Performance in the next pack?

[ex13]

Point 3 is the tricky one. The world above and beneath you needs to be loaded so that the game knows what blocks are above and below you. Imagine someone is above you in the sky and between him and you is an unloaded chunk. Now he places a bucket of water. Without the blocks in between the water will never reach you. But if the game would know which blocks are between you it could calculate the water-flow without updating nearby blocks or without updating anything else (not even it self) in these chunks. But with the actual engine something like this is not possible. A chunk is either loaded or not, there isn't something in between, which makes the structure of the chunk loaded, but nothing else. For Minecraft in it's current state a not loaded chunk simply does not exist. It may be saved somewhere but it can't change and can't be used for anything. If a giant floating island is above you but isn't loaded you wouldn't even see that island.

I think thats exactly what the cubic chunk thread proposes: Unloaded chunks simply don't exists, you can't see them and entities in there don't do anything. For your flowing water example, it can also be done right now: Two players (or chunk loaders) a few hundred blocks away so that there are unloaded chunks in between, an aquaeduct in between which decreases in height every 7 blocks, on player puts water at the one end - and it will never reach the other player. Its the same as vertically with cubic chunks.

I don't know any of Mojangs plans, if they indeed want to change this to also simulate the world in between, they definitely need to change more than I expected. But the original cubic chunks thread linked here simply ignores this issue.

 

[Bigglesworth]

Macrocosmic physcis doesn't work on statistics, it works on actuality. If I can see the sun, then it means there is definitely not something blocking it. It isn't based on a probability of there being a cloud or a moon or a giant spaceship, it's based on whether there's actually something there. The game would have to check, not guess.

What there is a million blocks "over there" is a lot less relevant to me than what's "up there". If the game started off drawing a blue sky with a sun based on the "probability" that there isn't a huge floating island (or a player-built frame machine) above me, but then started drawing a huge floating island a million blocks above me that then blocked off my previously-working solar panel farm, then that would not be acceptable.

It would only be checking the surface. What is under the 'skin blocks' is governed by the framework and needs a whole hell of a lot less rendering data untill uncovered (uncovering would take huge amounts of time) Not sure you read what I said or I didn't type it very clearly.

 

[Bluehorazon]

I think thats exactly what the cubic chunk thread proposes: Unloaded chunks simply don't exists, you can't see them and entities in there don't do anything. For your flowing water example, it can also be done right now: Two players (or chunk loaders) a few hundred blocks away so that there are unloaded chunks in between, an aquaeduct in between which decreases in height every 7 blocks, on player puts water at the one end - and it will never reach the other player. Its the same as vertically with cubic chunks.

I don't know any of Mojangs plans, if they indeed want to change this to also simulate the world in between, they definitely need to change more than I expected. But the original cubic chunks thread linked here simply ignores this issue.

Yes it is the same, but situations which have problems with this are far rarer on this axis. The best example is sunlight, or anything with gravity like gravel, sand or normal entities. If you drop sand from your world it will jam in between. So you can drop 200 blocks of sand down, which will jam this block, because the blocks can't pass the unloaded chunk. Once it gets loaded all the 200 blocks appear at once and create a 200 blocks high tower, which might be higher than the initial drop-height. Or the other option is they would not drop into unloaded chunks, but accumulate in form of a tower at the bottom of the last loaded chunk. I don't think either of the possibilities would really make sense. You could make it a bit easier by just loading chunks upwarts. And you don't actually need to perform any updates in this chunks, like running machines, but still it would require a major engine-change or even better not using java.

 

[Iskandar]

For those pointing at Robinton's Cubic Chunks mod and saying "See, it can be done!" there are a few caveats. One, while it did work, it never worked 100% correctly. Lighting, especially, was glitchy as hell And the expected performance increase was never realized, either, lag was actually worse in a lot of cases.

