cancel
Showing results for 
Search instead for 
Did you mean: 

Prime X670E pro BSOD issues

DaveG587
Level 9

I have built new system with the following:

  • Prime X670E Pro Wifi  (BIOS 1813)
  • AMD Ryzen 7950X3D
  • CORSAIR VENGEANCE RGB DDR5 RAM 32GB (2x16GB) 6000MHz CL30 AMD EXPO iCUE Compatible Computer Memory - Grey (CMH32GX5M2B6000Z30K)
  • Samsung 990 PRO 2TB PCIe 4.0 (up to 7450 MB/s) NVMe M.2 (2280) Internal Solid State Drive (SSD) (MZ-V9P2T0BW)

Things I've already done....
Updated all drivers, firmware, chipset etc for all of the above. to the latest.  Clean install of Win11 Pro
Just a few stress test utilities (OCCT, Adia64, Heavyload) and all complete with no issues.

The Issue....
If I do not enable EXPO, system runs fine.  If I enable EXPR 1 or 2, then I get BSOD and instant reboot.
Can't catch the BSOD to gain the information displays.

Does anyone else have this motherboard and is using EXPO okay, or had the same problem and resolved it, if so how?

Thanks

1,906 Views
1 ACCEPTED SOLUTION

Accepted Solutions

DaveG587
Level 9

Just to end this thread.
Replacement CPU has resolved the issue.

View solution in original post

29 REPLIES 29

Silent_Scone
Super Moderator

Hi @DaveG587,

Check to see the applied VSOC voltage when EXPO is enabled. Select EXPO II and raise VSOC in 20mv increments. The maximum allowed voltage is 1.3v.

Contrary to assumptions, XMP/EXPO doesn't ensure automatic overclocking. Tighter timings or higher frequencies may require manual adjustments for stability. Overclocking cannot be assured due to variance between parts when run outside manufacture specifications. XMP/EXPO success depends on the processor's Integrated Memory Controller (IMC) capabilities. Like CPU speed, memory overclocking varies due to variances between parts that exist when run outside of the manufacturer's specifications.

 

13900KS / 8000 CAS36 / ROG APEX Z790 / ROG TUF RTX 4090

Hi @Silent_Scone 
Thanks for the reply, and not being a techie in todays BIOS's I'll have a look at see.

I'm not sure I agree with "run outside manufacture specifications" as the company "sells" the memory to work at the OC speed, and the Board (Asus) says that memory (at that speed) on the board which it supports both the memory & CPU. 
As a consumer, that is all we have to work with, you have to take them at their word, I agree if you step outside simply turning the feature on or off, then that is a different story.  But turning on or off on a board that is suppose to be compatible with both memory & chip should be expected.

Overclocking is always outside the manufacturer's specifications, that's why it's considered overclocking. Due to variance between parts, nobody can make assurances for such things. This is all part and parcel of overclocking. EXPO is overclocking, thus we have to take into account CPU variance. The QVL offers a list of kits that have been validated to give users the best possible chance of success. Moreover, due to CPU variances the auto rules can only accommodate so much, and adjustment of relevant voltages can be necessary. Experiment with VSOC and see if this aids stability.

13900KS / 8000 CAS36 / ROG APEX Z790 / ROG TUF RTX 4090

Okay @Silent_Scone  I hear you and see the logic in what you've stated, not 100% in agreement, as if they sell something that is rated at CL30 6000 then it should work at that, unless your saying that all the AMD CPU's are only rated for 4800 !  

Anyway, putting all that aside, is it:  CPU SOC Voltage  I'm looking for as I also have 
VDDSOC Voltage Override
VDDSOC Offset Mode Sign
VDDSOC Voltage Offset
and more....

But using OCCT, I can see the CPU VDDCR_SOC Volt... is sitting at 1.240 with EXPO 1 on at the moment.

Silent_Scone
Super Moderator

Hi,
The processor officially supports up to 5200 in 1 DIMM per Channel configuration as per the AMD product page. Anything above this can be considered an overclock.

If the VSOC auto rules are applying 1.24v you can adjust the voltage up to 1.3v using the override function, however, the CPU may be more comfortable at a lower voltage than 1.24v. The rules try to cater for the worst-case scenario, some CPUs may be more comfortable with less voltage. You can adjust in 20mV increments and see if it aids stability.

Silent_Scone_0-1702911028216.png

 

13900KS / 8000 CAS36 / ROG APEX Z790 / ROG TUF RTX 4090

Okay, thanks - will try that later after work.

if anyone reading this thread has similar hardware, did you do this or change hardware, if changing hardware which is the likely candidate to change CPU !!

@Silent_Scone 
Thanks for your pointer on SoC

Have did what you said but still got a few BSOD.
What was strange is I got to 1.2800v and on checking both via HWHardware & HWInfo64 they both showed 1.280, but Aida64 showed the SoC at 1.328  which freaked me out a little, so reset defaults and then checked again the three stats. and again a difference, this time  HWINFO64 & HWHardware were showing 1.240 but  Aida64 was 1.280
So for now have just left everything as default.  (See images below)

Could not take image of this, but in the BIOS (Default - latest BIOS) it showed:
CPU SOC Voltage 1.320 AUTO
SOC Local Voltage Reading 1.281


Just for info, here is a list of the BSOD been getting, normally when system is not under any load, as all stress testing is fine.
System Service Exception
FLTMGR.sys
APC Index Mismatch
NTOSKRNL
Reference by pointer
Memory Controller
Unexpected kernal mode trap
Page fault un nonpaged area
Kernal security check failed

Thinking easiest thing to do is return the Board. CPU and Memory and change it for Intel kit !!

With 1.280 SoCWith 1.280 SoCWith Defaults LoadedWith Defaults Loaded

Your board is not equipped with die sense measurement so the motherboard's Super IO controller won't show as close to the real-time voltage. For the most accurate measurement inspect SVI3 in HWInfo, as this is the reading taken from the CPUs own internal telemetry.

You can try turning off memory context restore as suggested in a below post, however, this possibly may not help as it stops the system from retraining. If the applied settings are unstable it should be understandable why this wouldn't make sense. Be sure to check memory stability with TM5 or Karhu RAM Test when doing so

13900KS / 8000 CAS36 / ROG APEX Z790 / ROG TUF RTX 4090

Hi @Silent_Scone 
Cheers on the direction and info on HWinfo etc, will use that going forward.
As for the memory context restore and power down, this did not help as you said.  So I'm back on the hunt again.

Out of interest, since I have the option to return either or all of the new parts (Board, CPU and Memory) what would be your best guess as the to one with issue, or do all?
I ask and every stress, bench test I have done never produces any errors, the BSOD only happen when system is doing very little, which is driving me mad, I'd almost like in the tests for something to error out, at least I'd have a indication then.