Performance in the next pack?

[Bigglesworth]

I used to, actually, back in BC era You *CLEARLY* have never tried to code a Mod if you can make this statement as anything other than a troll. You also clearly have no blinkin' clue of what is going on in the code, and the differences between purely aesthetic changes (WoW addons) and code which is nearly as large as the original game (Buildcraft). Even if you set up an API to smoothly interface between the two, the sheer amount of code inherent in a mod like Buildcraft is a whole 'nother scale.

I've yet to see a Computercraft program longer than a few hundred lines. Try decompiling and deobfuscating a Mod and see how many lines is involved. You are impressed by something like a DPS Meter Addon for WoW. Someone who has never lived on an ocean coast is probably impressed by a large lake.

Besides, it's rarely the raw code that is the framerate lag, so much as the items it produces. For example, the sound and lighting updates a Timer produces every time it ticks. Dynamic Lighting requires tree-traversing algorithms, updating the light level for every block near the timer as it lights up for one tick, then re-calculating them as it turns off is what makes the Timer a potential lag-beast. You wouldn't experience things like that in a WoW Addon because the code is explicitly designed to prevent such. Plus WoW isn't running on Java.

You understand you’re furthering my point that MC needs an mature official API surrounding a game designed for extensibility. Also you missed the point, it seems, about why I mentioned DPS addons. It was not due to complexity, it was due to work the code is doing in relation to performance lost in framerates. Youre confusing things and ranting. When idle, these things (need) do very little to interact with the world besides saying "im here and ready, i am possibly emitting light", yet the performance lost is very great. So you can go on about amount of lines of code all day, but you’re missing the point in that complexity doesn’t equate to better. It equates to the enduser developing Tourette’s. I fully understand and respect the work (in large part to struggle with the API and changes every patch that break shit) that goes into the larger mods, but you seem to be comparing it against things like computercraft, instead of things like TukUI. At the end of it, Minecraft modding could be made much more concise and better preforming while retaining all of individual mod ability while improving stability if/when Mojang starts to actively make the API public (and maybe give us some real tool ontop of it but thats too much to hope for I think, eclipse is pretty darn good anyway).

By the same token you’re arguing about mod, you could argue about textures and why they have such a negative performance impact on the game, when a game like skyrim I can install 4096 textures with little to no impact. Of course they are different, and of course Minecraft renders things in a whole new way to create a world where everything is able to be broken and built, but why should that be the reason for it to stay that way? Its not impossible to imagine that the game might evolve past its original code to accommodate for such things it never originally intended while keeping its core sandbox gameplay regardless of all the excuse why it cant.

Your snipe about wow not running on java is irrelevant.

Aye. Having an API would solve nothing. We already have an API, that is what Forge is, that is what it does. Yes, it isn't "official", but that really doesn't mean much. It allows a mod to interface with Minecraft without having to alter the base code, that is it, that is exactly what an API is. There is nothing magical about that, and an official API from Mojang would be little different, functionally, than what Forge is doing now.

And yes, check the size of your mods folder. Then compare that to the size of your Minecraft jar, notice that the former is somewhere between six to ten times the size of the latter? Trying to compare that type of complexity to WoW add ons is beyond silly.

Again, the root of the problem is that each mod adds a bit of overhead, more loaded into memory, more CPU load in game. On its own, that impact is trivial, but compounded across 80 mods or more, what are you expecting? Nothing in life is free, and there is no magic wand you can wave here. You want more content in game, you pay for it in more burden on your system. That is the way it works, and it is neither Mojang's fault, Forge's fault, nor the mods's fault.

I agree it wouldn't mean much alone, Thats what I said Minecraft itself needs to allow for it officially. Meaning they would make active changes (likely) to the game itself so code is more directly integrated and not just another layer. Compairing the size of the folders as some sort of point though ... lolwut?? Bigger is better I guess?

I'm definitely in the "WoW add-ons do a lot less than MC mods" camp, and not just because they have an API. WoW mods are all just UI. You can't add new items, spells, abilities, professions, races, classes, mobs, world gen, and dungeons in WoW.

Even if you could there would be zero performance impact. Most of that stuff is simply stored in a database and just read, it is static for the most part. Im talking about what the add/mod is actively doing that requires a great amount of CPU power. Complexity doesnt mean better.


