Z590 with i9-11900K - Frequent freezes/blue screens - bug in Intel?

[sh10151]

Since Intel will do "advance replacement" and send me a new processor before I send back old one, I'm taking them up on the offer. Who knows if it'll help or what - it can't hurt I guess.

 

ASRock has no such advance replacement, and also will not guarantee it's a new board. So I'm not bothering with them for now.

[sh10151]

Got new processor today, installed it, and turned C-States back on, and set Windows power mode back to Balanced. We'll see if it freezes again.

 

Update: Going on like 20+ hours now with new processor, still no freezes. If I can go a week without a freeze, then I'm pretty sure this indicates there was something wrong with my old processor.

[Clausewitz]

On 9/6/2021 at 3:07 PM, iransofaraway said:

I don’t have a way to reproduce the freeze consistently. And it doesn’t happen when the system is loaded (the crashes have happened all when I was doing something mundane like Chrome or programming). They threw the same CPU and motherboard together and ran AIDA64 CPU stability test (stress FPU option) over a weekend. Said it had no issues. Which I can probably do the same test and not have it crash. 
 

Disabling c-states, running memory in XMP for days now and have not had a single crash. Really believe the problem lies there somehow.  But I have no way of knowing if it’s a bug with microcode on CPU, a problem with the chipset, or a defective motherboard or CPU. Hard to say, given others are having these same issues?

ASUS released a beta BIOS with a microcode update for all Maximus XIII boards: 

ROG MAXIMUS XIII Series 1102 BETA BIOS

1. Bug fix
2. Microcode Update

https://www.overclock.net/threads/overclocking-11700k-11900k-results-bins-and-discussion.1777365/post-28865081

I am not BSODing with c-states enabled. 

[Clausewitz]

On 9/8/2021 at 4:19 PM, iransofaraway said:

Got new processor today, installed it, and turned C-States back on, and set Windows power mode back to Balanced. We'll see if it freezes again.

 

Update: Going on like 20+ hours now with new processor, still no freezes. If I can go a week without a freeze, then I'm pretty sure this indicates there was something wrong with my old processor.

ABT or no ABT?

[sh10151]

29 minutes ago, Clausewitz said:

ABT or no ABT?

I am leaving ABT off for now since that’s what my BIOS defaults to. If I can confirm this is all stable with C-States enabled then I’ll give it a go. Have XMP and ReBAR on. 

[Clausewitz]

1 hour ago, iransofaraway said:

I am leaving ABT off for now since that’s what my BIOS defaults to. If I can confirm this is all stable with C-States enabled then I’ll give it a go. Have XMP and ReBAR on. 

Nice! I'm glad you were able to get a sample that works with C-states on (at least).

Please let us know if ABT works for you without issue!

[SnoopyCute98]

Following up to this thread, my 11700K will have stability issues (weird restart pattern) when 1.35/1.35 VCCIO_MEM/VCCSA.

 

Auto was 1.45V same with OP, then manually set 1.40V for both and will keep an eye out. Mobo was Asrock Z590 Steel Legend, ram 3600C17.

And yes, the stock thermal throttle point was 115C and I ran it once at around that temp, instantly turn off the benchmark as the old Hyper 212Evo cant cool it properly. 

[unclewebb]

28 minutes ago, SnoopyCute98 said:

the stock thermal throttle point was 115C

The Intel recommended thermal throttling temperature for an 11700K is 100°C. 

 

https://ark.intel.com/content/www/us/en/ark/products/212047/intel-core-i711700k-processor-16m-cache-up-to-5-00-ghz.html

 

Motherboards that set the thermal throttling temperature to 115°C at default settings are not setting this correctly. Intel specifically states that they cannot guarantee long term stability if you decide to operate your CPU beyond 100°C. Doing this just one time will probably not permanently damage your CPU but it might. 

 

ThrottleStop shows if a manufacturer is not setting the PROCHOT (processor hot) temperature correctly.

