Ryzen 5000 crashes - anyone getting them?

Techy_Ben · 8 May 2021 at 19:52

Getting crashes (only in desktop and web browsers, now Linux wont boot, but games fine strangely enough) now on a two month old 5600x.

Default settings, Asus ROG Strix B450i with latest bios. Tried 2 PSUs, but neither brand new. 3 different sticks of ram.
So I'm down to an RMA on the CPU, motherboard or a new PSU to test.

Techy_Ben · 8 May 2021 at 19:55

Cadder said:
I almost bought a new keyboard the other day but then it started behaving again. After reading some of the posts on this thread it appears I may be suffering some of these issues.

My Logitech wireless keyboard and mouse sometimes stutter, I was thinking it was the pad for the mouse and just age for the keyboard but maybe the USB Unifying receiver is being affected by this issue on my 5800X/X570 Tomahawk combination.

I hope it gets sorted soon. I'm on latest BIOS and updated drivers now and still noticed some stutter..

EDIT: But on the topic of CPU crashes my 5800X is behaving really well, no issues to report other than the USB problem.

Ah, I was getting mouse spikes a little too. Though less now the entire CPU is giving up.

Techy_Ben · 18 May 2021 at 08:43

Whelp. CPU is on it's way back. Tested and was fine.

smiley for not knowing the fault,

for I guess it not being the CPU.

So that means the motherboard, or other gremlins to find... RAM and GPU working fine on this system in the meantime... might have to bite the bullet and send the 450 platform back and get a 500 whatever.

Grim5 · 18 May 2021 at 09:04

For whatever reason Zen3 is less stable with ram than zen 2. Something in the architecture has made it more unstable cause it's using the same zen2 memory controller. I can right now out my 3950x back in my x570 board and do 3600mhz on my ram but I've tested 3 5950x and they all crash over 3200mhz on my memory

Jimmy Weirdarms · 18 May 2021 at 09:09

Techy_Ben said:
Whelp. CPU is on it's way back. Tested and was fine. smiley for not knowing the fault, for I guess it not being the CPU.

Went through the exact same process! When I got a new mobo it was able to identify that one of the sticks of RAM was bad, and when I got that exchanged then everything was happy.

Cyber-Mav · 20 May 2021 at 10:44

B2 stepping should fix zen 3 failure rate

Cadder · 20 May 2021 at 11:03

My 5800X has behaved really well so far.

Richdog · 20 May 2021 at 15:37

Cadder said:
My 5800X has behaved really well so far.

Mine too, its's super stable with my spec and powers through all my games. Would have loved a 5900x but they weren't in stock and realistically I'm not missing out on too much until I upgrade in a couple of years.

OverDrive · 20 May 2021 at 16:10

Richdog said:
Mine too, its's super stable with my spec and powers through all my games. Would have loved a 5900x but they weren't in stock and realistically I'm not missing out on too much until I upgrade in a couple of years.

Same here, the 5900X was not available when I built my last rig in Feb, so went for a 5800X when OCUK were selling them for £379. It’s been completely stable and I’m very happy. I’ll prob. upgrade in a generation or two.

kitfit1 · 20 May 2021 at 18:44

Grim5 said:
For whatever reason Zen3 is less stable with ram than zen 2. Something in the architecture has made it more unstable cause it's using the same zen2 memory controller. I can right now out my 3950x back in my x570 board and do 3600mhz on my ram but I've tested 3 5950x and they all crash over 3200mhz on my memory

I think the "less stable" statement is not something i recognize at all. I had a 3950x running in the same mobo i got now along with the same 4x8Gb ram. Both the the 3950x and my 5950x run 3733mhz rock solid 24/7. In fact the 5950x will run 3800mhz as well, i just havn't stress tested it at 3800mhz yet.

Richdog · 21 May 2021 at 06:38

Hah I am now getting some crashes... but only as a result of updating to the latest Windows feature pack. It scrubbed all of my restore points and now crashes when I try and create a new one.

The 5800x passes Cinebench R23 with flying colours so I know it's not to blame!

EDIT - Ran a system file check and seems to have repaired things and I can now create restore points. Phew.

AStaley · 21 May 2021 at 07:28

Had my 5950X/Dark Hero for 3-4 months and it's been rock solid, average temps when gaming are mid-60's. When pushing the CPU/980's it spikes into the mid-80's but quickly drops back down.

Grim5 · 21 May 2021 at 11:52

AStaley said:
Had my 5950X/Dark Hero for 3-4 months and it's been rock solid, average temps when gaming are mid-60's. When pushing the CPU/980's it spikes into the mid-80's but quickly drops back down.

that's decent temps if it's overclocked, if it's stock that's very hot

Cyber-Mav · 21 May 2021 at 13:11

Richdog said:
The 5800x passes Cinebench R23 with flying colours so I know it's not to blame!

cinebench is not ideal stability stress tester.