You're all starting to sound like you have some form of Forge Stockholm Syndrome. FSS. a terrible blight

 

[Iskandar]

Biggles, no offense, but stop trying to talk about things you don't understand. While it is true that there are things that could be done to optimize the way Minecraft does things, this isn't some kind of magic wand you can wave and all of sudden all the lag and slow downs will vanish.

Let me reiterate, the way Forge runs in game is EXACTLY how any official mod API from Mojang will function. Forge ALREADY optimizes Minecraft code, which is why it MUST be installed in the Minecraft jar. The only thing an official Mojang API will do differently is to already come prepackaged so you wouldn't have to install it manually, perhaps add some out of game functionality, nothing new or better would happen in game. You forget, unlike most other games, Minecraft's code can be changed directly, and Forge has done so. And no, you absolutely do not want mods to integrate with Minecraft directly. You'd get different mods trying to change the same bit of code, and you'd have incompatibility and crashes everywhere. We had that, before Forge got going, it wasn't pretty. You couldn't tell if two mods would work together, and most cases they wouldn't. Now, you can just drag and drop, and outside of block ID conflicts, you can pretty much guarantee any two arbitrary mods will work together. This is why you need that extra layer, that is what an API does. Yes it adds overhead, but that is NECESSARY not something you want to get rid of. Mods talk to the API, in this case Forge, Forge talks to Minecraft, everyone plays nice. That is the sacrifice you must make. You either choose easy compatibility between mods, or you choose speed, you CANNOT have both. And no matter how much you would like it to be different, that is the way it is.

Everyone calls Minecraft slow, and laggy. Thee are very valid reasons behind this. Yes, poorly optimized code is part of the reason, but the physics and rendering behind the game are the real reason. No other game has the freedom to change the world to any degree you see fit.

 

[Azzanine]

I used to, actually, back in BC era You *CLEARLY* have never tried to code a Mod if you can make this statement as anything other than a troll. You also clearly have no blinkin' clue of what is going on in the code, and the differences between purely aesthetic changes (WoW addons) and code which is nearly as large as the original game (Buildcraft). Even if you set up an API to smoothly interface between the two, the sheer amount of code inherent in a mod like Buildcraft is a whole 'nother scale.

I've yet to see a Computercraft program longer than a few hundred lines. Try decompiling and deobfuscating a Mod and see how many lines is involved. You are impressed by something like a DPS Meter Addon for WoW. Someone who has never lived on an ocean coast is probably impressed by a large lake.

Besides, it's rarely the raw code that is the framerate lag, so much as the items it produces. For example, the sound and lighting updates a Timer produces every time it ticks. Dynamic Lighting requires tree-traversing algorithms, updating the light level for every block near the timer as it lights up for one tick, then re-calculating them as it turns off is what makes the Timer a potential lag-beast. You wouldn't experience things like that in a WoW Addon because the code is explicitly designed to prevent such. Plus WoW isn't running on Java.

I was hoping someone with actual knowledge would set him straight.

With knowledge you confirmed my hypothesis that WoW plugins are doing a lot less so of course they are going to run faster on even sub par machines then a minecraft mod.
So to say "my comp can handle Nththousand WoW plugins why can't it handle a megre 80+ mods?! Must be the lack of API and not the scope of the code involved" is foolish.

 

[whizzball1]

Question, guys.

My processor is an AMD Athlon II x4 640 3.00 GHz processor. I'm trying to find the bottleneck in my framerate, since I'm only getting like... 40-60 FPS in good scenarios. It can get down to 20-30. By far the lowest FPS I have ever had in any game I've ever played on this machine.

You think 20 is bad?
I get 20 at best, and that's on a new world.

 

[ShneekeyTheLost]

You understand you’re furthering my point that MC needs an mature official API surrounding a game designed for extensibility. Also you missed the point, it seems, about why I mentioned DPS addons. It was not due to complexity, it was due to work the code is doing in relation to performance lost in framerates. Youre confusing things and ranting.

Absolutely wrong. Categorically and completely. DPS Addon basically surfs your chat logs for key phrases indicating dealing damage and looks for the number. It then keeps track of those numbers along with how long it has been keeping track of those numbers, then divides the first into the second. The code is mind-numbingly boring to write, but exceedingly straightforward. Compared to a mod like Buildcraft, it's 'baby's first coding experience'. You really have no bleedin' clue about this topic, why do you continually open your mouth and prove your ignorance?

