• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Resolved by Bios Update: 9800X3D/Asus 870E Hero Powering off Under Idle (Kernal 41 Error)

B

Blackstone

Supreme [H]ardness
2FA
Joined
Mar 8, 2007
Messages
4,098
Recommend updating to latest bios if on Asus 870E Crossfire Hero.
 
Last edited:
Blackstone said:
Bottom - line, system is rock solid in all stress tests including Prime, AIDA, Furmark, Games, Cinebench, ect. I do not get issues during load. Temps are good, performance is great, and I do not get crashes in applications.

When systems sits idle, about once a day, or once every other day, I find the system just shut itself down. I power back on, and in event viewer I have Kernal Error 41. Unexepcted shutdown.

I repasted CPU.
I reseated ram.
Flashed Bos and Updated to Latest Mobo Firmware
I replaced all power cables.
I replaced 850 Watt Seasonic Prime with 1300 Watt Seasonic Prime PSU
I tried to shut down all bloatware Asus installed. I accidently instaled armory crate. I shut those processes down.
Reseated GPU
Added additional 120 MM fan to CPU cooler (Noctua D15 G2)
I tried reverting to stock and removing OC, but this issue was happening before I ever started undervolting and it happens at stock speeds. I have +200 in PBO overdrive, Scalar at 1X, -20 curve, but it happens even with no undervolting at all.

I have just now increased minimum power state for CPU from 0 to 100 percent. I am thinking it is related to power state. But if I find that this fixes it, what can I do about it then? Just leave it at 100 percent so it never downclocks?

Can anyone help?

9800X3D
ASUS ROG CROSSHAIR 870E HERO
64 GB DDR 5 6000 (Corsair Dominance Titanium)
RTX 3090
Seasonic Prime TX-1300 PSU

Thanks,
Click to expand...
I had this issue with my old Intel 9980XE. Stable at 4.6GHz but unstable at idle. I had to end up setting a manual voltage because nothing I could do get the vDroop up high enough at IDLE when trying to set the automatic voltage with an offset, and thus it would droop too low at idle and crash. So I ran it with a manual voltage for 6 years.

In my new ASRock x870E Taichi, there is a BIOS setting that gives me an offset for voltages at idle. Does your BIOS have anything similar? I know it's a different board manufacturer, but I'm not sure if Asus provides it, probably in a different place. See here for more info. Pay close attention to the last chapter, Curve Shaper, of the video.


View: https://www.youtube.com/watch?v=3hNgEgmLJ8Q
 
If you have a power supply other than seasonic lying around try it, just to idle test. Any size will do.

Also, you didn't mention trying a clean install.
 
Blackstone said:
Yes and I have already dabbled with curve shaper. Basically, if I understand curve shaper correctly, I do like a -20 curve optimizer across the board and then go into curve shaper and do a positive offset for light work loads and idle. I already tried that a bit, but the positive voltages i applied were very modest. I did not negate the entire -20 curve optimzer for light work loads. I reapplied like 5 in shaper, but maybe I need to negate the undervolt completely for low frequency/idle.

I'm thinking if setting power state minimum to 100% in windows fixes the problem, that indicates I need more voltage at low frequencies and maybe if I dial it in right I can do away with 100% power state.

But why I am dealing with this at all on a $700 motherboard is beyond me.
Click to expand...

Blackstone said:
I did not clean reinstall yet. Testing with another PSU beyond the two seasonics is not in the cards. I am pretty confident it is not PSU.
Click to expand...
In that case, I would highly recommend you return that Asus board and go with ASRock. I did that this time around, and the entire experience has been absolutely flawless. No issues, and runs cool to boot. Asus just isn't what it used to be....

Also forgot to ask about your RAM. Are you running EXPO settings? What are the timings?
 
Blackstone said:
Expo 2 but dropped CAS from 30 to 28. Again, stable in every single stress test and game. Just not at idle for extended periods.

By the way no change no matter what I do with memory.
Click to expand...
Quite possible you need to RMA the mobo or the CPU. Shouldn't be doing this with everything on stock and no OC whatsoever.
 
