Vanilla Minecraft News Discussion

[SterlingRed]

Yep. He also said he'd tried different video cards in case that was the reason. I think he's trying hard to fix the issue, but can't reproduce it.

Yup, and that's definitely an advantage he has as a professional over me when I run into such problems. I've got one system, one video card, one internet connection, and somehow have to figure it out from there :)

My apologies for jumping the gun on my interpretation there. Good reminder for me to keep the pessimism in check. Glad it's being looked into.

[finite8]

OMG. The lag is horrendous! That machine i complained about before that replaced my farm (machine gun farm), if i run it on any level of smooth lighting, the game turns into the most stuttery piece of shit... and that is on a SMP so the redstone logic isn't even being handled locally. I normally cruise at 80fps+ and it becomes unplayable with smooth lighting.

Damn right they optimized it. They optimized it so much, it's like it's not even there now.

[FlowerChild]

OMG. The lag is horrendous! That machine i complained about before that replaced my farm (machine gun farm), if i run it on any level of smooth lighting, the game turns into the most stuttery piece of shit... and that is on a SMP so the redstone logic isn't even being handled locally. I normally cruise at 80fps+ and it becomes unplayable with smooth lighting.

Damn right they optimized it. They optimized it so much, it's like it's not even there now.

You know, it makes me wonder if they aren't seeing this not because of video-card/system but rather because they don't have any redstone builds in their test worlds.

As I've gone on about potentially removing in the past, redstone devices tend to emit light when on, but not when off, which causes excessive lighting updates for no good reason. Smells rather fishy to me.

[Sarudak]

Maybe someone needs to send them a copy of the world they're getting bad FPS in?

[FlowerChild]

Maybe someone needs to send them a copy of the world they're getting bad FPS in?

Or tell them to, you know...play their own fucking game :)

If all this is much ado about them not having any redstone builds in their worlds, especially after developing the "redstone update", well then...

[finite8]

Maybe someone needs to send them a copy of the world they're getting bad FPS in?

Or tell them to, you know...play their own fucking game :)

Spoiler

Show

[Sarudak]

Or tell them to, you know...play their own fucking game :)

If all this is much ado about them not having any redstone builds in their worlds, especially after developing the "redstone update", well then...

Come on man we can't have such unreasonable expectations about game developers.

[FlowerChild]

Come on man we can't have such unreasonable expectations about game developers.

I know man...I know. I also try to avoid assuming people in other professions should be willing to sample their wares.

Take chefs for example. I'm personally way more comfortable eating at restaurants if I know the chefs won't go anywhere near the food they prepare.

[BlueRavenGT]

I worked out the UI for placing chests a while ago.

[FlowerChild]

I worked out the UI for placing chests a while ago.

I said it's tricky, not that it isn't doable. I'm personally not a fan of how you did it there, which is an indication of how tricky it actually is ;)

[BlueRavenGT]

Any suggestions about how I could improve it?

[FlowerChild]

Any suggestions about how I could improve it?

Well, I personally don't think hard and fast auto-connect rules like that are the way to go about it, as they're not necessarily intuitive for the player, potentially resulting in a lot of breaking then replacing chests to get it right.

Personally, I'd favor a system where the player specifies through their own actions whether they want a chest to connect or not and to which one. Honestly, given the current interface conventions in MC, I'd probably lean towards shift clicking on a chest to connect to it while placing another, and that's probably the route I'll take if I decide to tackle this one myself.

[BlueRavenGT]

Ok. I actually did this before shift clicking was implemented in vanilla, and hadn't considered that. :)

[MoRmEnGiL]

One thing I really hate is that lighting updates in redstone devices.

Need a clock? Tough luck, choose between fps tanking repeater/torches, or fps tanking pistons :P

Worse thing is, making it light level 15 around the clock doesn't necessarily help.

EDIT: Insanity:

[FlowerChild]

One additional bit of irony that I'd like to point out in this is Dinnerbone's old and repeated statements about how what they had coming would be "better than Optifine".

Ahem...yeah.

[Ulfengaard]

One additional bit of irony that I'd like to point out in this is Dinnerbone's old and repeated statements about how what they had coming would be "better than Optifine".

Ahem...yeah.

More hype = less game. That's not universal, but true often enough in my experience.

[MoRmEnGiL]

Well I don't doubt they want to bring something better than optifine to the table (or at least wanted to), but I'm a bit puzzled by their apparent inability to sort the lighting system out. I don't think they are that bad in actual coding, which means that it is a matter of time allocation and priorities. Which is puzzling. Fix your game mojang!