When idle, these things (need) do very little to interact with the world besides saying "im here and ready, i am possibly emitting light", yet the performance lost is very great.

The 'possibly emitting light' is what is causing much of the lag, because of how Minecraft calculates dynamic lighting, due to the code structure that Java requires. To give you an idea... placing a torch in minecraft requires more computational power than every DPS Tracker in your entire raid. Now you have something that causes that to update constantly. THAT is what causes lag.

So you can go on about amount of lines of code all day, but you’re missing the point in that complexity doesn’t equate to better. It equates to the enduser developing Tourette’s.

So you disagree with me before you agree with me? I honestly have no idea what you are trying to say here. Complexity doesn't equate better, no. but it does impact CPU load, because the more lines are in the program, the more computing power it requires to run it. If you cannot grasp the concept that reading a multi-thousand line program takes longer than a couple hundred line program, then I really have nothing further to say to you.

I fully understand and respect the work (in large part to struggle with the API and changes every patch that break shit) that goes into the larger mods, but you seem to be comparing it against things like computercraft, instead of things like TukUI. At the end of it, Minecraft modding could be made much more concise and better preforming while retaining all of individual mod ability while improving stability if/when Mojang starts to actively make the API public (and maybe give us some real tool ontop of it but thats too much to hope for I think, eclipse is pretty darn good anyway).

Again, you misunderstand. I'm not comparing a Minecraft API to Computercraft. I'm comparing WoW addons to Computercraft programs, with regard to complexity and CPU load. Hell, they're even written in the same language.


In other words, one mod has the complexity of the entire WoW addon API. ONE mod. Think about that for a second. All the support for all the WoW addons in existance? One mod. And a fairly CPU-friendly mod at that, since it's mostly an interpreter. I'd make an 'ant, boot' comparison here, but I think it would be over your head. Suffice to say, you're missing a scale of several orders of magnitude here by trying to compare a mod to an addon.

By the same token you’re arguing about mod, you could argue about textures and why they have such a negative performance impact on the game, when a game like skyrim I can install 4096 textures with little to no impact. Of course they are different, and of course Minecraft renders things in a whole new way to create a world where everything is able to be broken and built, but why should that be the reason for it to stay that way? Its not impossible to imagine that the game might evolve past its original code to accommodate for such things it never originally intended while keeping its core sandbox gameplay regardless of all the excuse why it cant.

You cannot render textures the way Skyrim does and still have Minecraft be a game where every block is placable. The way Skyrim can handle the insane textures and have little to no impact is that it doesn't need to be able to render it dynamically, because objects are only interacted with in certain specific and scripted ways. In Minecraft? Not so much. That's the problem with a sandbox game. You have to render each and every grain of sand individually.

Your snipe about wow not running on java is irrelevant.

The mere fact that you can make this statement and apparently believe it speaks more for your ignorance on this topic than just about anything else I've ever seen.

I agree it wouldn't mean much alone, Thats what I said Minecraft itself needs to allow for it officially. Meaning they would make active changes (likely) to the game itself so code is more directly integrated and not just another layer. Compairing the size of the folders as some sort of point though ... lolwut?? Bigger is better I guess?

At this point, I'm not sure if you are being serious, or just trying to troll. I will, however, give you the benefit of the doubt and assume the former, and patiently explain this to you, AGAIN.

More lines of code = more CPU load to read the code. Reading 10 lines of code takes more CPU time than reading 1 line of code. Bigger does not mean better, but it does mean slower.

Even if you could there would be zero performance impact. Most of that stuff is simply stored in a database and just read, it is static for the most part. Im talking about what the add/mod is actively doing that requires a great amount of CPU power. Complexity doesnt mean better.

You're all starting to sound like you have some form of Forge Stockholm Syndrome. FSS. a terrible blight

And you are speaking out of ignorance on a topic you are apparently more ignorant on than my eight year old son, if you think that is all it is.

Believe what you like, I certainly won't be able to change your mind any more than anyone else has ever been able to change a zealot's mind. Go ahead, keep bleating 'four legs good, two legs bad'. It won't make it any more true.

