EXPO ram suddenly unstable ?

[Dicehunter]

Running a 7800X3D in an X670E motherboard, Latest bios with 2 x 16GB of Corsair Vengeance 6000 CL30 EXPO memory, Windows 11 is fully up to date, Motherboard and GPU drivers up to date.

It's been running rock solid for the last 2+ months and today out of nowhere I started getting system restarts during gaming, No blue screen, No warnings, Just like you pressed the reset button.

I read around and most fingers are pointing at ram, To test this I set it to the DDR5 base spec of 4800 and the system is rock solid once again but if I go over 5200 then any game activity just sees an instant reboot.

What I find weird is that at its rated speed of 6000 CL30 it can pass every single memory stability test there is, Memtest, OCCT, AIDA etc... but the moment I fire up The Division 2, As an example, And walk around, Instant reboot. I tried taking out my GPU and using the onboard GPU just to be safe, Same thing happens.

Nothing in my system has changed over the last 2+ months so I'm at a loss of what it could be unless ram can degrade running at its rated speed.

Anyone encountered this before ?

 

[Dicehunter]

Are you using EXPO or just setting a manual frequency? Manual frequencies require more configuration with DDR5 than what DDR4 systems did.

Yep, Set EXPO, Then lower the ram speed until the system is stable, 5200 is the limit, Any more and it's unstable.

I would expect ram instability result in game crashes, error messages, blue screens

Sudden reboot on AMD is a sign of CPU instability. Sort of you get if going too low in curve optimizer.

I disabled curve optimizer as I had a negative offset of 30 but ran into instability issues so returned it to default.

 

[Dicehunter]

should also try defaults on infinity fabric speed, SoC voltages and such.

and there were cases, at least on Zen 3 where CPUs were unstable on stock clocks and sometimes different LLC levels helped it.

Yep tried setting defaults, Although I'll look at the LLC next.

Maybes reset bios fully back to all defaults and go from there?

Already reset everything to default, Same thing happens, Any more than 5200 and the system does instant reboots in game.

How old is the PSU?

About a year, It's a Corsair ShiftX 1200w, I also removed any cable extensions and tried with the stock cables, Same behaviour.

 

[Dicehunter]

This isn't a case of EXPO SoC voltage being too high leading to degradation is it?

EXPO sets it to a max of 1.25 which as far as I've read is in the safe zone as AMD has stated the max is 1.3.

Maybe the RAM is getting hot, point a fan at it. I would also run sfc /scannow and also validate the game files. Ubisoft games do tend to hit RAM hard. Ghost Recon Breakpoint would overhead my AM4 DDR4, nothing else did.

Ram according to Hwinfo gets to a max of 45'c so temp shouldn't be an issue. I've done sfc /scannow and it came back clean, Same with game files.

Lowering clocks on the ram from 6000 to 5200, Which is apparently the max "official" supported speed by AM5 according to AMD, Makes the system rock solid stable so I don't know, Maybe my 7800X3D's IMC is poo, Maybe the ram is funky as it has both EXPO and XMP profiles, Maybe that causes some weird issues.

 

[Dicehunter]

This is why I don't bother tweaking RAM, it just too much hassle. I got a RAM cooler with a DDR4 kit years ago, I installed it yesterday as its hot in my room, RAM stays ~35C now.

This isn't about tweaking ram though, Just selected its profile in the bios and after a few months it became unstable for reasons unknown to me.

 

[Dicehunter]

I think I may have figured it out, Touch wood. In the end no matter what I changed my ram speed to it kept on rebooting randomly.

I had a 2TB HDD running as a basic storage drive for documents, Family photos, Music etc... and it was metal on metal with the case which I know "can" cause shorts if the wrong area is touched, Mainly the circuit board, And the exact type of behaviour I was seeing with the instant reboots and no error screen are in line with a quick short.

I must have forgotten to replace the rubber pad I cut underneath it last time I did a cleaning of the case.

Gonna do more testing but I'm hoping that was it, So far done an hour of testing and no issues at the memory's rated speeds and the system is fully stable.

 

[Dicehunter]

Could be that the warmer than usual temperatures have made something unstable, I had to adjust my ram timings and drop voltage a tad when I upgraded from a 1070ti to 3080 due to the excess heat the card generated.

My ram settings had previously been rock solid for over a year and would pass any memory stress test. The crashes would only in games when the gpu was loaded so it took me a while to pin down the issue.

Temp in the room my PC is in is fairly warm but it's not the heat. The temp was the same when it was unstable and then when I removed the shorting HDD to where it's now stable.

Lesson learned, Do not place conducting PCB's on metal surfaces

 

[Dicehunter]

Since turn off your negative offset is it stable now?

Most likely the ram is stable but your -30 offset on the cpu wasn't.

What you should do is test one thing at a time. Put the CPU to stock for now but just enable the expo ram. Use it for a few weeks and see how things go.

If it's stable then you can try a -25 CPU offset, then -20, -15 etc untill the crashing stops etc...

It wasn't the CPU or ram in the end, It was a shorting component triggering the PSU which caused a reboot. Component removed and now all good