Techy_Ben · 21 May 2021 at 17:43

Jimmy Weirdarms said:
Went through the exact same process! When I got a new mobo it was able to identify that one of the sticks of RAM was bad, and when I got that exchanged then everything was happy.

Not for me, error persists over 3 different sticks of ram (brands/sizes too!). It *might* be low power state crashes only, and Overclockers only did burn in tests on high power state software (prime 95/games/etc, which AFAIK will stop the cores going into sleep state, thus give a false "negative" on the CPU being faulty/ok).

In which case, it might be the Motherboard is failing and not supplying voltage to sleep state cores, thus crashing only Chrome and Linux boot up/etc, where *some* active cores are sleeping, then loosing their cache/etc. While games/Prime95 prevents sleep states, thus those cores never ever fail.

Or it's the CPU doing exactly that. I can only find out with a new motherboard, as Overclockers already returned the CPU, so I can't ask them to test it in low activity (they never tested Edge/Crome/MS word/Excel where sleep states would be allowed).

Joxeon · 21 May 2021 at 18:51

Techy_Ben said:
Not for me, error persists over 3 different sticks of ram (brands/sizes too!). It *might* be low power state crashes only, and Overclockers only did burn in tests on high power state software (prime 95/games/etc, which AFAIK will stop the cores going into sleep state, thus give a false "negative" on the CPU being faulty/ok).

In which case, it might be the Motherboard is failing and not supplying voltage to sleep state cores, thus crashing only Chrome and Linux boot up/etc, where *some* active cores are sleeping, then loosing their cache/etc. While games/Prime95 prevents sleep states, thus those cores never ever fail.

Or it's the CPU doing exactly that. I can only find out with a new motherboard, as Overclockers already returned the CPU, so I can't ask them to test it in low activity (they never tested Edge/Crome/MS word/Excel where sleep states would be allowed).

My monies on the CPU as the type of crashes you're getting are the sort which happen when overclocking with curve optimiser and using to much boost + so even though you're only running stock settings I'm guessing it's right on the edge of stability.

Personally I'd RMA the CPU directly to AMD.

Techy_Ben · 21 May 2021 at 18:55

Joxeon said:
My monies on the CPU as the type of crashes you're getting are the sort which happen when overclocking with curve optimiser and using to much boost + so even though you're only running stock settings I'm guessing it's right on the edge of stability.

Personally I'd RMA the CPU directly to AMD.

Thanks, that's kinda what I'm thinking too. But no idea how to rule out the VRMs or something just being dud on the motherboard. I might get a newer PSU in, but again, that's even more cost on replacing parts that work fine with an intel system when tested (GPU/psu/ram/ssds etc).

No idea though, if it's boost causing the crashes. Can I limit the CPU to a low clock state and see if chrome (easiest software to test) is stable then?

As said, overclockers sent it back as "working". So I'm reluctant to do another RMA on it. At this point, I'm tempted to send both back, and get a 550 board and a 5800x... But as they are "tested ok", how do I prove anything.

Joxeon · 21 May 2021 at 19:54

Techy_Ben said:
Thanks, that's kinda what I'm thinking too. But no idea how to rule out the VRMs or something just being dud on the motherboard. I might get a newer PSU in, but again, that's even more cost on replacing parts that work fine with an intel system when tested (GPU/psu/ram/ssds etc).

No idea though, if it's boost causing the crashes. Can I limit the CPU to a low clock state and see if chrome (easiest software to test) is stable then?

As said, overclockers sent it back as "working". So I'm reluctant to do another RMA on it. At this point, I'm tempted to send both back, and get a 550 board and a 5800x... But as they are "tested ok", how do I prove anything.

Because of the nature of the fault you won't pick it up in normal stress tests so I'm guessing if you send both back to OCUK they will just run some benchmark tests on it then say it's fine and just return it to you again that's why I recomend sending it direct to AMD as they will likey just replace it and then you will know for sure.

Techy_Ben · 21 May 2021 at 20:11

Oh well. I might give it a try then. :/

AStaley · 22 May 2021 at 09:07

Grim5 said:
that's decent temps if it's overclocked, if it's stock that's very hot

CPU is running stock, but using 3800Mhz/CL14 RAM (1.5v). Think before I changed the RAM I used to see max temp around mid-70’s. Otherwise think my temps are about right, idle is current mid-30”s to low 40’s with heating on in the house.