 

[tatopolos]

IMO, a native API is always better, software industry wise. Why need someone else to do this layer (and I hope you agree with me on this, and not oblige me to list you all the drawbacks and issues coming with that), especially if you are willing to help your community. Less cost? at what price..
But lack of API may not be the main source of problems. But again, I see no reason why Mohjang doesn't make one for its community.

For the rest, you are both saying the same thing.. but too angry focused to see it..

and , yes, Minecraft is far from being the most optimized code.. and MJ should make their ugly looking cubical game less resources demanding than the latest far crys

 

[Iskandar]

IMO, a native API is always better, software industry wise. Why need someone else to do this layer (and I hope you agree with me on this, and not oblige me to list you all the drawbacks and issues coming with that), especially if you are willing to help your community. Less cost? at what price..
But lack of API may not be the main source of problems. But again, I see no reason why Mohjang doesn't make one for its community.

For the rest, you are both saying the same thing.. but too angry focused to see it..

and , yes, Minecraft is far from being the most optimized code.. and MJ should make their ugly looking cubical game less resources demanding than the latest far crys

See, that is the thing, and everyone does this, people look at Minecraft's simple blocky exterior and try to compare it to whatever the latest graphic intensive game is, and say "Well, I can run this game fast, why is Minecraft so slow?" Stop that, you are comparing apples and oranges. Far Cry and most other games are demanding graphically, Minecraft is demanding in a whole 'nother direction.. As much as it may fall short in some categories, Minecraft comes the closest at simulating real world physics in real time in a dynamic, interactive environment than any other game out there yet. You can explore nearly at will, instead of being limited to where a developer wants you to go, rearrange the terrain at will instead of being limited to deformable meshes, lighting is both dynamic and can be changed on the fly, liquids behave pretty much as you expect real world liquids to behave, and you can place and remove them nearly at will, gravity and physics are mostly present and accounted for and behave in a reasonable, if somewhat abstract, manner. Other games may do a few of those things, but at this time, none do all of them to nearly the same degree as Minecraft. Add to that the plastic nature of Java code which has allowed an unprecedented mod community to grow and thrive, and well, Minecraft remains a unique game that no one has managed to duplicate yet. Trying to measure performance metrics against something like Far Cry is more than just a bit silly, it is down right ignorant. Could Minecraft be optimized better? Certainly. Would that magically solve all the lag out there? Well, no, barring a rewrite from the ground up, and that is more than a little unfeasible. Even then, no matter what, Minecraft will always, always be a CPU and memory intensive game, and adding tons of mods on top of that are just going to make that worse.

Sorry about the rant, but it just irritates me when people judge Minecraft at face value, instead of looking at what it is doing under the hood. Sorry folks, but there is no magic wand that will suddenly let everyone run Minecraft at blazing speeds. If the bottleneck is your graphics card, then Optifine will help, but if you don't have the CPU and memory to keep up, nothing will change that much.

As far as an API goes, any benefits an official Mojang API would bring over Forge would almost entirely be out of game stuff. Forge has had several years now to refine their API, Mojang would be playing catch up. Hell, Forge has recently removed the one advantage Mojang could have brought to the table, allowing mods to be (mostly) independent of vanilla Minecraft updates. Add to that an official API would require mod authors to completely rewrite their mods to be compatible, and the turmoil that would cause in the modding community, I'm perfectly happy with the current state of affairs.

 

[quantumllama]

Modded Minecraft has performance issues because of mostly the same reasons why very large Dwarf Fortress .. well, fortresses die a FPS death. Too much stuff to constantly track and compute.

 

[tatopolos]

Sorry about the rant, but it just irritates me when people judge Minecraft at face value, instead of looking at what it is doing under the hood. Sorry folks, but there is no magic wand that will suddenly let everyone run Minecraft at blazing speeds. .

Iskandar, you should also stop thinking that you are the only person that likes minecraft, and knows its true value ... And no need to throw obvious statements as ultimate arguments,, everybody knows that performance issues are no easy matter, and it is not resolved by magic. very often, to make material performance leaps you need very deep changes of your soft/game...(and you seem ok with that).

Well, no, barring a rewrite from the ground up, and that is more than a little unfeasible.

Even far bigger Games and softwares had to do that, when ti comes to surviving. the bas news is that Minecraft is victim of it's success and is the cash cow for Mojhang. it means minimum investment to keep cash coming in.. i don't say it's evil, it just how it works..

