Hey everyone,
so recently I’ve been thinking about the combination of Forge and Sponge. While I love Forge and I love the idea of combining Sponge and Forge / building Sponge upon Forge, we all know that Forge is not the very best when it comes to performance.
Although this usually isn’t a big problem for modpacks and such, it may turn into a problem for huge servers with thousands of players like Hypixel, Mineplex etc.
On the other hand if you only have Sponge as the only Forge mod installed, it will likely not be as noticable as with big modpacks with 100-150 mods.
So this goes mainly to the developers: can you make any prediction on how much of an effect on performance using Forge will have on Sponge servers? I would expect it to be slower than Vanilla Minecraft and Bukkit servers, but I cannot predict how much or if it will turn into a problem.
Forge is a wrapper over Minecraft server, therefore, no technical performance degradation unless mods do load crap over Minecraf tserer.Only specific aspect can be measured.
Sponge mostly uses vanilla classes and patches them at runtime, therefore will get a good performance close to vanilla.
I would like to see some performance tests with Forge against vanilla because I have a feeling there is little difference. Lag on forge is mostly caused by mods by third parties, but with sponge alone there should be little difference.
Note: I have no evidence, this is just what I’m thinking right now.
again, performance only measured in specifics. there is no wrap around methd for forge becuase no modification of the sever ocurs until runtime, use for forge only.
I think bringing over some of Cauldron’s performance patches would be a very good idea, and it should be safe, too, since the forge specific patches to which you are referring are separated/separable from the bukkit side of Cauldron.
“Performance” is not a nice, clean, constant, matter of fact, finite object that many people believe it is. It is affected by everything between hardware and the minecraft server you are running and all its contents. That means your operating system, drivers, any other software running parallel, and any other components that might be helping your system.
Sure, benchmarks can be done and programs can be built with performance as a goal, but at runtime, how it performs is in part the software, and in part the system it is running on.
In the case of forge being alleged as not as efficient, I would have to agree with the former replies. Yes, a forge server has a longer startup. That’s because the MC server isn’t multithreaded, and even if it were, forge loading async to the main thread would be the epitome of thread unsafe (at least I have to think this is the case). During operation, it’s going to be mods that slow down a forge server, not the wrapper of forge.
Okay thanks for your answers guys ^^
I’ve always used Forge for big modpacks only, so I have no experience with how it performs without / with very few mods.
Oh, and @bdubz4552 I know performance is dependent on all those things, but that doesn’t mean you cannot compare performances of different algorithms
Forge with no mods (I.e. effectively vanilla) performs almost identically to vanilla. The threading model is effectively the same and the class mappings add essentially no measurable overhead. The only thing is that startup takes a little longer because Forge’s bootstrap is more complicated.
If the cauldron/spigot optimalisations could be implemented that would make me really happy.
Not only is cauldron way faster than vanilla server, it offers great features as in limiting mob spawns etc…
Without cauldron: anyone walking through extreme hills would cause server ticking behind 30seconds. With cauldron not an issue, no lag at all.
Without cauldron more than 300 mobs would cause 10 secs ticking behind. With cauldron no issue.
If i try to run my server with vanilla forge with some players online my server ticks behind 60-90 seconds. With cauldron no issues like that.
I’d be very relieved if cauldron would live on within sponge.