Not to say I wouldn't love to have that kind of freedom, mind you. I kept an eagle eye on that mod for quite some time, until it stopped updating. In the end, I suspect the current game engine simply isn't suited for it.

This game was created by one person, and then changed, upgraded, added to, and tinkered with for four years now. I'm fairly sure there is a lot of legacy code that was good enough when it was first written, but is desperate need for a rewrite now. What really needs to happen now is "Minecraft 2", basically looking at what we have now, what needs to be done better, and what needs to be added, and just doing a ground up, start from scratch, rewrite. Doubt it will ever happen, though.

 

[PeggleFrank]

I play ultimate with Optifine and the default texture pack and I get 10 FPS. I've adapted to the low framerate, but I noticed in something like Tekkit that's really old, I get up to 200 FPS easily, and in old vanilla I get really low framerates. Also, I get around 200 FPS in 1.4.7 vanilla, but when I do the ultimate pack my FPS drops significantly for some reason. It might just be some kind of overload of items. Should this be able to run ultimate with atleast a solid 60 FPS?

Oh dear god. This has turned into a tech support thread.

 

[DrRed]

Don't worry, 1.5.x Forge is *much* faster

 

[brujon]

For those pointing at Robinton's Cubic Chunks mod and saying "See, it can be done!" there are a few caveats. One, while it did work, it never worked 100% correctly. Lighting, especially, was glitchy as hell And the expected performance increase was never realized, either, lag was actually worse in a lot of cases.

Not to say I wouldn't love to have that kind of freedom, mind you. I kept an eagle eye on that mod for quite some time, until it stopped updating. In the end, I suspect the current game engine simply isn't suited for it.

This game was created by one person, and then changed, upgraded, added to, and tinkered with for four years now. I'm fairly sure there is a lot of legacy code that was good enough when it was first written, but is desperate need for a rewrite now. What really needs to happen now is "Minecraft 2", basically looking at what we have now, what needs to be done better, and what needs to be added, and just doing a ground up, start from scratch, rewrite. Doubt it will ever happen, though.

Yeah, but... It was done for a beta version of minecraft, by only one person, who only had access to obfuscated code. Lots have changed since then. I believe it can be done much more effectively now, even if it's made into a mod rather than a game rewrite. In any case, most of my criticism is directed towards the fact that Mojang, and not just Mojang, mind you, but any game company, prefer to focus on adding inane but "cool" features instead of focusing on actual game improvement and polishing, because it's been proven over time that the average gamer is happier with new features that make the game look fresh, rather than with engine rewrites, code updates and the like that improve the overall experience. It's also because doing so would mostly benefit the modding community, since vanilla has already achieved a point where lag is pretty manageable outside of really crazy redstone. Games are no longer made to be perfect on release, they're made to be good enough.

 

[DrRed]

Yeah, but... It was done for a beta version of minecraft, by only one person, who only had access to obfuscated code. Lots have changed since then. I believe it can be done much more effectively now, even if it's made into a mod rather than a game rewrite. In any case, most of my criticism is directed towards the fact that Mojang, and not just Mojang, mind you, but any game company, prefer to focus on adding inane but "cool" features instead of focusing on actual game improvement and polishing, because it's been proven over time that the average gamer is happier with new features that make the game look fresh, rather than with engine rewrites, code updates and the like that improve the overall experience. It's also because doing so would mostly benefit the modding community, since vanilla has already achieved a point where lag is pretty manageable outside of really crazy redstone. Games are no longer made to be perfect on release, they're made to be good enough.

That's just because the 'average gamer' doesn't know what they want. If you explained the inefficiencies to them, they would definitely prefer a bugfix/optimise update vs 'The Scary Update'.

 

[DrRed]

I feel like I'm cheating on my current 1.4.7 world...At the moment I have gone off FTB playing for the sole fact that I am going to be making a new world in a week anyway with the 1.5 pack. I have already started to compile a 1.5 pack myself just to see if the FPS increases are more than a myth.