not saying it cant be the board, but many here point the finger at msi, without even applying the basic bios"tweaks" that should be done for AMD/AM5,
like disabling C-states to prevent usb related issues.
your usually better of setting things like voltages (soc/ram/vddp etc), clocks/timings/ram manually, dont have any insides, as i stopped working in shops before AM5 came out.
many forums like guru3d/tpu and others have basic guides for setup.
@
jakeift153102db
reset bios to defaults, power down, do bios flashback, load defaults, reboot, then see if you can change stuff.
@
Piledriver
QVL has ZERO to do with stuff working or not, just a list that certain ram was tested on that board.
after +50 builds/+1000 repaired, i can say enough kits that were on the list of the mb (Gskill) didnt even post past jedec settings, and those "oh so horrible" corsair kits (not on QVL), everyone says to stay away from, worked as specced on the same exact board.
in short, unless you have used the same rig and tested +1000 ram kits for each "side" (QVL/not QVL), you have no statistical relevance that QVL makes any difference.
@
Globespy
the "lower latency" is an msi feature, doesnt do much for ppl that can manually tune ram.
have you verified yours is stable (TM5/HCI and something like core cruncher for CO)?
+200/-20 is not that extreme, but short of fine tuning a lot regarding CO, usually -10-15 is what most can do (if just enabled).
at same clock/timings/voltage, the cost of the ram has no direct correlation to how far i can tweak it, its the die being used that does.
you can gain more on amd cpus tweaking primary/secondary timings, than what most will get from PBO on, even possible to do better than specced.
e.g. while im on ddr4 my corsair kit (3600C14/1.45v) is performing about 10% better with slower primary (C16) but lower secondary, while running at 0.1 lower voltage.
@
ghal155a02dd
many bios are now released in beta (or would you rather have no update for fix/new hw?), and many are being renamed to non-beta, without any further changes (besides testing), not just on msi.
did you disable msi driver install in bios?
did you do gpu driver install/cleaning while connected to internet?
unless the beta improves/fixes anything, use the latest non-beta is a good idea.
no, flash doesnt always clear cmos. i personally load defaults prior to flash, as well a right after, on some Gb i even had to use flashback to erase previous settings.
i have done 5 flashs in the past 12 month and redo almost the all settings incl redoing fan profiles for 5 fan groups each time.
take pictures or write down settings, if thats too much "work" (in relation to "gains" on perf/noise), buy a mac, and be done with it
disable C-states option in bios, see if it helps with usb, if not, check SoC v, too much/not enough tends to impact it.
and no, memtest86 (or any other variant of it) will NOT find errors from oc. its designed to detect DEFECTIVE dies, not instability.
part of the reason why im not surprised with the usb issue, as i assume its not really stable (timings/clocks/voltages).
test 24h with TM5/HCI or similar, without getting errors, THEN your good.
@
laurentiu.v
disable win driver update and disable MPO in win
disconnect from internet (for wifi, worst case, turn off router), uninstall driver, clean with ddu in safe mode,
and reinstall the driver, connect internet again.
disable win driver auto update
disable MPO