Question nvlddmkm Event ID 14, but only with 3700x

[tkurgpold]

this is a bit of a weird one. i'm posting this in the cpu section given that despite the fact that it's a gpu error, the error has proven to be be dependent on what CPU I have in the computer, and i've confirmed that the gpu is fine and the situation isn't changed by updating/downgrading gpu drivers. i've had my 660 installed with my previous CPU, a 2200g, for two days now and had no issues whatsoever. as soon as i put the 3700x in, bam, totally unusable. and it's not entirely clear whether or not the cpu is bad, given that it's fine in safe mode or without gpu drivers.

anyway. copying the main text from a thread i made elsewhere just because i haven't had any luck anywhere else trying to figure this out. sorry if it's too long or anything, i've tried to remember everything i can that would be relevant.

specs:
ryzen 7 3700x [purchased in november]
evga gtx 660 sig2 superclocked [latest drivers]
16gb ram
asus prime b450 a/csm [bios 2409]
thermaltake gf1 750w [purchased sometime around july 2020]

the problem:
"Event ID 14 from source nvlddmkm cannot be found"
desktop arbitrarily freezes up, though i can still move the cursor. lasts maybe 10-15 seconds, then a black screen, then desktop returns to normal. event viewer shows the dreaded "Event ID 14 from source nvlddmkm cannot be found". generally starts happening as soon as i login. works fine in safe mode, or when i dont have gpu drivers installed. also works fine if i use the 660 with my old 2200g.(which is an improvement over running the 2200g with the igpu, but ultimately i do need the 8 core to be able to get work done. music producer, trying to make headway on an album i've been working on.)
i'm fairly confident i've managed to corelate this to when the gpu goes up to boost clock, i watched precision x1 and had it freak out 6-7 times and each time it was within a few seconds of ramping up to 1045mhz.
started manifesting itself maybe a few months after i picked up the 3700x. it started out rarer, and then got steadily more frequent to the point that i stopped using my pc for several weeks and tried to buy a new card thinking mine was dead.

what i've tried:
different versions of the gpu driver going from the latest all the way back to 2019 with ddu each time
updating chipset drivers
reformat+fresh windows[seemed to work at first, but after a reboot it came back]
precision x1 boost lock
high performance mode in nvidia control panel
updated bios to version 2409 which, i believe, has agesa 1006. dated september 2020.
underclocking as far as precision x1 will let me(at some point it ended up holding itself slightly under full boost clock at one point the other day, and was stable the whole time, but it's hard to say if that was a fluke or not. when i reinstalled windows earlier that day, it was also totally fine until i rebooted. without having really installed anything, by the way, so that shouldn't be the issue.)
increasing tdr delay to 10 seconds
resetting bios settings
disabling SMT
installing gpu drivers without geforce experience, hd audio, physx
running armory crate uninstaller multiple times on the previous windows install. thought that was the problem, was stable with the 2200g, then swapped to the 3700x and it was unstable again.

at this point, as far as i can tell it's down to either an incompatibility between the 660 and the 3700x, or the cpu is bad somehow. since i bought the 3700x during black friday week when bestbuy had a sale on it, i would assume that i can still rma it. should i? is there any specific reason to think that it's the cpu thats bad as opposed to the card and the cpu just disagreeing with eachother? i'm fine to rma it if that'll fix the problem, since i can't use the thing rn anyway, but would like to hopefully not have it be a total shot in the dark, which is where i'm at right now. i've seen a lot of evidence of nvidia+ryzen 3000 chips throwing this error, but they also seem to be mainly turing cards.

thanks for reading.

 

[rgd1101]

why not the latest bios?

 

[tkurgpold]

why not the latest bios?

my fault. last time i updated my bios, it bricked the board and took out half my ram on the way out(still not sure why, everything was done correctly. maybe the ram took out the board, who knows.), followed by a month and a half of back and forth with Asus support. obviously i was very apprehensive about trying again, but when i found out i had the network update feature, i went ahead and did that. wasn't until i checked the version number with the website that i realized it gave me an out of date version, but i never updated it any further since i was already past the version of AGESA that i heard had fixed the problem.

if going back and getting the fully up to date version will fix this though, absolutely willing to do so.

 

[tkurgpold]

not much of an update, but i've managed to confirm the gpu is definitely entirely functional. i've had it running with the 2200g for the past few days, subjecting it to normal use, making sure it's stable while boosting, and left gta 5 running on it for a couple hours. which means it's definitely something weird with the 3700x, and i'm not crazy. i still can't seem to find any other evidence of anyone else having this specific incompatibility, which is a little disheartening to put it lightly.

i find it extremely strange that this would crop up only after a.. couple months? maybe? of using the new cpu without any issues. i never overclocked it, i never overclock the 660, and i can't recall any specific event that occurred when the freezes started happening.

very stumped, very frustrated. if this were strictly a gaming pc, it wouldn't be as much of an issue, but i really need the extra cores for creative work.

 

[tkurgpold]

yeah. i don't know what to do anymore. nobody else seems to be having this problem, and now i'm stuck having dropped $300 on a cpu i need but cant use, and the shortage looks like it's going to last until next year. i am screwed.

 

[rgd1101]

and you did update the bios?
check for bend pin on the cpu?