IMC on the CPU is bad or the motherboard is bad. I'd RMA both and replace the board with an ASRock x870E Taichi.
 
If you haven't, install the latest AMD chipset drivers, from AMD.com

What is your system agent voltage? 64GB is more taxing on the IMC. It should probably be at least 1.25v. It may need to be higher than that. But, it should never be higher than 1.3v. If its higher than that, turn it down. Higher than 1.3 legitimately risks damage to the CPU. And it can actually make things less stable.

Turn off your Curve Optimizer/undervolt, until you figure out your idle issues. Its only complicating things.

IMO and IME, a clean install would be best, as well.
 
Last edited:
Blackstone said:
My ram is not on the QVL for this motherboard. Is that a problem? But it is an expo kit.
Click to expand...
That's entirely board specific. Even though the 800 series boards are the same silicon----memory compatibility seems worse, for now. It will get better as more BIOS updates release.
 
Blackstone said:
My ram is not on the QVL for this motherboard. Is that a problem? But it is an expo kit.
Click to expand...
While I can't 100% guarantee it's not the memory until you swap it out for another set to test, it's highly unlikely not being on the QVL is causing this problem.
 
Blackstone said:
There has to be more I can do to test before I RMA. I will certainly clean install.

I’m running memtestc86.

I’m going to see if not allowing CPU to downclock helps. Maybe that will narrow my field of search.

Maddening.
Click to expand...
What is the SA voltage?
 
Dude, I applaud you for trying to troubleshoot, I really do. But you absolutely should not be having these problems at stock, period. To save yourself the pain, I would just return the board and the CPU, and while you're at it, switch out the board for ASRock. Asus has lost their way. If you want something that won't screw with your CPU at stock, and do exactly what you tell it to, get ASRock. There's a high probability it's just the board that's acting wonky, because Asus doesn't know how to write a BIOS anymore.
 
I wouldn't buy Asus at this point, but running C28 on 6000 C30 RAM isn't stock? I'd seriously try full stock, no Expo. If that is stable, then try the Expo setting (maybe the other profile). THEN, if that is all stable, work at getting C28 stable on your kit.
 
Blackstone said:
It is unstable at C30 too.
Click to expand...

Try setting your VDDP manually at 0.950 to 1.000V range, just for curiosity's sake. I remember having mega problems with Asus boards when AM5 launched and it kept cranking this up way too high (like 1.6-1.8V), AMD default is 0.950V. I went around in circles and RMA'd a bunch of parts just to come back to this. Later, a BIOS update did eventually fix it.
 
When troubleshooting, remove everything you don't need. Use just one stick of RAM (single channel), use a cheap low-wattage graphics card, reset BIOS, and the only other thing you should have is a drive for a fresh OS install, a mouse, keyboard, and ethernet. Yeah, it sucks, but better to do it now, than later.

Once in your OS, see if you get the idle crash. Either way, swap the RAM sticks and see if the result is different. If you have a known good RAM stick you're using on another computer, test it with that, too.

Once you rule out the memory, it's either the CPU or motherboard. 99% of the time it's the motherboard, or the BIOS and/or settings.

Sometimes, it can be a corrupt OS install. I assume you're using Windows. If you don't want a fresh OS install, boot into Safe Mode. If you don't crash on idle, then it's a driver or software causing the crash.
 
Looking on the net, could be BIOS, motherboard (RMA), Vanguard anti-cheat software, Windows power profile, etc.
 
Blackstone said:
Passed memtest.

Do you mean SoC voltage?
View attachment 702198
Click to expand...
Oops, yes. SOC voltage.

I would turn off your curve optimizer undervolt. Put SOC voltage at 1.29. And see what happens.

If it's stable, I would slowly lower SOC voltage, until you no longer have stability at idle/low load.

Once you find stability, then try curve optimizer.
And it may also endnup that your CPU can't do even -20 all core. I've seen a few which had to do more like -15, or resort to tuning the offset for Individual cores.
 
