Ryzen 5000 crashes - anyone getting them?

Retro89 · 22 Feb 2021 at 20:06

Retro89 said:
Built my new system yesterday and had a couple of random reboots when i wasn't doing anything.

Since disabling PBO i've not had any. (knock on wood)

Jinxed it
"A fatal hardware error has occurred.
Reported by component: Processor Core
Error Source: Machine Check Exception
Error Type: Cache Hierarchy Error
Processor APIC ID: 6"

Cyber-Mav · 22 Feb 2021 at 22:40

Wont be surprised amd do same thing they did with first gen ryzen cpu compiler bug. Ignore the issue and just issue out rmas.
They never came clean about what caused the issue

Joxeon · 22 Feb 2021 at 23:32

Retro89 said:
Jinxed it
"A fatal hardware error has occurred.
Reported by component: Processor Core
Error Source: Machine Check Exception
Error Type: Cache Hierarchy Error
Processor APIC ID: 6"

That usually occurs when the CPU isn't getting enough voltage.

Try either updating the bios or adding more voltage to core 4 with curve optimiser.

ZeroG · 22 Feb 2021 at 23:55

I recently upgraded my mobo and cpu to an Asus ROG Strix B550-F mobo and an R5 5600x and it's been working fine for the last month approx. However today I was playing World of Tanks and my pc suddenly restarted. The event log shows the following:-

A fatal hardware error has occurred.

Reported by component: Processor Core
Error Source: Machine Check Exception
Error Type: Cache Hierarchy Error
Processor APIC ID: 9

The details view of this entry contains further information.

On Googling this I see it crops up quite a lot however there doesn't seem to be a definitive fix. I have done no overclocking at all, everything is running at stock. So far it's just the once but reading online I might expect it to happen again?

Joxeon · 23 Feb 2021 at 00:37

This error is common when using to much negative curve optimiser and seems to happen more at idle/light loads, if it's happening at stock then it could be down to the bios not delivering the correct voltage which is why it's worth updating the bios or it could be dodgy CPU, you might able to fix it by increase the voltage slightly but it's not an ideal solution.

The APIC ID lets you know which core caused the crash so for ID 9 it would be core 5 as the cores show in pairs counting SMT starting from 0/1 core 1, 2/3 core 2 etc so you could try to add a slight increase of voltage to that core through curve optimiser but the first thing I would try is updating the bios to an agesa 1.2.0.0 if your not already on it.

Jimmy Weirdarms · 23 Feb 2021 at 01:34

Had to RMA my 5950x today.

Been running it stock, no tweaks. 4x16GB RAM in a system that has been rock solid for a year and a half with a 3800x in it. It ran fine for about six weeks then out of nowhere I started getting hard lockups several times a day (under linux), and occasional boot failures with the RAM debug light on. Tried all the DIMMs individually with memtest86 and they worked great, tried one DIMM across all slots and got intermittent boot failures on one slot and memtest86 errors on another. Weirdly only on the "Random pattern" test. Tried a 3700x and the test passed.

I figure something's gone wrong with the memory controller on the 5950x, somehow. Wishing I hadn't sold my old 3800 in january now...

Retro89 · 23 Feb 2021 at 01:55

I'm at stock settings with XMP enabled, all drivers and bios updated to latest. The crashes have happened when at idle or just using Firefox.

Think i'm going to send it back and get a refund. I don't want to keep having to tinker with settings. I'll go back to my 4790K then pick a CPU later on in the year, hopefully all the bugs have been ironed out by then.

OverDrive · 23 Feb 2021 at 17:25

My 5800X has been completely stable. I’ve not had any crashes at all. I have 32GB (2x16) of 3600 RAM with XMP enabled and I have also enabled smart memory access for my Radeon 6800 GPU. I’m running the latest chipset drivers from AMD.

EDIT: Running at stock.

mattup · 23 Feb 2021 at 19:17

My 5800X is running at 4.85Ghz with 32Gbs of ram clocked to 3600Mhz CL14. Not one crash.

More Stable tan any other CPU I have had over the last 35 years I would say.

Don't get carried away with the current DOA rumour.

Matt

Richdog · 23 Feb 2021 at 20:47

mattup said:
My 5800X is running at 4.85Ghz with 32Gbs of ram clocked to 3600Mhz CL14. Not one crash.

More Stable tan any other CPU I have had over the last 35 years I would say.

Don't get carried away with the current DOA rumour.

Matt

Do you mean running at stock or an all core overclock?

mattup · 23 Feb 2021 at 21:15

All core PBO -25

All cores get to 4.85 whilst gaming and 4.750-4.775 during CB20 @ 100% load.

