Question Experiencing crashes after CPU, Mobo and RAM upgrade

[huggybear92]

Problem:

I installed new hardware (CPU, RAM and Mobo) a few weeks ago and been experiencing fairly frequent application level (can be several an hour, or little as one per 2 hours) and occasional system level crashes (maybe two times a day from 5ish hours of gaming but can be none).

Both system and application crashes only happen when playing games and some games seem to be worse than others, like space marine 2 is worse than god of war Ragnarok.

My system temps don't seem to high, the CPU can peak in the low 80s on rare occasions but mostly in a few hours of gaming wont exceed mid-high 70s, everything else is quite a bit cooler.

After reinstalling windows the crashes are still their but with less informative/different log messages.

System level crash logs:

Event 41 Kernel-Power

The system has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed, or lost power unexpectedly.

When rebooting:

Event 161 Volmgr

Dump file creation failed due to error during dump creation.

Event 6008 EventLog

The previous system shutdown at 03:12:57 on ‎29/‎09/‎2024 was unexpected.

Application level crash logs:

either:

Event 1000, Application Error

Faulting application name: GoWR.exe, version: 1.0.612.4312, time stamp: 0x66eb4ed6
Faulting module name: GoWR.exe, version: 1.0.612.4312, time stamp: 0x66eb4ed6
Exception code: 0xc0000005
Fault offset: 0x0000000000295275
Faulting process id: 0x0x3458
Faulting application start time: 0x0x1DB1256242B7049
Faulting application path: C:\Games\God of War - Ragnarok\GoWR.exe
Faulting module path: C:\Games\God of War - Ragnarok\GoWR.exe
Report Id: 5f486fb2-ca9d-44fc-9b92-31a4f3644956
Faulting package full name:
Faulting package-relative application ID:
or:

Event 1002, Application Hang

The program Warhammer 40000 Space Marine 2 - Retail.exe version 3.0.0.1 stopped interacting with Windows and was closed. To see if more information about the problem is available, check the problem history in the Security and Maintenance control panel.

Things I have tried:

I have run memtest and passed with 0 errors.

The event viewer logs mentioned drivers so I ran DDU without much change, then tried reinstalling window entirely.

In general benchmarking I am seeing very good performance as I would expect from this system.

I have also updated the bios to the latest stable version a few weeks ago.

System:

Processor Name: AMD Ryzen 7 7800X3D

Motherboard Model: ASUS TUF GAMING B650-PLUS WIFI

RAM: Corsair Vengeance Black 32GB DDR5

GPU: MSI RTX 4090 Ventus 3X OC

PSU: Corsair RM1000e

At this point I would assume the issue is likely hardware based and balance of probability more likely the newer hardware. However after passing memtest I am less sure as inconsistent errors seem to be more likely a RAM issue.

I am going to power down and check all the power cable connections but after that am uncertain how I could go about checking CPU/GPU/PSU and other hardware health in a manner that would allow me to narrow down which device.

 

[Misgar]

I know your RAM passed MemTest (how many complete passes?) but are you running XMP timings (technically an overclock) or have you tried setting the RAM back to Jedec standard 4800MT/s?

Similarly, are you running PBO on your 7800X3D. If so, switch it off temporarily.

Do you have all the relevant Windows patches for your hardware. Look in the list of "Optional updates" in Windows Update. Sometimes an NVidia driver update can fix things.

Finally, there's the 4090. Are you overclocking the Core or the Memory with Overdrive?

If you have compatible parts, swap the RAM, followed by the GPU and finally the CPU.

If your system only fails on some games and not on standard (more well behaved) apps and it passes stress tests with flying colours, you'll just have to wait until the game developer or NVidia issue a fix. Your new (faster) system may just be less stable in games than the old build. Life is like that!

 

[huggybear92]

I know your RAM passed MemTest (how many complete passes?) but are you running XMP timings (technically an overclock) or have you tried setting the RAM back to Jedec standard 4800MT/s?

Similarly, are you running PBO on your 7800X3D. If so, switch it off temporarily.

Do you have all the relevant Windows patches for your hardware. Look in the list of "Optional updates" in Windows Update. Sometimes an NVidia driver update can fix things.

