Instability with STRIX Z390-E GAMING+i9-9900K+Corsair Vengeance LPX 32GB

[Toffypops]

Hello all.
Writing here in hopes someone might be able to help.

I have had problems with stability even since the start with my new system.

I get random blue screens referring to IRQ errors, WHEH uncorrectable error, clock watchdog timeout.

My Specs:
Motherboard:
ASUS ROG STRIX Z390-E GAMING, S-1151
Processor: Intel Core i9-9900K
Memory (part number): Corsair Vengeance LPX DDR4 3200MHz 4x8GB CMK16GX4M2B3200C16
Graphics Card #1: POWERCOLOR Radeon HD 7990 6GB
Sound Card: Asus Xonar DX 7.1
Monitor: 3x ASUS VG278HE 144Hz
Storage #1: Intel 120GB ssd (system)
Storage #2: Samsung SSD 840 Pro
CPU Cooler: Noctua NH-D15
Power Supply: MIST 1000W
Keyboard: Microsoft Ergo 4000
Mouse: Logitech G402
Headset: Sennheiser PC 363D
Headset/Speakers: Logitech 5.1
OS: Windows 10 Pro 64BIT
Accessory #1: Samsung 970 EVO 500GB x2 M2.NVM

I ran memtest86 and the first scan gave the following results, with all 4 memory sockets populated:



Then I tested all 4 modules, 1 by 1 in the first DIM socket with 4/4 passes all tests..and there were no errors at all for either module.

Then I ran 1 module in each DIM socket in sequence, to check if maybe the DIM's had faults, but I did not get any errors then either.

Then I rearranged the memory modules and tested again with all 4 modules populated and got the following results.




I am very confused now, as to how the comprehensive testing for each module did not give any errors, yet with all 4 populated it gives errors.

I have tested booting with default settings in bios as well as with AI-Overclock , both with and without XMP profiles enabled, and the machine just does not want to be stable.

CINEBENCHR15.038 crashes randomly, and the Intel Benchmark test software crashes if I stress the memory.

In BIOS i have enabled 4x mode for NVM instead of default 2x and have not populated SATA ports 5 and 6.

My cooler gives me a score of 191 in the auto-clock overview in BIOS, and max temp i have seen was 78 degrees with 5200 overclock on all cores in a CPU stress test with the Intel benchmark utility. The CPU cooler never seem to get hot at all and feels cold to the touch at all times.

So I am starting to wonder if it is the motherboard that is unable to run with all 4 memory slots filled?

The memory i use is a set of 2x8GB with product number: CMK16GX4M2B3200C16
https://www.komplett.no/product/893831/datautstyr/pc-komponenter/minnebrikker/corsair-vengeance-lpx-...

As far as I know, the modules should be compatible with the motherboard.

Any advise or tips would be greatly appreciated.

Regards

Bård
Norway

[mdzcpa]

Here are a few things to consider or review:

The memory you have IS on the QVL, so that is good.

It is always best to get the matched 32GB set of memory than pair up 2 16GB sets. I suggest trying only a pair of sticks and see if it is stable with 16GB of RAM in dual channel mode.

Make sure you are running the latest BIOS. 0602 has been out awhile and is very solid. 0702 just released 3 days ago.

ALWAYS get your memory stable before overclocking your CPU. Go back to CPU default speeds until your memory problem is resolved. An overlocked CPU can make it more difficult for a struggling Memory Controller (that resides on the CPU) to find stability. Run your core multiplier and cache multiplier at default for now.


Run the RAM at DEFAULT speeds to start as well. XMP at 3200 is not Intel spec and is technically overlocking. See if your ram will run default 2133mhz. Then move it up. Most CPUs can handle 3200, so you should get there. But start with 2133.


Double check your Cooler installation. 78c is warm, but actually a pretty low temp for non water cooled CPU at 5200GHZ. Your CPU may be throttling. That said, get your system stable at DEFAULT speeds first.


Side note. Your NVME will automatically run at x4 if your M.2. is X4 capable on installed in either of the M.2 slots on the board. That NVME setting you are referring to is for putting in a M.2 card into the PCIE_3 slot that can be set to X4 from default of X2. Do you have something installed in PCIE_3? If you are using the M.2 connection on the board you do not need to adjust the setting in the UEFI. This should not effect your stability either way however.

I suggest you run the system of default CPU and memory speeds (2133) and run memtest with 2 sticks in dual channel and see if you get an error free result. If so, drop in two more sticks and see what happens. Also if 2 sticks run fine at 2133, move the speed up to 3200 for 2 sticks and see what happens. You can determine if the issue is the speed or number of sticks that trigger the errors.

[Toffypops]

mdzcpa wrote:
Here are a few things to consider or review:

The memory you have IS on the QVL, so that is good.

Good 🙂

mdzcpa wrote:

It is always best to get the matched 32GB set of memory than pair up 2 16GB sets. I suggest trying only a pair of sticks and see if it is stable with 16GB of RAM in dual channel mode.

These are two identical sets, of the same type of ram. Each set comprised of 2x 8GB module.

mdzcpa wrote:

Make sure you are running the latest BIOS. 0602 has been out awhile and is very solid. 0702 just released 3 days ago.

Thnx for the tip. Did not know a new one was out. Will update it.

mdzcpa wrote:

ALWAYS get your memory stable before overclocking your CPU. Go back to CPU default speeds until your memory problem is resolved. An overlocked CPU can make it more difficult for a struggling Memory Controller (that resides on the CPU) to find stability. Run your core multiplier and cache multiplier at default for now.

