Bobby
NebelNidas opened this issue · 16 comments
Mod name
Bobby
Curseforge link
https://www.curseforge.com/minecraft/mc-mods/bobby
Modrinth link
https://modrinth.com/mod/bobby
Other link
https://github.com/Johni0702/bobby
What it does
Bobby is a Minecraft mod which allows for render distances greater than the server's view-distance setting. It accomplishes this goal by recording and storing (in .minecraft/.bobby) all chunks sent by the server which it then can load and display at a later point when the chunk is outside the server's view-distance.
Why should it be in the modpack
Because players can see further than what the server's view-distance is set to, if they want to.
Why shouldn't it be in the modpack
Starlight currently has an issue with it, but Phosphor works fine: PaperMC/Starlight#38
Categories
- Performance optimization
- Graphics optimization
- New feature
- Optifine parity
- Fixes a bug/dependency
- Replaces an existing mod
I've yet to compare these on my machine, but I'll probably do a vote between Bobby and FarSight similar to Starlight-Phosphor once most mods of 1.18.2 are updated.
@TiboOpGithub That's a good point, and regardless of which mod I add, I will disable the chunk saving feature anyway. (As I wrote above, this was added to Bobby recently: Johni0702/bobby#25 ) But my current question is performance - does it decrease performance by any means? When loading cached chunks, when loading new chunks?
Oh I didn't know that! In that case, it's actually an improvement since in most cases your computer can load chunks before the server has yet to provide them. It certainly does not lower the performance and can be an added value in FO
Edit by Madis0: formatting
Hmm. Well, Bobby surely has the network benefit, but regarding the rendering itself, I'm now aiming towards #240 or equivalent, as those seem to provide a more tangible benefit for computers with lower performance. Maybe even alongside Bobby.
It's a very good mod that I use often but I don't think it fits in the modpack as it stores all chunks that take up a lot of space. After playing with the mod for half a year or so, I already have about 8GB of chunk data in the bobby folder. I think Farsight is a better option since it removes the chunks after playing and so it hardly takes up space but is also less useful.
@TiboOpGithub That's a good point, and regardless of which mod I add, I will disable the chunk saving feature anyway. (As I wrote above, this was added to Bobby recently: Johni0702/bobby#25 )
But my current question is performance - does it decrease performance by any means? When loading cached chunks, when loading new chunks?
From Discord:
I’ve used Bobby with this pack for a while now on a private realms server and I’ve had some performance problems with it trying to render large mountain ranges all at once when it’s caching them but other wise it’s been great 👍
Anyone else have tried it with latest FO releases?
Voting has ended, results are below!
Github: 4👍 2*👎 2😕
Discord: 8👍 0👎 7😕
Sum: 12👍 2👎 9😕
* Also counted the bug report #260
So 52% thought that Bobby is better than vanilla (together with other optimizations made in that beta), 9% thought it is worse and 39% did not see a difference.
Now it's time to see how Farsight goes against Bobby. The voting will end at the end of the next week, after which the mod to use will be decided.
(please don't vote on both Discord and GitHub, pick one)
To vote:
- Join a survival server in 3.3.0-beta.3. Set the rendering distance to 10. Explore it a bit, observe the performance.
- Join a survival server in 3.3.0-beta.2. Set the rendering distance to 10 and explore again.
- Now compare the two. Was the performance equal? Were there any glitches?
👍 3.3.0-beta.3 was more performant, had better FPS/network latency.
👎 3.3.0-beta.2 was more performant, had better FPS/network latency.
😕 Performance was similar or equal, did not notice a difference.
This vote has ended, scroll down for another one!
It is time for another public test, this time it's Bobby vs FarSight!
Both mods aim to improve render distances on servers by caching the chunks you visit and keeping them in display while you move further, regardless of the render distance set by the server. In other words, as long as you visit the distant chunks first, you can have any rendering distance on any server! In this test we'll see if the mods have equal performance on similar rendering distances.
Regardless of which one I'll pick, they will not fill up your storage space as FO will disable any permanent storage options.
Feel free to discuss the vote in this topic. This voting will end at the end of this or next week, depending on the amount of votes/feedback received, then FarSight will be added for another round.
To vote:
- Join a survival server in 3.3.0-beta.2. Set the rendering distance to 10. Explore it a bit, observe the performance.
- Join a survival server in 3.3.0-beta.1. Set the rendering distance to 10 and explore again.
- Now compare the two. Was the performance equal? Were there any glitches?
👍 3.3.0-beta.2 was more performant, had better FPS/network latency.
👎 3.3.0-beta.1 was more performant, had better FPS/network latency.
😕 Performance was similar or equal, did not notice a difference.
Farsight results are in!
Github: 0👍 1👎 1😕
Discord: 14👍 3👎 1😕
Sum: 14👍4👎2😕
70% of voters thought Farsight runs better than Bobby,
20% of voters thought Bobby runs better than Farsight and
10% of voters did not notice a difference.
Therefore, Fabulously Optimized will use Farsight from now on.
If you were one of the 20%, feel free to follow this tutorial:
https://github.com/Fabulously-Optimized/fabulously-optimized/wiki/Adding-more-mods
This isn't exactly fair. Voting in 2 places means some people will vote 2 times and others will vote only 1 time, skewing the results (considering the low amount of people who actually vote)
And some people on discord will just press 👍 without reading the message, maybe next time host it on discord only and use F and B emojis as reactions.
Thanks for the feedback, considering it for the next time.
Overall this isn't the place to discuss the meta of voting though, so you are also welcome to join our Discord if you'd like to do that.
This isn't exactly fair. Voting in 2 places means some people will vote 2 times and others will vote only 1 time, skewing the results (considering the low amount of people who actually vote)
And some people on discord will just press 👍 without reading the message, maybe next time host it on discord only and use F and B emojis as reactions.
We can verify on discord and on GitHub who reacted so there's no problem on running platforms, if some people vote on both we'll only count one.
And if people want to use dupe accounts to vote multiple times they can theoritically do it so it's flawed to but why would anyone do that ? To get their mod in FO ? Well we are not electing someone president, i feel taking our time to try to do too much control wouldn't be worth it.
For people pressing +1 on discord well same can happen on GitHub and it could still happen with your F and B technique and same point as right up, we trust the users for global testing, if users decide to random vote, well they are gonna get random perfs. Random voting is a disadvantage for everyone but again taking to much time for controlling everything would be a waste of time.
About only doing F and B reaction
Currently for testing we do compare 1st mod to vanilla (aka no tested mods), then second mod to vanilla
Then see which one outperform vanilla the most. This has the good point of being able to show if some mod is even worth it and comparing against a clear reference.
After you have read that just post any reaction so we can hide this cause it's not the good place to discuss it.
It is an interesting mod, but seems to have the same problems as #7 and more (huge caches). I imagine that being especially confusing in multiplayer minigames, where parts of the world are reused.
Edit: also heard claims about chunks not loading
Edit 2: if incompatibilities are fixed and Johni0702/bobby#25 exists, I can reconsider