 

 

[sh10151]

Another update - since installing replacement processor I’ve still not had a freeze. Been multiple days at this point (since 9/8). Running the following config, BIOS defaults except:

- XMP Profile 1

- CPU PLL Spread Spectrum: Disabled (BIOS set this when enabling XMP)

- CPU TJ Max: 100

- Clever Access Memory: Enabled

 

- Windows power profile is Balanced

If this keeps staying stable, I’m going to turn on Intel ABT and see how it goes. 

[SnoopyCute98]

On 9/13/2021 at 12:32 AM, iransofaraway said:

Another update - since installing replacement processor I’ve still not had a freeze. Been multiple days at this point (since 9/8). Running the following config, BIOS defaults except:

- XMP Profile 1

- CPU PLL Spread Spectrum: Disabled (BIOS set this when enabling XMP)

- CPU TJ Max: 100

- Clever Access Memory: Enabled

 

- Windows power profile is Balanced

If this keeps staying stable, I’m going to turn on Intel ABT and see how it goes. 

 

May I know what's your VCCSA & VCCIO_MEM voltage at?

[sh10151]

5 hours ago, SnoopyCute98 said:

 

May I know what's your VCCSA & VCCIO_MEM voltage at?

VCCIO is 1.056V

VCCIO_MEM is 1.448V

VCCSA is 1.462V

[Clausewitz]

On 9/14/2021 at 11:44 AM, iransofaraway said:

VCCIO is 1.056V

VCCIO_MEM is 1.448V

VCCSA is 1.462V

Please try ABT.

[sh10151]

5 hours ago, Clausewitz said:

Please try ABT.

I’ve turned it on and we’ll see how it goes. 

[Clausewitz]

15 hours ago, iransofaraway said:

I’ve turned it on and we’ll see how it goes. 

Okay awesome!

[sh10151]

6 hours ago, Clausewitz said:

Okay awesome!

It worked for 21 hours lol - then I had a Windows blue screen with the watchdog timeout exception, which I've seen others have with ABT. So I think ABT still isn't quite stable, or my cooler isn't adequate. Turning it back off for now.

[Clausewitz]

43 minutes ago, iransofaraway said:

It worked for 21 hours lol - then I had a Windows blue screen with the watchdog timeout exception, which I've seen others have with ABT. So I think ABT still isn't quite stable, or my cooler isn't adequate. Turning it back off for now.

Well, I've tried several samples and some work with ABT on some don't. It's really odd. The components and BIOS are the same and one will run ABT the others will bsod. 

I don't know what to say.

[Mister Woof]

14 minutes ago, Clausewitz said:

Well, I've tried several samples and some work with ABT on some don't. It's really odd. The components and BIOS are the same and one will run ABT the others will bsod. 

I don't know what to say.

is ABT like PBO?

[Clausewitz]

17 hours ago, Mister Woof said:

is ABT like PBO?

Yes, it's a floating turbo. 

However, I have some samples that can operate ABT without BSOD while others BSOD.

[SnoopyCute98]

On 9/14/2021 at 11:44 PM, iransofaraway said:

VCCIO is 1.056V

VCCIO_MEM is 1.448V

VCCSA is 1.462V

Same as mine.. 

Got an email from Asrock Support, seems its fine for 3600MHz.

 

 

[SnoopyCute98]

20 hours ago, Clausewitz said:

Yes, it's a floating turbo. 

However, I have some samples that can operate ABT without BSOD while others BSOD.

ABT is only for i9 right?

[sh10151]

Welp, I got a CLOCK_WATCHDOG_TIMEOUT just now. I went a lot longer without issues, but yeah. I'm guessing this is the same C-State bullshit. Disabling C-States again. I can't deal with this.

[chrischevrier1]

On 9/24/2021 at 2:10 PM, iransofaraway said:

Welp, I got a CLOCK_WATCHDOG_TIMEOUT just now. I went a lot longer without issues, but yeah. I'm guessing this is the same C-State bullshit. Disabling C-States again. I can't deal with this.

Were you ever able to come to a resolution?!?! My friend is having the exact same issue. Seems like 11th gen is a S&$* show.... We tried switching RAM was gonna try CPU but you said it didn't work for you either. Maybe it's the motherboard?