Finally, there's the 4090. Are you overclocking the Core or the Memory with Overdrive?

If you have compatible parts, swap the RAM, followed by the GPU and finally the CPU.

If your system only fails on some games and not on standard (more well behaved) apps and it passes stress tests with flying colours, you'll just have to wait until the game developer or NVidia issue a fix. Your new (faster) system may just be less stable in games than the old build. Life is like that!

I am running with xmp timings and for memtest I ran default which from memory is 2 or 4 passes (around 2hr 40 run time).

Ill have to check my bios PBO setting but both GPU and CPU are running stock.

Will have a look at updates and look into possibly swapping out parts if i can.

 

[Misgar]

Ill have to check my bios PBO setting

I found the PBO settings in my Asus BIOS somewhat confusing when I built my 7950X system in December 2022. It wasn't obvious if PBO was on or off by default.

I suspect your problem is more likely to be with XMP than the GPU. Some AMD systems are a bit fussy about DDR5 RAM. There's the old XMP/EXPO argument, but specific AMD EXPO settings are rarer than Intel XMP settings in SPD. XMP settings usually work with AMD, but not always.

(around 2hr 40 run time).

That's around the minimum length of time I'd use. Some people recommend running MemTest overnight. When I have doubts, I'll run 2 or 3 passes (typically 4 hours or 6 hours). Even then, MemTest might not always spot an error.

I had some OCZ gold-plated DDR2 DIMMs running at 7-7-7-20 (fast) on a Phenom 955 for years and couldn't understand why USB file transfers got corrupted. It was only when I swapped from OCZ to Kingston the problem went away. OCZ developed a bad reputation for unstable RAM. You might be unlucky and have incompatible RAM too.

I didn't bother about the QVL (Qualified Vendor List) which can often be rather limited when the motherboard is first released, so I ended up with two "bog standard" 4800MT/s Corsair DIMMs (2x32GB) with no faster settings, because DDR5-5600 and DDR5-6000 16GB DIMMs were too expensive in 2022.

 

[huggybear92]

I found the PBO settings in my Asus BIOS somewhat confusing when I built my 7950X system in December 2022. It wasn't obvious if PBO was on or off by default.

I suspect your problem is more likely to be with XMP than the GPU. Some AMD systems are a bit fussy about DDR5 RAM. There's the old XMP/EXPO argument, but specific AMD EXPO settings are rarer than Intel XMP settings in SPD. XMP settings usually work with AMD, but not always.


That's around the minimum length of time I'd use. Some people recommend running MemTest overnight. When I have doubts, I'll run 2 or 3 passes (typically 4 hours or 6 hours). Even then, MemTest might not always spot an error.

I had some OCZ gold-plated DDR2 DIMMs running at 7-7-7-20 (fast) on a Phenom 955 for years and couldn't understand why USB file transfers got corrupted. It was only when I swapped from OCZ to Kingston the problem went away. OCZ developed a bad reputation for unstable RAM. You might be unlucky and have incompatible RAM too.

I didn't bother about the QVL (Qualified Vendor List) which can often be rather limited when the motherboard is first released, so I ended up with two "bog standard" 4800MT/s Corsair DIMMs (2x32GB) with no faster settings, because DDR5-5600 and DDR5-6000 16GB DIMMs were too expensive in 2022.

So yet to disable PBO but have turned off XMP, I looked into it further and my RAM CMK32GX5M2X7000C34 doesn't appear to be officially supported which as I understand isn't that important usually but can lead to greater instability for XMP clock speeds.

My brother has an identical combination (different GPU) and has had considerably better stability though still a few crashes. Think ill run an extensive memtest tonight see what I find, may need to consider returning them for the slightly different brands that are supported potentially or just RMAing them if memtest finds errors.

 

[Misgar]

A trick I used on an AMD 2600X to get it to run faster than the sanctioned speed 2933MT/s with DDR4-3000 was to relax the CL(CAS) timing from 18 to 20. If your DDR5 has a CL timing of, say, 36 at 5600MT/s, try manually adjusting it to 38 or 40. It helps to make XMP overclocks more stable.