[FlowerChild]

Well I don't doubt they want to bring something better than optifine to the table (or at least wanted to), but I'm a bit puzzled by their apparent inability to sort the lighting system out. I don't think they are that bad in actual coding, which means that it is a matter of time allocation and priorities. Which is puzzling. Fix your game mojang!

My opinion on it is that they are focusing on the "proper" way of doing things in terms of structure, rather than the optimized way, and are of the mistaken impression that will somehow sort itself out and result in good performance.

Notch was definitely aware of optimization being what makes games possible. I don't get that impression about the current team, especially with some of the pie in the sky ideas they have expressed with regards to the mod API.

Programmers do not equal game programmers man. It's a very specialized skill set, and one I am beginning to despair the current team doesn't possess. At present, I see them rewriting large swaths of Notch's code, and it makes me extremely nervous, as I think they've yet to deliver on any of their promises of improved performance, while hurting it on more than one occasion (1.3 being a prime example).

Anyways, just my two cents as to how my opinions are shifting with time.

[MoRmEnGiL]

Well, notch may have made one or two mistakes (one of which led us all here ;] ) but I cannot remember him being lax with such issues. He even gave up one of his weekends working on fixing stuff, when the issues were big with the current release.

Imho, 1.3 is when all hell broke loose. Wish they never merged ssp with smp in the first place.. It has only gone downhill from that release on..

[FlowerChild]

Yeah, Notch knew what he was doing, there's no question there. Sure, there are a few things I didn't agree with him on (like wolves), but I always respected the shit out of him and was perfectly comfortable designing my mod on top of his work. In other words, I trusted him.

It's just not the case with these guys, and that's extending to more and more aspects. Previously I thought it was just a matter of them not being game designers, but I'm not so sure that's the extent of it anymore.

Anyways, as usual, just further motivation for me to get the fuck out of Dodge :)

[dawnraider]

I am so glad that you are actually competent (far more than that) in game design and keep turning out awesome stuff like HCH. I honestly would have left MC back in beta 1.8 and never looked back. I didn't like that update (mostly from terrain changes. Why did they change what wasn't broken. This also seems to be a recent (year long)trend they have followed) , and while 1.0 may have had some cool stuff, that was when it all went south. So thank you very much for keeping me on board with such a great game (which I effectively consider you the designer, and Mojang the fucker-uppers). I have thoroughly enjoyed playing BTW and I know RTH will be even better.

[FlowerChild]

Ok, someone mentioned on MCF that 1.6 is some kind of "mob update". Has anyone seen any confirmation of this from a reliable source?

If witches and bats are any indication, this makes me extremely nervous.

[TheGroovyWorkshed]

There will be new mobs in 1.6, but not 1.5

From Jeb's twitter.
Source: https://twitter.com/jeb_/status/304301377818066944

I'm hoping this doesn't mean they're going to 'improve' mob AI again. Who am I kidding?

[FlowerChild]

From Jeb's twitter.
Source: https://twitter.com/jeb_/status/304301377818066944

I'm hoping this doesn't mean they're going to 'improve' mob AI again. Who am I kidding?

Fuck me.

I think I was holding out lingering hope that it might just be a rumor.

Can anyone think of a nice safe area of the MC design that we can all start bitching at Mojang about needing attention to try and distract them from working on more critical aspects? I might have thought biomes or something, but after they went and coated the world in a ridiculous number of jungles, I'm not so sure that's safe either.

[magikeh]

To be noted Jeb has spent some time on optimizing the new lighting system and on my lappy there is a difference. (in the good way) the fps has gone up, there are no more lag spikes, block lag and my cpu fan is a lot slower ;) But what really gets me, is why the fuck was this not in the actual release?? Why did it take them releasing the update to the world to fix this one damn thing!

Jeb's twitter for reference:


Back at work... Slept poorly tonight =( The whole day I'll spend on adding metrics to Minecraft to track the smooth lighting problems :/

First test show nearly no difference between smooth "minimum" and "maximum", but it's 10 times faster to have it turned off...

Finally changed something that made average time to tesselate block with smooth lighting from 3.2 to 2.6 μs. Flat light still at 0.3 μs...

Got the average time down to 1.8 microseconds, going to create a test build for people on the bug tracker

Ok, it turned out the problem was the hotspot optimiser. We had too many paths (?), splitting into 2 methods halved the calc time to .6 µs

I've uploaded a temporary snapshot for testing performance. If you have time, please give it a try: http://pastebin.com/yAQV493W

Edit: Clarity