[sh10151]

42 minutes ago, chrischevrier1 said:

Were you ever able to come to a resolution?!?! My friend is having the exact same issue. Seems like 11th gen is a S&$* show.... We tried switching RAM was gonna try CPU but you said it didn't work for you either. Maybe it's the motherboard?

Nope, I gave up and left C-STATES off entirely in BIOS (and I also run "High Performance" power profile in Windows), and haven't had any stability issues since.

 

I haven't seen any BIOS updates for my motherboard since 2021/8/12 - so I assume they don't think there's a problem or don't care, and/or Intel can't fix it.

[chrischevrier1]

1 hour ago, iransofaraway said:

Nope, I gave up and left C-STATES off entirely in BIOS (and I also run "High Performance" power profile in Windows), and haven't had any stability issues since.

 

I haven't seen any BIOS updates for my motherboard since 2021/8/12 - so I assume they don't think there's a problem or don't care, and/or Intel can't fix it.

Interesting. I will try all the steps you tried. Thank you!

[Falkentyne]

3 hours ago, iransofaraway said:

Nope, I gave up and left C-STATES off entirely in BIOS (and I also run "High Performance" power profile in Windows), and haven't had any stability issues since.

 

I haven't seen any BIOS updates for my motherboard since 2021/8/12 - so I assume they don't think there's a problem or don't care, and/or Intel can't fix it.

 

I went through some of your original screenshots and I found a few problems already.

I noticed that your "VCCST" is reporting 1.6v. This is a LN2 voltage setting and should NEVER be this high.

Also you have another setting called "CPU Standby 1" which is reporting 1.056v, which is the correct value.

I'm honestly not sure what's going on here.

There are two standby voltages.  One is "Standby (VCCST) and one is Standby Gated (VCCSTG).

Normally, the one that HWinfo reports as VCCST is the regular standby voltage.  Standby Gated (VCCSTG) is almost never reported by hwinfo unless the "Embedded Controller" field for your motherboard reveals this, and that's always in its own section "E.g. EC" or "Embedded Controller."

 

Anyway, both voltages should normally be set to 1.05v, which is also the same default voltage for "DMI voltage".

 

Another issue I saw is your VCC PLL.  I'm going to assume that this is "PLL Termination Voltage".   Usually Gigabyte motherboards have traditionally called this "VCC VTT", VTT meaning Termination, much like, for DDR4 memory overclocking, you may have heard of DDR VTT voltage", or DDR Termination Voltage", which is normally set to 1/2 of "Memory VDIMM" voltage.  But these voltages are often called different names depending on OEM (Asus, Gigabyte, Asrock, MSI etc). 

 

Anyway...

PLL Termination Voltage should ALWAYS be set to the SAME Value as Standby voltage, which is 1.05v.  This is yet another LN2 setting.  Sometimes, raising PLL Termination Voltage and CPU Standby (VCCST) voltage together a few ticks (e.g., from 1.05v to 1.20v) can help stabilize cache overclock a little bit, on air/water cooling, but this is never needed for stock operation or regular light overclocking.

 

What we found on both Z590 and Z690 motherboards (ESPECIALLY on Z690 motherboards), is that PLL Termination Voltage needs to be within a certain range of Standby voltage, or the system can completely hard lock (or even report "No CPU detected" or "Post code 00" error after a power off).  I think on Z590, PLL Term had to be within a certain amount of millivolts of Standby voltage, while on Z690, they need to be set to the same value.

 

I'm honestly hoping that hwinfo64 is reporting an incorrect value, as a PLL Termination Voltage of 1.20v, while not unsafe, is not standard at all.

 

The second problem is your VCCST.  That's reporting 1.60v and this is absolutely NOT correct and NOT safe.  However I know nothing about Asrock boards so I don't know what's going on here.  This is a LN2 setting (designed to prevent "cold bugs" when trying to run subzero cooling).

 

Are you using the very, very latest version of HWinfo64 from the hwinfo website?  If not, try using that and see if you still get a 1.60v reading.

Otherwise, if you can find these values in your BIOS (VCCST and VCCST Gated), you can change them both to 1.05v, and change VCC VTT / PLL Termination Voltage to 1.05v and see if that improves anything.