Microstutters on memory clearing
Whenever the memory gets cleared in Minecraft, I get 1-2 frames of microstutters. This happens with DH + Iris, and with FerriteCore.
CPU usage is around 60-80% on world loading, is stable around 40% during gameplay.
Allocated RAM is 8gb, Memory usage in Minecraft at 5-22%, global RAM usage >50%.
GPU usage 80-100% (100% aswell for only Sodium + Iris + Shader, no stutters there)
AMD Ryzen 5 5600X
Nvidia GeForce RTX 3060Ti
40 Replies
Recording with phone because capturing programs affect performance
GitHub
UsefulMods/JavaArgumentsOptimization.md at main · TheUsefulLists/Us...
Just a list of useful mods. Contribute to TheUsefulLists/UsefulMods development by creating an account on GitHub.
Thats what I personally use, they work great in my experience
I will take a look and try it out, thank you
The game crashes without any logs upon starting the game with all the flags enabled
What are your flags rn
I'm sorry, I made a mistake and now the game starts up correctly. I forgot to remove the "java" term at the beginning of the flags. CurseForge only asks for additional Flags, which java is not additional
I will test the performance now with the flags enabled
It sadly did not resolve the microstuttering
https://www.graalvm.org/
You can try using GraalVM in addition to the flags
GraalVM
GraalVM is an advanced JDK with ahead-of-time Native Image compilation.
Do I just use the javaw.exe in the bin folder as java version in the launcher?
Seems to have worked
Yup
Loading into the world, the microstutters seem to be gone
I will keep the world running for a bit
Big thanks already for the effort!
Np! Have fun playing
After having the world open, the stutters come back after 2 minutes, sadly
aifjuhdrsgidhg
In that case, I dont know :(
How good is this?
The best one Ive tried
I never have issues with stutters unless Im running out of RAM
And last time that happened was while I had 6 gigs allocated with DH and 6 ppl connected to me via e4mc
It is a JDK alternative?
How does this compare to Adoptium OpenJDK, or classic Oracle JDK?
what i do is just give it a lil more ram so it never hits 100%
just as a snack
i sometimes have that problem too
I would run out of RAM if I would do this
OP isnt running out of ram
Neither am I, I have 16 GB of RAM, and I just set it from 4 to around 6.5 so it wouldn't reach 100%
Upon further investigation I found that the stutters only occur once a certain amount of chunks have been loaded. Immediately after loading the world they don't occur, yet after a minute of loading they start and get progressively worse over time
How high is your real render distance and DH render distance?
Oracle also has a graalvm but idk if that's the same to the graalvm link he provided
I havent done any thorrow testing but its a lot better than the stock everything
I never have lag spikes anymore
You can always just try it, see if it works well for you and if not just revert to how it was setup before
I never had any big lag spikes or noticeable stuttering, I just enabled custom GC flags because it is teoreticly better, and can switch to graalvm if it should be better, but I don't think that I could pin point the difference.
I curious though why is that better, while setting the custom flags I've read what every flag does, and why I should use it, I can read why graalvm is good, but I don't know where to find what I'm replacing, a.k.a. why Adoptium is bad
They are both good
Just use anything other than the default stuff lmao
Why does modernfix team recommend azul zulu
But the default stuff should be best on average, if there is a better alternative it should replace the default value
.
E.G. iIrc G1GC was by default running on single thread, but at some point it was changed to give it more threads by default
and ZGC sould eventually replace it
.
If graalvm is better, but experimental I can understand it not being default, but I didn't see any experimental warning or other stuff
graalvm compiles stuff more aggressively, or smth afaik
Ask modernfix ig
Yes because default minecraft runs best /s
Minecraft is dumb and stupid and if you want it to run optimally, you have to test what works best for you
But this is getting #off-topic
but more then what?
It is easy to say my terrible Python app is better ten competition is by competition I mean a 30 year old Python 1.0 app. But the real competition mught be faster then my app.
If there is no named comparasion data, with explaination why and how was this result achieved, I can myself fabricate any data I want
6 and 48
I've found this comparison: https://whichjdk.com/
I might give GraalVM a chance
I've also found that some parts of GraalVM are being implemented in other OpenJDK releases, i don't know what part does what thought, so I don't know how to value this fact
Sorry, didn't see there were other messages sent after the original question. I'm using 6 MC render distance and 48 LoD
And this: https://www.reddit.com/r/feedthebeast/comments/uz21wc/graalvm_speeds_up_modded_minecraft/
But this is 2 years old :/
Also do I understand correctly that GraalVM does not support the new improved ZGC?
https://github.com/oracle/graal/issues/8117
GitHub
[GR-45919] Add support for Generational ZGC on HotSpot · Issue #811...
TL;DR Add support for Generational ZGC (JEP 439) to the Graal compiler. Goals JDK 21 added support for Generational ZGC, which Improve(s) application performance by extending the Z Garbage Collecto...
Well that is a problem I see