I'm perfectly happy with the current state of affairs.

Things can always be better. Don't forget that!!

Anyway, very interesting discussion here..

and btw does any have more details about what have been changed/improved to have the FPS increase in 1.5?

 

[Quesenek]

I'm gonna try it but the forum threads posts dont look nice lol.

Edit it actually seems to be helping a lot I don't get Any world tick lag from spawning in to the server or from going back and forth from mystcraft worlds and my TPS is 20 now apparently according to the commands.

I have tried the tickthreading mod for a couple of days and have mixed feelings. On one hand I have Zero world loading lag and 20 TPS constantly. On the other hand it unloads the chunks that I have chunk loaded and I have run out of power more than once even being in the same area of the problem. I personally do not think it is done but it certainly works and it works very well.

 

[Gr3n]

Ok guys, lets just stop all these arguments right here. Slowpoke and many others on the ForgeCraft2 server have already said that there is massive improvement in performance in 1.5 because of forge optimizations. Don't believe me? Go watch Direwolf20's ForgeCraft2 episode 13. The answer to the question in the OP is yes, performance in 1.5 pack will be improved because of optimizations to rendering.

 

[DrRed]

Ok guys, lets just stop all these arguments right here. Slowpoke and many others on the ForgeCraft2 server have already said that there is massive improvement in performance in 1.5 because of forge optimizations. Don't believe me? Go watch Direwolf20's ForgeCraft2 episode 13. The answer to the question in the OP is yes, performance in 1.5 pack will be improved because of optimizations to rendering.

Beat me to it.

CPW has made it that instead of the ENTIRE texture sheet being called up every time there is a change, only the changed parts are pull up for 'much faster' performance.
I am so excited!

 

[brujon]

A lot of things can be improved in minecraft. Take chunks, for an example:
http://www.minecraftforum.net/topic...upportersnew-poll-please-resubmit-your-votes/
This forum post explains most of the things better than i could ever explain myself... But the gist of it is, the way minecraft handles worldgen is vastly unoptimized. In fact, it pretty much hasn't changed at all since inception, with a few developments here and there. And here's this bloke with a very well thought out idea for fixing it, with knowledge backing it. Why it hasn't been improved already, with all the time minecraft is in development and with all the money they're making?

I mean, come on. Higher, or even infinite build height is something that the community has been asking for ages.

Right now, they improved over one of my long standing complains which was the way minecraft dealt with rendering, and just look how much performance improved. A lot of other things aren't handled well in minecraft, or could be handled better, at least.
Particle effects such as rain, sound and light updates... Block updates, as well, such as when you start breaking a block and it updates to reflect how close it is to breaking. Minecraft has poor optimization. It doesn't do approximations, it doesn't cut corners, it does all of the calculations, and that *is* very taxing on the CPU. It is a problem which is exacerbated because it doesn't play very nice with threaded optimization either, so it doesn't utilize the full power of the multi-core CPU's that are the staple right now. But that is a problem which is common to many(or even most) programs, too. It also shifts most of the calculations to the CPU, instead of using the GPU. Many improvements could be made in many different areas.

[MassiveRant]
API would be nice? Yes. But as mentioned, Forge is doing a pretty good job so far. More pressing concerns.
Minecraft Realms? Damn awesome. Not relying on third party server providers or using all my bandwidth for the month in a couple of days setting up a server is a damn fine addition. Still doesn't fix performance, though.

But that is my opinion... They're focusing on that. No matter how hard i wish they focused on other things.

It's easy to explain, though. It's a business. Those are business decisions. What would make Minecraft sell more/make more money? Setting up an official (and controllable) modding environment and server-renting business (with monthly payments), or fixing code and performance issues that people obviously don't care enough to stop buying the game for?

Obviously, adding in more features (inane or awesome as they may be), keeps the community at large more satisfied than actual performance increases, especially since, with performance tweaks, the most impact is made on the community playing heavily modded minecraft, which, although it doesn't make up an insignificant part of the community at large, does not represent the majority. Most people who bought the game will never get into modded minecraft until Mojang adds in an official API and a way of downloading it in-game with buttons that glow and easy instructions and semi-official support. Which is what they're going to do. Most people playing minecraft as it is right now don't play on servers, either, unless it is one set up by a knowledgeable friend or a lan environment. Adding in an API will please BOTH groups, making more money for them in the end. What's not to like (for them)?