Blackstone said:
Well if setting power minimum performance state to 100% in windows fixes it, what do i make of that?

I’m thinking RMA.
Click to expand...
Had this happen when I upgraded from my E5-1680V2 to 7800X3D without a clean install. Would crash after a few hours at idle without performance state set at 100% no matter what overclock and RAM setting I chose.

Clean install, now I can leave minimum performance state at 5%, PBO at -20, EXPO RAM settings, and have not had a single CPU-related crash.
 
This sounds an awful lot like the old "Typical Current Idle" thing that early Ryzens suffered from; I think the label was usually Power Supply Idle Control, and if you didn't have it set to Typical Current Idle, you could see weird crashes when idling. But surely that's long gone? Do you even have the setting any more? (I forget where it was, probably AMD CBS or maybe PBS.)
 
Idle crashes are typical with undervolt, is curve optimizer supposed impact it at idle as well as load? perhaps the system is applying it erroneously at idle.
 
chameleoneel said:
Oops, yes. SOC voltage.

I would turn off your curve optimizer undervolt. Put SOC voltage at 1.29. And see what happens.

If it's stable, I would slowly lower SOC voltage, until you no longer have stability at idle/low load.

Once you find stability, then try curve optimizer.
And it may also endnup that your CPU can't do even -20 all core. I've seen a few which had to do more like -15, or resort to tuning the offset for Individual cores.
Click to expand...

1.29 is a shitload of SOC voltage. Somehow after that whole frying chips thing people got stuck in the mindset that you put it to 1.30V and call it quits, but most Ryzen chips are more stable and OC better with it around 1.100-1.150V. There are technical reasons for this that I can't remember off the top of my head but it has to do with the interaction between the IOD and the fabric.
 
LigTasm said:
1.29 is a shitload of SOC voltage. Somehow after that whole frying chips thing people got stuck in the mindset that you put it to 1.30V and call it quits, but most Ryzen chips are more stable and OC better with it around 1.100-1.150V. There are technical reasons for this that I can't remember off the top of my head but it has to do with the interaction between the IOD and the fabric.
Click to expand...
It all depends on what you are trying to do. In this case, the OP is trying to run 64GB of RAM, which is more difficult to do with XMP/EXPO.

On my own 7800X3D build, I needed 1.25 for stable 32GB DDR5 6200 with low main timings like CL30 and lots of very tweaked sub-timings. Or DDR5 6000 with CL28 and tweaked sub-timings, etc. -20 for curve optimizer.


Anyway, looks like the OP did a clean Windows install, which will get rid of any hard to chase problems. Which can sometimes look like hardware issues.

And I also suggested 1.29 temporarily, to see if the SOC voltage is the issue or not. And to slowly lower it, to find the correct point of stability.
 
Last edited:
Blackstone said:
Expo 2 but dropped CAS from 30 to 28. Again, stable in every single stress test and game. Just not at idle for extended periods.

By the way no change no matter what I do with memory.
Click to expand...
It's an issue with the ASUS motherboards. There is a tweaking section where you can adjust the power delivery up to extreme/ 200% and you can try that and see what happens. Just up it 1 step more aggressive and see what happens. I had to do the same thing but I can't remember the exact menu cuz I'm not home right now.
 
Geezer000 said:
It's an issue with the ASUS motherboards. There is a tweaking section where you can adjust the power delivery up to extreme/ 200% and you can try that and see what happens. Just up it 1 step more aggressive and see what happens. I had to do the same thing but I can't remember the exact menu cuz I'm not home right now.
Click to expand...
Yep. Like I said previously, I had this exact issue on my old Asus based system, and could never get it stable.

Went ASRock this time and can't believe things were this easy. It just....WORKS. Like Asus of the old (circa 2010). I won't be going back to Asus, that's for sure.
 
chameleoneel said:
It all depends on what you are trying to do. In this case, the OP is trying to run 64GB of RAM, which is more difficult to do with XMP/EXPO.

