Hey! So as the title states, I am running a fresh 2950x build, but sadly am having some issues with stability.
The BSODs happen very infrequently, sometimes on shutdown, sometimes when I close Chrome, sometimes even just when I boot up. The messages are also not the same every time - In order of frequency:
DRIVER_IRQL_NOT_LESS_OR_EQUAL (sometimes without the driver part)
KERNEL_SECURITY_CHECK_FAILURE
DRIVER_CORRUPTED_EXPOOL
KMODE_EXCEPTION_NOT_HANDLED (with various .sys files: USBXHCI, ndis)
SYSTEM_THREAD_EXCEPTION_NOT_HANDLED
INTERRUPT_EXCEPTION_NOT_HANDLED
Interestingly enough, the Minidump folder is empty, if there is an option I have missed to enable the crashdumps, that’d be much appreciated!
A quick overview of my specs:
CPU: 2950x, not OCed or anything
RAM: 64gb of Vengeance LPX 3200
MOBO: AsRock x399 Taichi
GFX: MSI 1080Ti Gaming X
Storage: 860 evo (Data), 970 pro (system), 4tb wd blue (backups)
You got the minidumps you can toss at us? They will generally provide more information on top of what has been provided to us.
In the meantime, what is the RAM clocked at in BIOS? if it is set at 3200, have you tried setting it to 2400 or 2666 and seeing if the computer is stable then?
Is this a the first Windows install? Is reinstalling from a new image an option?
Hey, thanks for the quick reply! This is kind of the second install, after I formatted the partition to get rid of my “Pro N” install after I’ve noticed that my Pro Key wouldn’t work with it. The partition has been cleaned completely (even with the extra partitions like backup) before the “reinstall”.
The RAM is clocked at 3200mhz, i have not changed anything in the bios yet apart from loading the XMP profile so their timings are correct.
The Bios is up-to-date running v3.30
The computer in general is pretty stable, as the problems only occur whenever I want to shut down or am about to start work. Never just while I’m doing “stuff” (coding, rendering, casual browsing etc).
I have the options you’ve described enabled already, but both locations are empty. Could it be that a windows update cleared them out? If that’s the case I sadly would need to wait for another crash to come
What speed is the memory running at? Try 2933? 8 sticks of 8gb or 4 sticks of 16gb?
Is that v4 or v5?
The hynix memory will be… not stable… at 3200.
2133 can also be unstable, I’ve found, so you want to be running 2400-2933 imho. You might be able to manage 3200 with 8x8, maybe.
There is also an idle power setting in the bios, somewhere, that may need to be tweaked.
there is a legit issue where using too little power will cause issues with certain PSUs. It actually is a thing. But that’s not the only thing if you’re one of those poor unfortunate souls.
I am using a Bitfenix Whisper M 850W PSU, which I hope should be a good one.
I have bought the 3200mhz ram as I’ve read that the higher clock speeds are helping out the ryzen CPUs.
This is my first PC build in over 9 years, so the landscape has quite a bit changed and I had to research quite a bit beforehand. I will try to find the idle power setting as well as dial down the RAM to 2933 and see how it goes. So far, the pc has been stable for over 5 hours, so we’ll see.
Which value would you suggest I use for the Idle power setting?
Edit: I have looked for a SAMSUNG RAM for exactly the reason that there seem to be issues with non-samsung RAM sticks. It looks like I picked a “good” model, but eh.
I actually don’t know if it’s about the RAM or not. The computer works fine with 2 sticks as well, just quad sticks seems to have issues from time to time.
Additionally, I have noticed that when I tried to install Pop!OS on a separate partition, it froze completely unless I only had one stick installed.
RAM looks good.
Thanks @Novasty for recommending Thaiphoon burner. I usually recommend it myself for diagnosing memory issues like these.
What this could be, (IF 2 memory channels are unaffected) is that potentially the CPU isn’t seated just right in the socket.
Or alternatively and don’t discount this when building a new system. One or a few of the RAM sticks are bad out of the box. Granted that’s rare but it does happens sometimes.
So test the DIMM’s 2 at a time going through all the channels and see where it goes wrong.
If those all work and it’s only the fully populated configuration that fails then one can continue to figure things out from there.
At that point generally one has to reduce the speed and or tweak voltages and specific sub timings and figure out what can be done to get the system to be happy.
I’ve had the same hunch, but no luck so far. I have tested them in single (all one by one), then rotate them around (1, then another one, another one, … - 2, another one, etc.) - took me quite some time.
Even after the memtest, i had no errors, which I find weird.
If it’s the CPU, phew. not that easy to get it out as I don’t have any alcohol pads handy to get the paste off and such.
Let’s see how the timings do, but it still bums me out that I cannot run them at the speed i bought them at (if it really is a speed issue).
I would like to refute this with my anecdotal evidence that my board is fully populated 4x8GB sticks on my Ryzen 1800X running at 3400 using the XMP profile.
That aside, it isn’t a bad formula to live by, but it isn’t a guaranteed either. I think it is generally better to wait and see what other people are able to achieve with a broad range of hardware combinations and make and educated guess off that.
Hey, thanks for the hint - the tool exited successfully and found no errors. It would have been weird to have problems anyway I guess, as the BSODs already happened during my very first install unless I only had 32gb installed.
What does the memtest not cover when windows and linux fail during bootup? Also, it never even once crashed during a benchmark (heaven, pcmark, cinebench) or a gameplay session, always when I least expected it (think spanish inquisition).