Thing is, that were just tests. I even tested using my system with default settings, and ram at defaults, and it still crashed.

mdzcpa wrote:

Run the RAM at DEFAULT speeds to start as well. XMP at 3200 is not Intel spec and is technically overlocking. See if your ram will run default 2133mhz. Then move it up. Most CPUs can handle 3200, so you should get there. But start with 2133.

I see varied views on this. Since the ram has been certified by the manufacturer as 3200Mhz, and supports an XMP profile as such, it is said that overclocking only occurs if you go above that. I do know though that the base speed is 2133mhz though, but that goes for pretty much all XMP memory these days.
Manufacturers certification should state it is stable at the given speed, or it is misleading marketing. I will try run a full scan with the memory at 2133Mhz.
But really weird if I cannot run it at the certified speeds with this motherboard. It is an overclockers motherboard.

mdzcpa wrote:

Double check your Cooler installation. 78c is warm, but actually a pretty low temp for non water cooled CPU at 5200GHZ. Your CPU may be throttling. That said, get your system stable at DEFAULT speeds first.

On idle/windows use the cpu rarely ticks over 30 degrees celcius. This Noctua NH15 is amazing at cooling.

mdzcpa wrote:

Side note. Your NVME will automatically run at x4 if your M.2. is X4 capable on installed in either of the M.2 slots on the board. That NVME setting you are referring to is for putting in a M.2 card into the PCIE_3 slot that can be set to X4 from default of X2. Do you have something installed in PCIE_3? If you are using the M.2 connection on the board you do not need to adjust the setting in the UEFI. This should not effect your stability either way however.

I am not sure what you mean here. The motherboard has 2x NVME slots, and i have a drive in each. Default settings in bios is 2x for PCIE according to bios settings.Thnx for commenting on it though. Good to know

mdzcpa wrote:

I suggest you run the system of default CPU and memory speeds (2133) and run memtest with 2 sticks in dual channel and see if you get an error free result. If so, drop in two more sticks and see what happens. Also if 2 sticks run fine at 2133, move the speed up to 3200 for 2 sticks and see what happens. You can determine if the issue is the speed or number of sticks that trigger the errors.

OK, i will try that. thnx for the tip.

Will update once tested.

Cheers.

Bård

[albertcjp52579]

I have same issue with my Asus ROG MAXIMUS XI EXTREME with 4 x Corsair 8192 MB (DDR4-2137) - XMP 2.0 - P/N: CMW16GX4M2C3600C18 :(.

I would like to know how you get on and fixed it.

Albert

[Phillyflyer]

albertcjp52579 wrote:
I have same issue with my Asus ROG MAXIMUS XI EXTREME with 4 x Corsair 8192 MB (DDR4-2137) - XMP 2.0 - P/N: CMW16GX4M2C3600C18 :(.

I would like to know how you get on and fixed it.

Albert

If you are using (2) dual kits..you will encounter compatibility issues...especially with the Corsair brand.

[MrAgapiGC]

Phillyflyer wrote:
If you are using (2) dual kits..you will encounter compatibility issues...especially with the Corsair brand.

I can confirm that. I did get a 4 stickt kit CMK32GX4M4B3600C18. BUT! on XMP i got a error on Real Bench On OpenCL Some type of anomaly REX meaningless Aside that.... ALL GOOD. I will recheck these once i finish the resintalacion since i change my nvme for more space.

But on test and other bench! 0 errors. I did test code xi with stock, stock with xmp 1 and 2, manual oc 5.0 and the AI OC to 5.1 and crash at 5.2 at my 4th cine rune. I am find with stock since i got a good score 5.0 is stable with 81C on 1 core using a x74 aio in 24C ambient temp. I am impress VRM on point most news are wrong.


The idea of 4pc kit was a friend of mine 2 years ago i got 2 3200mhz and 3600mhz I keep the 3600mhz.

[albertcjp52579]

Phillyflyer wrote:
If you are using (2) dual kits..you will encounter compatibility issues...especially with the Corsair brand.

Thanks for the reply.

Funny enough even I only install the 2 of them (both from the same kit) memtest86 still came back with errors.

[Toffypops]

Turns out it was two of the memory modules that were at fault.

1 module worked without errors at stock speed 2133Mhz non-XMP activated. No errors in memtest86
I tested it with another matching module, which was confirmed working fine in both stock and xmp 3200mhz profile.
Whenever i ran that one module, it would fail with errors in XMP activated speeds.

It also turned out one of the other modules were also bad, and that failed both in stock speeds 2133mhz and xmp-speeds 3200mhz.
It did not matter which modules I paired it with, it just was no good.

So for me the permanent fix was buying a set with 4x 8GB modules
https://www.komplett.no/product/954635/datautstyr/pc-komponenter/minnebrikker/corsair-vengeance-rgb-...

These worked perfectly in XMP-3200 dual channel mode. No errors at all.

So....lesson here is, NEVER EVER buy the 2x packs if you plan to upgrade, it is a really really bad decission.
If you ever decide to go 4x later on...swap out ALL the modules in a certified 4x set like the one above. You save yourself a lot of hazzle.

After I started using the new set above, my machine has not crashed once.

Best regards.

Bård

[Toffypops]

...and i can also add that my i9-9900K now is running stable at 5Ghz on all 8 cores with 3200 mhz XMP activated. 🙂

[mdzcpa]

Toffypops wrote:
So....lesson here is, NEVER EVER buy the 2x packs if you plan to upgrade, it is a really really bad decission.
If you ever decide to go 4x later on...swap out ALL the modules in a certified 4x set like the one above.

This is always the rule to follow to minimize memory issues. Glad to hear you have it sort out.