• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Unstable RAM OC consequences and SFC scan worries

Joined
Jul 11, 2023
Messages
205 (0.31/day)
Processor Ryzen 7 5800X3D
Motherboard Asus Strix B550-A
Cooling Be Quiet! Dark Rock 4
Memory Gskill Trident Z DDR4-3200 (16GB x 2)
Video Card(s) Sapphire Pulse RX 7900 XT 20GB
Storage Samsung 970 Evo Plus NVME 1TB (Boot), Samsung 970 Evo Plus NVME 2TB, Samsung QVO Sata 2Tb
Display(s) Aoc 31.5" 1440p 75hz; Asus 24" 1080p 75hz (secondary)
Case Be Quiet! Silent Base 802 White
Power Supply Corsair RM750X 2021 w/ Corsair Type 4 Sleeved Red Cables
Installed a 32gb kit of ddr4-3200 cl16 from g skill a couple months back with my 5800X3D and stupidly just set the speed to 3200 instead of turning on DOCP. PC ran ok but 3dmark was having some issues freezing, which seems to have been fixed by getting rid of my old boot SSD as well as doing a fresh install onto my NVME drive this time. All accompanied by a bios update and cmos clear, DOCP has been on w/ RAM speeds set to default so that DOCP is the only setting managing them, and a 4.5hr 4-pass test with memtest86 passed with no errors. However last night when I ran SFC Scannow via command prompt after a windows update, it detected and repaired some "corrupted files" (which is why I ended up running mem86) but hasn't flagged any more in subsequent scans. I always run DISCM first.

It was suggested on another forum that running my RAM wrong like I did *could* have done something, but I didn't even touch any voltages, it's just the timings that were off. It was also suggested that either my CPU, Chipset, or storage device could be to blame for these corruptions (the 970 EVO plus has been short and long SMART tested, along with CHKDSK, and runs cool under an EK passive heatsink).
I dunno how much of an issue any of this actually is and I'm having a gard time trying to understand what's going on. Can anyone help?
 
Now that you got the memory testing out of the way (I recommend using testmem5 with anta777 extreme1 profile, just let it run 1 cycle, should take 30 mins for 32GB - it's very heavy, if that doesn't error you're probs golden there), even if you already have done so, it can't hurt to be sure, run again:

dism /online /cleanup-image /restorehealth
sfc /scannow
shutdown -r -t 0

(the shutdown command reboots your PC clean without data from fast boot/partial hibernation - this can be the source of inexplicable problems sometimes) and you're probably not gonna have any further problems with Windows. After doing this observe for a few days, if you get crashes, run sfc again in a while and see if there is any corruption detected. Be mindful that a lot of things can trigger a sfc check failure, such as modifying a DLL on your Windows folder or changing a few settings.
 
It was suggested on another forum that running my RAM wrong like I did *could* have done something, but I didn't even touch any voltages, it's just the timings that were off.
Who ever said this really meant to say was memory errors can lead to data corruption. Your components weren't harmed.

By having the incorrect voltage and timings, just makes the memory unstable is all. Can't really harm the memory by adjusting timings, only if you raise the voltage beyond the XMP profile. The voltage part is more of a guideline because it's much more complicated. Easier to understand if you're aren't into overclocking memory to the extreme and stuff of that nature to say don't pass the XMP voltage.
 
Who ever said this really meant to say was memory errors can lead to data corruption. Your components weren't harmed.

By having the incorrect voltage and timings, just makes the memory unstable is all. Can't really harm the memory by adjusting timings, only if you raise the voltage beyond the XMP profile. The voltage part is more of a guideline because it's much more complicated. Easier to understand if you're aren't into overclocking memory to the extreme and stuff of that nature to say don't pass the XMP voltage.
oh yeah, all I did was set memory speed in bios to 3200 instead of just turning on DOCP like I should have, pretty sure it was still displaying RAM voltage at 1.35v anyway, it's just the timings were horrid. I might run another batch of Memtest86 tonight when I go to bed but it passed fine last night.
 
Worst case scenario is that you won't be able to fix your OS and you will have to install Windows again. Good idea to use a spare OS drive when testing RAM since it can lead to corruptions. Now that you have it stabilized hopefully that won't happen again.
 
@OP

Instead of doing a fresh install, should you not be using DSIM. You can't use this feature if you have bad memory. Fix this first then use DSIM.
 
Back
Top