[SOLVED] Daily crash with XMP enabled

[tegzz]

Hello,

I built a PC in late 2020 and it works great, but I get 1 and only 1 crash per day after the first boot when it gets its first heavy load like starting to play a demanding video game. After I restart the machine it works fine with no crashes, until I turn it off and boot up the next day. The RAM is in the 2 and 4 slots as the manual recommended. I'm 90% sure its RAM related, cause when XMP is turned off and its running on 2133 mhz the crash does not happen.

My build is:

B550 Aorus Elite V2
Ryzen 5600x
RTX 3800
GSKILL Trident RGB 3600 CL 16 2x8GB
Corsair RM750

Things I tried so far with no result:

Updating BIOS
Turning off XMP and setting values manually
Setting just the clock to 3600 mhz and leaving the other settings untouched
Increasing the voltage a bit above the recommended value
Ran a memory check and no errors were found

these are the BSOD stop codes I got so far while trying the solutions mentioned above:

PAGE FAULT IN NONPAGED AREA
DRIVER IRQL NOT LESS OR EQUAL
ATTEMPTED WRITE TO READONLY MEMORY
MEMORY MANAGEMENT
dxgkrnl.sys
KERNEL SECURITY CHECK
IRQL_NOT_LESS_OR_EQUAL

Do you guys have any experience with an issue like this?

Help would be much appreciated.

 

[mOb69]

I had exactly same issue but whit ryzen cpu...

 

[Ralston18]

Actually I would suspect the PSU. Especially if the PSU has been heavily used for gaming.

When cold there is some PSU failure to provide suitable power to RAM.

Once warmed up the PSU is then able to provide the necessary power and voltages.

Look in Reliability History and Event Viewer to determine if Windows is capturing any related errors.

Start with the more user friendly Reliability History. Reliability History uses a timeline format that (if relevant errors are indeed being captured) should match the startup problem sequence that the PC is experiencing.

That said, "warming up" could just as well be applied to some other component in the PC.

Could be just a loose connection within the PC.

====

Power down, unplug, open the case.

Clean out dust and debris.

Check by sight and feel that all cards, connectors, RAM, and jumpers are fully and firmly in place.

 

[tegzz]

Actually I would suspect the PSU. Especially if the PSU has been heavily used for gaming.

When cold there is some PSU failure to provide suitable power to RAM.

Once warmed up the PSU is then able to provide the necessary power and voltages.

Look in Reliability History and Event Viewer to determine if Windows is capturing any related errors.

Start with the more user friendly Reliability History. Reliability History uses a timeline format that (if relevant errors are indeed being captured) should match the startup problem sequence that the PC is experiencing.

That said, "warming up" could just as well be applied to some other component in the PC.

Could be just a loose connection within the PC.

====

Power down, unplug, open the case.

Clean out dust and debris.

Check by sight and feel that all cards, connectors, RAM, and jumpers are fully and firmly in place.

Interesting. The PSU is the only component, that has not been upgraded in the new build. It has been constantly used for the past 5-6 years. Do you think upgrading it to a 850W would solve it?

Thank you for your insight.

 

[mOb69]

Interesting. The PSU is the only component, that has not been upgraded in the new build. It has been constantly used for the past 5-6 years. Do you think upgrading it to a 850W would solve it?

Thank you for your insight.

Definitely not in my opinion

 

[Ralston18]

Like many other products PSU's have a designed/built in EOL (End of Life).

At some point in time after "X" numbers of cycles, usage, etc., some component will wear out and fail. Either gradually with minor, growing problems or catastropically - just dead.

Varying and growing numbers of errors are, to me anyway, indicative of a problem PSU.

That is why Reliability History and Event Viewer should be checked. Reliability History's timeline format tends to make the error patterns more apparent.

You can do some testing on the 5 - 6 year old PSU if you have a multi-meter and know how to use it. Or know someone who does.

https://www.lifewire.com/how-to-manually-test-a-power-supply-with-a-multimeter-2626158

Not a full test as the PSU is not under load. However, any voltages out of tolerance would make the PSU suspect.

====

As for upgrading: More wattage (if that is the context of your question) is only a solution if the current 750 watts is not enough to keep up with power demand peaks being made by the overall system.

850 watts may provide some additional margin but my thought is that a new PSU even at 750 watts is likely to prove successful.

FYI:

https://www.tomshardware.com/reviews/best-psus,4229.html

Read the above link and use 2 or 3 of the calculators to determine the applicable PSU wattage to match your system.

Also do a manual total using each installed component's listed wattage requirements. If a wattage range is presented use the high end value. Then, once totaled, add 25% more. Plan ahead if you intend any additional upgrades.

 

[punkncat]

In the above suggestion about Event Viewer look for "41". It can be a result of bad PSU shutdowns (among other).

I would consider that the issue is RAM related since you state you can turn off XMP and eliminate the issue entirely.
In a case like this I would likely turn off XMP, but use those settings as reference. Slowly work from the base speed and run it up along with tightening timings until the issue returns. I would personally go stock voltage up to that point, then bump the RAM voltage a little to see if the issue resolves.
In my own case I can run my timings as tight as they can be, but have to relax the actual speed to 2933 (IIRC) on my RAM rated to 3200.

 

[Third-Eye]

Hello,

I built a PC in late 2020 and it works great, but I get 1 and only 1 crash per day after the first boot when it gets its first heavy load like starting to play a demanding video game. After I restart the machine it works fine with no crashes, until I turn it off and boot up the next day. The RAM is in the 2 and 4 slots as the manual recommended. I'm 90% sure its RAM related, cause when XMP is turned off and its running on 2133 mhz the crash does not happen.

My build is:

B550 Aorus Elite V2
Ryzen 5600x
RTX 3800
GSKILL Trident RGB 3600 CL 16 2x8GB
Corsair RM750

Things I tried so far with no result:

Updating BIOS
Turning off XMP and setting values manually
Setting just the clock to 3600 mhz and leaving the other settings untouched
Increasing the voltage a bit above the recommended value
Ran a memory check and no errors were found

these are the BSOD stop codes I got so far while trying the solutions mentioned above:

Do you guys have any experience with an issue like this?

Help would be much appreciated.

Do you have PBO enabled? I noticed on my system that PBO was causing bsods and sometimes memory errors as well even though my ram oc passed memtest86 four times with 4 full passes. I'd have to dial in all new settings to fix it, if it even would fix it and I barely notice any different in fps with PBO off.

I reset everything to defaults and reapplied all my settings except with 3600 CL16 ram overclock instead of the 3733 I had been using with my 4x8GB before, but skipped all CPU OC options. I've had zero issues or bsods since doing this and the system has been stable. I'm probably not even going to bother trying 3733 again since It barely makes a difference in fps.