Out of Video memory - Common error and my thoughts on a workaround - Discussion

HayuPhukmiWife · February 4

Mods, if this gains any traction and should be moved, by all means. I am just sharing my experience here for discussion. I posted this as a reply in a GPU thread, and it turned into the post I was gonna make here, So I just copied it over.

I was gonna post a 'my experience' writeup on this topic. Seems to be happening to a lot of i9 people. I've done a crapload of research and came to a few common workarounds, and then a solution for myself that differed from the common solution. Maybe my insight will help. I'm not a guru like the people on here, but I did have this issue and wanted to post about it in general. In searching, I saw your thread and figured I'd post quick before I wrote up in a bit more detail. Everyone keeps saying look to the GPU, the drivers, the SSD firmware, fresh OS install etc.... Here are my thoughts.

For context, I'm running a 14900kf and 4090 OC edition with 24GB of VRAM. There should be no way I got this message either.

Started in a few games, then spread to other games. I did some looking and similar systems with high end GPUs and more specifically i9 chips see mto be the only ones I found affected. i9 intel cpus seem to be the common denominator here with Generation varying and GPUs varying. Everyone seems to have the same story, Started in a few programs, then started happening widespread. People had tried OS reinstall, warranting the chip, the mobo etc, problems would fix, but then come back eventually. etc etc. Just as I experienced w/o the warranting of hardware.

Seems the error is misleading. IMO It never ended up 'actually' being video memory. I'm personally running 24gb of video mem, and when being run in these games that crashed with the error, never had less than 20gb left available. Games would usually crash at initial loading screen.

A lot of the GAME-SPECIFIC threads had a common solution that seems to work for many by underclocking their CPU. eg going from a 45 multiplier to a 44. or reducing clock speeds by as little as .25 have been reported to work in some cases I read up about. Great solution, right? well maybe. if you're into that. But it seemed to be a functional workaround.

So I didn't try this, and its running like a top now with an alternate solution. Nothing crashes anymore. The common solution seemed to point to the stock clock or overclock settings stressing the CPU and needing to dial back ever so slightly to avoid the issues. For reference this happens with my KF CPU with no integrated graphics, and others that have it.

In monitoring I notice core throttling and high temps, as do everyone with these 12-14 and more 13+14th gen fireballs. Seemed to relate (although maybe just in my rationale). Instead of downclocking, all I did was disable MOBO voltage enhancements and run Intel's stock settings. (Also updated UEFI, as it seems in some MOBO cases provide different voltages to the CPUs, however, this was sorting previous issues not this specific one). On my Z790 board. Strix z790E I simply click 'DISABLE ASUS MULTICORE ENHANCEMENT - ENFORCE ALL LIMITS' (Intel limits for cpu applied) and it solved the problem. Basically undervolting... I assume this is similar to a downclock, as it will put less stress on the cpu? I dunno. Maybe the guys here can offer some additional input, but this works fine for me. When I turn the voltage back up or try and push the clock to 6.2 from 6.0 or something, it works fora bit, then starts to degrade and the errors come back. If i leave it alone, it leaves me alone. True story.

Maybe this will help before people run down the rabbit hole.

Community, Thoughts?

Skiiwee29 · February 4

<-- Thread Locked -->

Your previous thread was moved from General Discussion to Graphics Cards since the conversation revolves around GPUs and Video Memory. Please continue in your existing thread. No need to repost.