-
Notifications
You must be signed in to change notification settings - Fork 18
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
VRAM leak and JVM crash #52
Comments
Update: Manually removing the |
Can you please tell more on how to implement it? I dont really understand what does that mean but it seems like i am getting the same issue, same crash, same reproduction but i can't get rid of it |
Remove the mod until it's fixed by the developer. Or clone the repo and build a modified version of the mod yourself, and remove this line: https://github.com/anthxnymc/MagnesiumExtras/blob/1.20/src/main/resources/rubidium_extras.mixins.json#L18 |
@tom5454 i was trying to figure out why leak is caused by that mixin but i didn't found anything, and i can't replicate that issue |
Minecraft version: 1.19.2
Forge version: tested with both 43.2.19 and 43.2.17
Modlist:
rubidium_extras-1.19.2_v1.4.3.jar
rubidium-0.6.2b.jar
Using
ComplementaryReimagined_r2.2.1
shader with OculusHappens without Oculus installed with a lesser effect (the object count still grows at the same rate, but uses less VRAM?)
Oculus version:
oculus-mc1.19.2-1.6.4.jar
The worst case scenario in my testing is flying in the Twilight Forest with shaders on, leaks 6 GB of VRAM every 1000ish blocks.
Crashes around 1500-2000 blocks (tested multiple times with NVIDIA RTX 3060).
Running the same test wthout rubidium extras only uses a maximum of 2GB of VRAM, and goes up and down as needed.
Flew 4000 blocks in the same chunks, between two waypoints.
Reproduction:
Create a new world and start flying in one direction
See your Device memory and objects increase.
If you reach your GPU's max VRAM amount, the minecraft log has this message spammed
Then the jvm crashes with
EXCEPTION_ACCESS_VIOLATION
if you keep loading new chunks.Relevant stacktrace from the hs_err file:
The text was updated successfully, but these errors were encountered: