Testing Every Modloader's Performance || The Lore of Modded Minecraft FINALE

Want more? Check here...

-=Socials=-

Second Channel: @LunarVods

Want access to the full interview and support my work?

Thumbnail by: @gutermaniac
Edited by: Kiz

-=Summary=--

In the beginning of its history, Minecraft had Forge, the one big modloader. Now we have Forge, Neoforged, Fabric and its fork Quilt, how did this happen? Where did it go wrong? Who ate my sandwich? Lets dive into the history of Modded minecraft together!

#forge #minecraft #moddedminecraft
-=Credit&Sources=-

Footage by ItsDrowned

Music provided by Infraction No Copyright Music

Other Music:
Tyler - Sweden Lo-Fi Remix

If You've Read This Far Comment Pebble :)

[Speiger]

FYI, tiny mistake in your testing. (Chunk Pregenerator dev here)

If you did really a 1000 chunk radius, you would have waited more like 12 hours per pregeneration test.
Because you would have pregenerated 4 million chunks instead, and that takes up a LOT of disk space.
Chunky Uses block values for everything so you did a 1000 block or 62.5 chunk radius.

But it doesn't change the result, i have seen basically over forge/neoforge the same performance in pregen speeds during my testing of ChunkPregenerator.
I usually do a 1 million chunk test, which takes roughly 4 hours to complete on my hardware, and forge and neoforge literally perform the same.

I can't talk about fabric, but I expect it to perform the exact same.

Overall pretty good video.

[appletree]

I think Mojang needs a "performance rewrite" update. If you don't play with mods and don't have essentially a quantum computer, you're basically screwed.

[MairanYT]

I don't need an increase in fps to make 2000fps out of 100 fps, but to make 60fps out of 20 fps with 500 mods.

[BurzowySzczurek]

I can tell you something interesting.
One of Iris and Sodium devs has been working on adding NeoForge support to Sodium for few months, so it looks like official NeoForge support is coming to Sodium.

[SimplyTheRedMelon]

i think when it comes to "fabric is better in performance" i think people mean it more on the modded side as for the base code idk if any of that even significantly changes anything
(and ofc fabric is modded i mean the mods that come with it just so nobody gets confused)

[paulbrancieq]

To be honest... Now that the subject is on the table, I never heard someone say that forge preform less than fabric. I heard that forge is heavier, with more layers and complexity. But it's interesting to know that without mods they seem to be equivalent. Also, I understand the frustrations when someone say "that thing is bad" without really knowing anything about the subject.

[Yasser-]

Over 60 fps isn't something people really need to play on all the time, it's more like an indicator that your game can handle heavy work and doesn't get too laggy when you start doing advanced stuff or use heavy mods. those avg 500 fps aren't useless as you think, if your game couldn't pass 60 fps in a fresh world, imagine how would it be if you played in that world?

[noelka]

I am not trying to make 1000 fps out of 100 fps, i'm trying to make 30 fps out of a pile of shit that crashes on startup... God bless the mods.

[BobzBlue]

I use over 150 fabric mods just for vanilla minecraft, runs great. I just scroll through modrinth client side mods or click the optimization tag and install anything that makes sense.

[nyxw]

Not a Fanboy, but I have always had a particular liking for Fabric. At the end of the day, however, the decision of which mod loader to use is really on the hands of the mod developers.

The "choice" for consumers/players really just depends on what mods they want to use, as it can be said that the only difference between Mod Loaders is the Developer Experience.

I'm not a Mod Dev, so these are all assumptions I made based on my limited knowledge of programming in general.

Great video as always. ^^

[piiiiko]

"you don't need more than 60 frames" (paraphrasing here), it's the same argument the cities skylines 2 dev made and that aged like milk instantly. none the less, love the videos ;)

[RaziMemeGod]

"What happened to a nice 60fps" high refresh rate monitors.

[SirPytan]

I think the main reason why so many people say Fabric is faster, is because they have already all the performance improvement mods in mind.
But I was a bit stunned, as others probably as well, that the bare bones mod loaders already increase the performance so much and that the results do not really differ. But your test does not show the real world performance everyone is referring to. That's what I assume, at least. Because who would just install a mod loader, except to prove a point.
So I would wish for a video, that actually compares performance mod configurations for the different mod loaders to have a good overview of the baseline performance you can maximal get, before slowing everything down with poorly optimized or just a tone of mods.

Like comparing:
1. Forge with Optifine + other performance mods
2. Fabric with Sodium, Starlight, and other performance mods.
3. Fabric with Nvidium
4. Fabric with the Vulkan mod, even though that one has not a lot of mod compatibility.
5. Quilt with, performance mods.
6. NeoForge with Optifine etc. if they are still compatible and have not diverged yet.

Do a performance Test in different categories, like Chunk Generation, Different amounts of Entities, Redstone tests, Multiplayer tests, Shader performance of the top 3 shaders or so, etc.
Because some mods have specific influence on a specific aspect, maybe show how these mods affect the numbers.
If there are two mods that kinda do the same thing, which one does perform better among the others or in different configurations. Is a performance mod, if it exists for two types of mod loaders, works better in one than in the other one?
These are the questions we ask us.


Then we have real world examples with a baseline of must-have performance mods for each mod loader and which mod loader can achieve the best performance at the moment with what we have as mods for them.

A thing that would really interest me is the performance difference with shaders between forge and fabric with the best performance setups and how for example Patrix Texturepack in different resolutions would affect the performance.
Are there maybe performance mod configs that would decrease the performance in some cases?
What about distant horizons, can other mods also somehow increase its performance?

Might there be mods that are more tailored for old hardware, than current hardware, so they would perform worse on newer systems?

These are the interesting questions to ask. I know some of these topics might require their own video, and they are more work to produce, but it would really be worth the test I believe. So I hope you read this and hopefully do test at least some of my suggestions and share the results with us. I would be glad to watch it. Thanks for your effort. You are doing great!

[crossscar-dev]

I have made many client side mods (where the fps lies) and in my experience fabric gave the best developer experience. This is because forge mixins still don't work half the time so you have to use weird workarounds and the forge api always just feels weird.

[Nord_Act]

I guess people manually meant situations when the same mod performs differently on different loaders despite having same functionality gameplay wise (i.e. Create).
Performance while playing with a lot of mods manually depends on what those mods are doing to do their stuff. Like... manually written mod vs the same mod made via MCreator very likely gonna perform better to some degree just because MCreator autogenerated code often consists of some real technical heresy (not like manually written mod can't have it too and to even worse degree)

P.S. Fabric doesn't have any patches for performance. It only adds an entrypoint for mods to load

[redouble_]

I mostly watch these videos because I'm trying to make more room on my SMP for extra players and it's hard to figure out how to best accomodate the most players without compromising experience

[ChaseMMD]

Another thing you can think about is the hardware running the modloader would impact the test. For example using a Intel or AMD processor. Running older hardware vs newer hardware. They're designed differently from generation to generation and performance scales differently depending on what the focus of the microarchitecture is running. Such as Intel being faster on integer calculations and AMD being faster on floating point calculations. In addition newer vs older instruction sets, with newer usually allowing for speed boosts if properly utilized in programming. So it's worth also noting that all that has impact on performance.

[bawnorton]

3:37 Fabric loader, by itself makes no changes to vanilla's code, performance patches or otherwise. It's made to be as lightweight as possible

Edit: Also forge and neoforge still heavily rely on patches instead of mixins, neoforge just has support for them built in now, which was a high source of contention between forge and neoforge and one of the many reasons for the split.

[surge]

its not about what loader is the best its about which loader can run the mods you want

[CobaltSpace]

Before we start, I assume that without any mods, they’ll all perform within margin of error of each other.