Stunning chip.

Matt

danq · 23 Feb 2021 at 21:24

I've managed to push my 5800X to 4.8Ghz.. without any 'fine tuning'. Working great

ollies · 23 Feb 2021 at 22:10

Something I've seen mentioned elsewhere is that when you set negative curve and then choose to return to stock or PBO auto then you need to set all the curve values back to 0, otherwise these values can persist and continue to cause issues. If you're getting WHEA when playing with curve then check the APIC ID and adjust curve for that specific core (keep in mind that APIC ID refers to thread number, so #0 and #1 is core 1 in curve optimiser).

luismenendez · 24 Feb 2021 at 21:33

Jimmy Weirdarms said:
Had to RMA my 5950x today.

Been running it stock, no tweaks. 4x16GB RAM in a system that has been rock solid for a year and a half with a 3800x in it. It ran fine for about six weeks then out of nowhere I started getting hard lockups several times a day (under linux), and occasional boot failures with the RAM debug light on. Tried all the DIMMs individually with memtest86 and they worked great, tried one DIMM across all slots and got intermittent boot failures on one slot and memtest86 errors on another. Weirdly only on the "Random pattern" test. Tried a 3700x and the test passed.

I figure something's gone wrong with the memory controller on the 5950x, somehow. Wishing I hadn't sold my old 3800 in january now...

I also believe my 5800x has a buggy memory controller also (apart from running very hot!). Memory is Samsung Bdie I had run at 3800Mhz before even with this ryzen at the beginning. New bios and had to set the memory at 3733. Starting to get random crashes, sometimes just surfing. Tried 3600 and also random crashes. To troubleshoot, I just loaded default bios and ram set to XMP of 3200 yesterday. No crashes so far but running only at 3200 with CL16 with b-dies

Using the latest bios in case someone suggests updating it.

Jimmy Weirdarms · 24 Feb 2021 at 22:34

luismenendez said:
I also believe my 5800x has a buggy memory controller

I would be interested to know what happened with memtest86 on your system, particularly the "random pattern" test.

But yeah, sucks

luismenendez · 24 Feb 2021 at 22:51

Jimmy Weirdarms said:
I would be interested to know what happened with memtest86 on your system, particularly the "random pattern" test.

But yeah, sucks

I run it when I tweaking the memory with Dram calculator for Ryzen usually for an hour or 90 minutes. When I get errors, I just relax timings until I don't get errors any more. However, passing memorytest didn't stop the processor from crashing. In the beginning, I thought it was the graphics card that I bought used in January as a few times I just got out of the game with a Radeon driver crash message so that's why I didn't think in returning the 5800x as I was blaming the newish 5700!

barrost · 3 Mar 2021 at 18:08

Hey everyone, I'm back. After requesting a long list of tests, AMD approved my RMA last week or so. I have received my 5950x back today, installed and I'm stable with/without XMP, which makes me super happy.

My WHEA 18 random reboots were fixed by replacing the 5950x.

humbug · 3 Mar 2021 at 18:39

barrost said:
Hey everyone, I'm back. After requesting a long list of tests, AMD approved my RMA last week or so. I have received my 5950x back today, installed and I'm stable with/without XMP, which makes me super happy.

My WHEA 18 random reboots were fixed by replacing the 5950x.

Good Show

philo-sofa · 3 Mar 2021 at 19:00

luismenendez said:
I also believe my 5800x has a buggy memory controller also (apart from running very hot!). Memory is Samsung Bdie I had run at 3800Mhz before even with this ryzen at the beginning. New bios and had to set the memory at 3733. Starting to get random crashes, sometimes just surfing. Tried 3600 and also random crashes. To troubleshoot, I just loaded default bios and ram set to XMP of 3200 yesterday. No crashes so far but running only at 3200 with CL16 with b-dies

Using the latest bios in case someone suggests updating it.

Intriguing. My 5800X was also running some Samsung B-die, four slots, at 3800 MHz. Failed to boot one day after two weeks of running perfectly stably and eventually discovered that removing any one DIMM from any slot fixed things.

I changed out all four DIMMs and found the same problem; finally changed to another set and found it worked perfectly again, but had to settle for 3733 like you.

Does seem the memory controllers are a bit of a weak-point for us. Thankfully performance is still awesome-sauce.

luismenendez · 3 Mar 2021 at 19:35

Just updating my post above regarding the 4x8GB memory speed problems. I swap for a 16Gb TG 8 pack kit (2x8gb) and now running fine at 3800Mhz 16-16-16 without any problems at all.

Good to hear barrost resolved his problems with a new processor.