TheUsefulLists / UsefulMods

Just a list of useful mods
https://bisecthosting.com/UsefulLists
MIT License
1.84k stars 135 forks source link

[Correction] redundant mods #239

Open ColonelGerdauf opened 1 year ago

ColonelGerdauf commented 1 year ago

Checklist

What mod is this correction for?

LazyDFU and Starlight

What section(s) of the list is this correction for?

Performance

What version(s) does this correction apply to?

1.20.X (Forge), 1.20.X (Fabric)

What needs correcting?

Both of these mods are marked as redundant and unnecessary by their authors as well as others in 1.20 (LazyDFU from 1.19.4).

The author of LazyDFU has expressed that he has placed 1.20.x support due to people blindly nagging him to update the mod and ignoring his written notices.

Input relevant link(s) which can be used to verify the correction.

https://github.com/astei/lazydfu/issues/32 https://github.com/astei/lazydfu/issues/35

https://github.com/PaperMC/Starlight/blob/fabric/TECHNICAL_DETAILS.md < at the very top

(OPTIONAL) Any extra information about the correction.

No response

ItsukaHiro commented 1 year ago

I know about LazyDFU, and now I prefer ThreadTweak (maintained fork of SmoothBoot).

But Starlight still have some impact on performance, not much as pre-1.20, but still better than vanilla.

NordicGamerFE commented 1 year ago

I am not removing starlight, and i've contacted the LazyDFU dev for more information

ColonelGerdauf commented 1 year ago

The note of redundancy seems to have been comparative to the original improvements. Also, it was written, plus what the devs were feeling, prior to the 1.20 Starlight update. I have done some rudimentary calculations based on provided numbers (place then remove) :

Original improvements (pre-1.20): phosphor: 1.35x - 26%, 1.34x - 25% Starlight: 37.5x - 97.35%, 13x - 92.5%

New improvements (1.20 and beyond): Starlight: 2.4x - 59%, 2.3x - 57%

With these kinds of values, I can understand how Starlight can still be seen as useful. It's improvements tanked relative to the old days, but still performs better nowadays than Phosphor did originally. Also, as an absolute, getting under 0.1 millisecond for time consumption is always nice, as that is practically instant.