3080ti @ 4k @ 120 FPS, High rez disabled & lowest settings enabled. 120% GPU VRAM usage. Why?! 13GB character screen, 16GB in game. GPU only has 12GB. VYSNC and FPS cap enabled

The game definitely has some kind of vram memory leak. When I start the game it uses maybe 75% of the vram and then after a time (sometimes hours) it will go up to 100% and micro stuttering starts until you restart the game or change the texture settings and save them, which seems to force it to reload and start over. It seems to build up more slowly possibly with lower settings so I just have textures on medium now and it doesn’t occur as often or perhaps builds up much slower so I’m never playing it long enough to encounter it at that setting.

Alt tabbing may also be an issue with it. I noticed it especially when I was running WoW and D4 at the same time to camp rare spawns in WoW and tab to them when I heard rare scanner make a sound. After awhile when D4 maxes out vram it can cause tabbing back to WoW to freeze up the computer briefly and then D4 dies and you get a Fenris crash report. Maybe people who don’t experience or notice the vram leak don’t alt tab much?

If you hear the fans spin up while talking to vendors, you’re likely experiencing a bug many other people are that’s talked about a lot in this thread: Game Freezes/Crashes GPU fans screaming when opening specific UI Elements - June 2023

Capping your FPS doesn’t seem to help with this particular issue, possibly due to what one poster describes in that thread:

As people have pointed out, the most likely scenario is the vendor windows and main menu are being rendered on their own ignoring all options and likely FPS cap, which is resulting in the card not only double rendering, which is already very bad and stupid, but also the card being overcalled due to this runaway threading, this is similar to what happened with new world with its main menu overcalling the card to the point it was trying to hard render thousands upon thousands of FPS directly.

This behavior is near the point of being a power virus furmark kind of load causing this out of control load hit on cards that shouldn’t happen in any way for a properly coded game. Regardless of hardware faults being uncovered by this, it doesn’t matter, this behavior SHOULDN’T happen and yeah, needs to be 100% fixed by the games coders. Nvidia cannot fix this with drivers, AMD cannot fix this with drivers, Windows cannot fix this with an update, only the games coders themselves can fix this and are the ones that need to get a handle on this.

As far as temps go outside of the vendor bug that can’t be helped much, I found disabling NVIDIA reflex lowered the temperature more than anything else at other times, though it didn’t stop the vendor bug from ramping up temps when you have their UI open. I also capped my FPS at 100.

Turning NVIDIA reflex off lowered temps by 10-20 degrees outside of the vendor bug for some reason. I assume there’s something wrong there as well as I don’t think that setting is supposed to have that kind of impact.

1 Like