Those are the right business decisions - making those two improvements a priority is something that will make Minecraft sell more copies, and make them yet more money in the form of server rentals with official support. Bugfixes are ALWAYS a priority, but fixing things that aren't broken, but work unoptimally? That is just a bonus. That can wait. That costs money but doesn't make that much more money. Always room for a Minecraft 2 with all the performance increases in the world . I may not understand all the coding and programming very well, although i do know a little here and there, but i do understand money, and that is what makes the wheels turn.

I mean, i can complain all i want about how x64 is something that is available since forever, and how native x64 programs run so much more smoothly with all the ram and multi core processing power available, and that didn't change the fact that we still live in a 32bit world, and even the new consoles are still going to be doing stuff that way. Or how the focus has dramatically shifted from PC optimization to console optimization, and that most games are poor unoptimized console ports that run poorly on a system that could handle so much more. Didn't change a thing. Because consoles make more money. Because companies don't want to change over when everything works well enough. Mojang is a company...
[/MassiveRant]

 

[Bigglesworth]

It's easy to explain, though. It's a business. Those are business decisions.

http://xkcd.com/1032/

 

[brujon]

http://xkcd.com/1032/

 

[PhilHibbs]

I mean, come on. Higher, or even infinite build height is something that the community has been asking for ages.

Wouldn't that make the "clear sky" calculation (referred to internally as "can lightning strike", I think) more expensive?

 

[brujon]

If clear unobstructed line of sight to the sky is needed for lightning to strike, then no matter how high the build height there will always be the same number of blocks in a given area where lightning is able to strike. Granted, the vertical area that will need to be scanned in order to ascertain whether or not there are blocks above a certain area during a thunderstorm is going to be increased, making lag a possibility with infinite build height, but then again, that can be changed and limited, such as making it so that the cloud layer is at a fixed Y-level range and no lightning can strike above that area. Also an incentive for building high, because lightning is a nuisance, and also makes sense, since lightning comes from clouds. Building above the clouds means no lightning... Thus the max number of vertical blocks needed to be scanned can be limited to a manageable degree. Cubic chunks, basically what makes infinite build height a possibility, would also help with this, since at any given moment you could be at, say, Y level 1024, and not all chunks below you would be loaded all the way down to Y level 0, so that all blocks below a certain Y level that is not loaded would simply not factor into the calculations for where lightning could strike, making for less blocks involved in the calculation, easing things for your computer.

If you read the thread, you will see that this was already taken into consideration to some degree... With how Sunlight is dealt with in Minecraft. Same as Lightning storm, Sunlight also checks for obstructions, and it also takes into consideration if those obstructions are opaque or clear blocks, and if they filter light or not. The lighting system in minecraft is very laggy. In an infinite build height environment, lighting lag is much more a concern than rain or lightning lag, although both are also things to be concerned about. The "heightmap" system suggested in the OP of the forum post i linked goes over some of that. What works for making Sunlight work without a hitch in that environment could also be used to make rain and thunderstorms work as well. It's... Pretty clever.

EDIT: It's also so possible that there's a mod (very old mod), that does just that, but limits the build height to about 65k blocks. Link: http://www.minecraftforum.net/topic/310298-100robintons-mods/#CubicChunks . With access to unobfuscated code and engine rewrites to accomodate the new system, i'm sure everything would work out much, much better, especially with all the other improvements that have been made since 1.00.

 

[tatopolos]

I mean, come on. Higher, or even infinite build height is something that the community has been asking for ages.

Yeah.. this might bring a lot of new discoveries.. a 6th or 7th continent.. may be we can confirm finally the the overworld is spherical.. can't wait

 

[brujon]

What makes me most excited about Cubic Chunks is how everything could be improved. From terrain generation to lag... Infinite or virtually infinite build height is awesome and all, but... Functional and awesome underground/skyland biomes, performance increases, even higher possible view distances... Damn.

 

[Bigglesworth]

I doubt infinite sky will ever be a thing. The game works differently than a game like, say, EVE. I could see 65K sky heights and going to the moon or local worlds in a Minecraft Solar System or something. Might be interesting to make a 2K block radius round world, or shoot a creeper into a 20k round world sun.