[filpo]

20 hours ago, QuantumSingularity said:

Anyone with any idea how to stress the PSU specifically?

you could try it in your own system

[Mark Kaine]

ima be blunt: its likely the ram, ram tests are useless for Ryzen memory compatibility issues, and the symptoms fit. 

your friend is drinking the amd coolaid and thats how that tastes (unreliable drivers) get a Nvidia gpu.

focus psu is awful at best (source: i had one) replace. (recommended: RMX, RMi, Leadex III)

 

 

one or all 3 of these things will fix it almost guaranteed. 

 

On 3/6/2024 at 7:49 PM, QuantumSingularity said:

3600MHz g.skill

one thing tho, exact model number? 

[QuantumSingularity]

Okay i think i have another suspect - can it be the cablemod cables? Their connectors are like brand new with no wear on them, they make perfect contact, but the moment i put the system back together in the case, all these issues start to appear again. When i have it all out and stress testing with the original PSU cables, there seem to be no issues. 

 

As for the PSU, tested it with the same mobo, a 5950X and a 3090Ti - absolutely no problem whatsoever. Currently i'm trying to hook it up with the 7900XT and the 6900XT together and try to run them both simultaneously with FurMak on the same system.

 

As for the memory - his kit is this exact one: https://www.gskill.com/product/165/326/1562840211/F4-3600C16D-32GTZNC

 

On 3/8/2024 at 3:05 AM, Mark Kaine said:

ima be blunt: its likely the ram, ram tests are useless for Ryzen memory compatibility issues, and the symptoms fit. 

your friend is drinking the amd coolaid and thats how that tastes (unreliable drivers) get a Nvidia gpu.

focus psu is awful at best (source: i had one) replace. (recommended: RMX, RMi, Leadex III)

 

 

one or all 3 of these things will fix it almost guaranteed. 

 

one thing tho, exact model number? 

Honestly idk what you are talking about. I've built over 100 Ryzen rigs in the past 3 years and with the exception for the early DDR5 kits and mobos with early BIOS-es, none of them has had a single issue. This is the first time i am encountering something that kind of problems with any system. Same with the GPUs - pretty much all of the issues i've had with GPUs have been with Nvidia and 90% of those were with Asus GPUs.

 

Kay... so update... the PSU runs perfectly fine with 7900XT and 6900XT together running FurMark, so it's certainly not the PSU. Will assemble the system now w/o the cablemod cables and see if there is any difference.

[QuantumSingularity]

Okay guys. turns out it was really the cablemod cables that were giving the issues. TBH, kinda embarrassed by myself that it took me a whole week to test one of the most fundamental components. Extracted them cables from his system (i had routed them in such way for maximum aesthetics that i had to take everything out and disassemble half of the case to free them) and tried them on mine and instantly started getting the same issues. So i guess keep that in mind the next time you meet an issue like this one. Now it will be interesting to see what their RMA process is.