On my own 7800X3D build, I needed 1.25 for stable 32GB DDR5 6200 with low main timings like CL30 and lots of very tweaked sub-timings. Or DDR5 6000 with CL28 and tweaked sub-timings, etc. -20 for curve optimizer.


Anyway, looks like the OP did a clean Windows install, which will get rid of any hard to chase problems. Which can sometimes look like hardware issues.

And I also suggested 1.29 temporarily, to see if the SOC voltage is the issue or not. And to slowly lower it, to find the correct point of stability.
Click to expand...
Yeah, I used higher SOC voltage on my 7800X3D just because it defaulted to 1.3 and I had lowered it to 1.28.

With my 9800X3D and the new BIOS's, it defaults to 1.1V (same board) with my 96GB of 6400 C32 RAM, and works just fine at 6000 C30 with Buildzoid's every Hynix A/M die kit can do this secondaries.

Not sure if it's just the newer BIOS that caused the much lower stock SOC voltage, or if it was also the fact that I'd upgraded to a 9800X3D.
 
Blackstone said:
I went Asus precisely because I have been using them for years and years and have had so few problems. Oh well.

It does seem like system is running better after the clean reinstall. I was actually getting artifacting/flickering in windows before. The whole system just seems faster and more stable. Either the issue will pop back up or it won't.
Click to expand...
Best thing to do after a Windows install with a SSD is remove 10% from the main partition. Then run CMD in Admin mode, chkdsk /f /r, reboot, and finally optimize the drive in Windows Defrag (runs TRIM), and have it run daily. After every major Windows Update, run chkdsk again.

Glad it's stable now!
 
Let us know how you solve it. I am interested in the resolution. FOR SCIENCE!
 
Blackstone said:
Overclocking this chip just seems unecessary as the system is just SO fast at stock.
Click to expand...
i thought that with the x3d chips it was "enable pbo and call it a day" as the oc was finicky or pointless.

at least youre back up and runnin
 
in my experience random kernal 41s are usually not enough juice to the ram, so maybe the expo didnt apply right, idk. waiting does seem like a good plan though...
 
I hope that fixed it but I kinda have my doubts. Though if you had the ol' Asus bloat on there that definitely could have been the culprit.
 
Bloat or leftover old drivers not completely cleaned out from the system. Like I said, I would get random crashes until I did a clean install; no amount of clearing old drivers and uninstalling old software solved it. ASRock here, all ASRock software reinstalled with no issues whatsoever. I'm fairly sure it's just a Windows bug.... I mean feature.
 
For overclock I would say it is indeed pointless unless you run Cinebench daily and pat yourself on the back. I will however, suggest some tighter timings on the ram and setting PBO to -15 all core and calling it. Mine is set at -20 as anything over has not been stable for gaming for me. I did try everything and I am not sure people play the variety of games that I play to check for gaming stability using lower PBO settings (I run 15-20 games to test some of which have shown me to be unstable with lower PBO in the past on all my AMD builds).
 
Blackstone said:
Still stable. I feel like the reinstall did the trick. Will know for sure in a few more days. I am more and more thinking either it was software or my undervolt corrupted the OS. My event viewer certainly is a lot cleaner now.

Now, bring on the 5090!
Click to expand...
Yeah, my brother-in-law's system I built for him, and my main had similar issues. All the other systems I built haven't had any issues though. A fresh Windows install fixed the problems. The ONLY thing in common with our two systems is that they were cloned Windows installs with Universal Restore. I'm thinking that was the main culprit. Fresh Windows installs for me from here on out.
 
Blackstone said:
It is incredible how fast you can get everything back online after a clean reinstall with solid state drives and a gig speed internet connection. I used to do it as part of a regular maintenance routine on an annual basis but as I got more busy I started to slack. The system really does like it though!
Click to expand...
Even faster when you just boot after a platform/cpu swap and it still works. ;)
 
just remeber to remove the chipset drivers before swapping the cpu.
 
let it idle and set it to blank out the monitor. without rereading... placed with llc yet?
 
You must log in or register to reply here.
Back
Top