1.1 should be an optimisation update

[bartekkru99]

Almost everyone who uses mods, can agree with me, that game is running too slow and crashes too frequently, maybe it's time for us to focus on more technical side of development, rather than new content and gameplay, for that one update? I know, I know, there are problems with Unity, but does it apply to everything, that make game run like Crysis 3 on pc from 2002?

[Kelderek]

Well, you can assume that eventually there will be a conversion to Unity 5 and presumably there would be many optimizations done then, but that may not be possible in 1.1, it might be later on. It wouldn't make much sense to focus too heavily on optimizations until that conversion is done, though some might be possible. Optimization tends to happen along side of bug fixing - they usually try to do both in every update to some extent.

[Superfluous J]

Talking about what should be in 1.1 when we haven't even seen 1.0 seems - at best - premature.

[bartekkru99]

Talking about what should be in 1.1 when we haven't even seen 1.0 seems - at best - premature.

But we know, what going to be added in 1.0...

[Superfluous J]

But we know, what going to be added in 1.0...

I would appreciate a link to where they said they were not going to optimize the game.

[Gkirmathal]

I would appreciate a link to where they said they were not going to optimize the game.

5th, personally I agree with the OP.

But what you are asking is never really discussed in any of the DevNotes, by any of the devs. Only thing I ever read are, vague notions, about some bugs they are fixing. But not which ones and what those fixes reflect vs performance/stability.

Sorry if I sound a bit sour.

But after 2 years of following and withholding myself from getting KSP, because I knew the issues the game had pre 0.90. I hadn't realized they were as bad as they are when you start adding more content (albeit from mods, or eventually new stock added content).

A bit OT and a bit of a rant, 2nd sorry.

I now have my 'perfect' mod setup, cut down on parts, reduced textures + dds/ATM. Even then it boggles, irritates me, that the ram profile is constantly creeping up, instead unloading on scenery/instance changes.

This is even without Karbonite, which I still want, but I do not dare to install more stuff because of the above.

So I'm, a bit worried about 1.0 stuff they want to add in stock and the lack of info on bug/performance fixes.

[regex]

1.1 should be optimization? No, 1.0 should be optimization. They're essentially releasing the game as "production-ready" when it can't manage its memory. They're even adding new parts and systems to this "production-ready" software that can't manage its memory.

But what you are asking is never really discussed in any of the DevNotes, by any of the devs. Only thing I ever read are, vague notions, about some bugs they are fixing. But not which ones and what those fixes reflect vs performance/stability.

To be fair its recently been better than "vague notions", but I do agree that more detail on bugfixing would be better.

[egoego]

1.0 might contain enough optimization. From my experiance the stock game without mods does not have a memory problem. (To give you a measurement, the biggest thing I had in space was a 500t, 1000+ part space station, assembled during 20+ launches.) I play the 32bit windows version on highest graphic settings and don't delete debris and never ran out of RAM.

Maybe, when one has more memory used by his mods, than by the stock game, he should not blame the stock game for his memory problems.

[regex]

Maybe, when one has more memory used by his mods, than by the stock game, he should not blame the stock game for his memory problems.

You're absolutely correct, sloppy memory management and memory leaks are completely excusable because the stock game plays just fine, and no one ever mods a game that specifically caters to modding made by a company that actually celebrates and hires, even part term, modders to add stuff to their game.

[McPoisoned]

I personally think that except for the lack of ram (which is a 32bit problem and not a game problem) the game runs fine even when using a lot of mods (assuming you have active texture management and a decent setup)

[Gkirmathal]

Maybe, when one has more memory used by his mods, than by the stock game, he should not blame the stock game for his memory problems.

The memory problems also occurs on stock, be it after more time. Few hours on stock, rather than within 120min at best when modded. Latter is on manually limited part packs and reduced textures.

Mods only enhance, like a magnifying glass, this issue which is present in stock but less visible.

To say the game is not to blame, is like saying Toyota is not to blame for their faulty elc handbrake failures on the first gen Prius's, occurring in 1/1000 cases, because drivers probably loaded their groceries wrongly for the COG so they should buy less groceries.

[Kiro]

I would much rather see optimization and bug fixes in 1.0 than any other features or add-ons. I love KSP but without the mods it not quite the same. Especially the beautification mods. I have an incredibly beefy PC, able to run the newest games in ultra settings and its so sad to see it come to a crawl in KSP when flying bigger ships or stations. Now that Unity 5 is out, I really REALLY hope that more resources are used to optimize this game. Lack of even simply hardware support/utilization is really not something that should be part of an official release. Additionally, even though I installed Linux 64 so that I can get the most use of this game, lacking full 64bit support in Windows in 2015 is rather laughable.

[Superfluous J]

5th, personally I agree with the OP.

But what you are asking is never really discussed in any of the DevNotes, by any of the devs. Only thing I ever read are, vague notions, about some bugs they are fixing. But not which ones and what those fixes reflect vs performance/stability.

While true, it doesn't change the fact that we don't know. 1.0 could be rock solid optimized, or it could be a mess. My money's on that it'll be somewhere in between and some people will say it's a mess and others will not understand where they're coming from.

IF it is not properly optimized, I will consider optimizations of primary import. But (like regex) I consider them of primary import NOW. I hope that Squad understands that releasing a buggy, crashy, slow game as a 1.0 release is a supremely bad move and if they don't, I hope they'll read this post and agree with me. If not, then I don't know if a 1.1 "bugfix/optimization" update will save the game.

Sorry if I sound a bit sour.

...

A bit OT and a bit of a rant, 2nd sorry.

I now have my 'perfect' mod setup, cut down on parts, reduced textures + dds/ATM. Even then it boggles, irritates me, that the ram profile is constantly creeping up, instead unloading on scenery/instance changes.

This is even without Karbonite, which I still want, but I do not dare to install more stuff because of the above.

So I'm, a bit worried about 1.0 stuff they want to add in stock and the lack of info on bug/performance fixes.

We agree 100% on this. I installed a memory monitor mod so I could be warned when the game was in imminent danger of crashing. I can play 90-120 minutes before that happens on my current favorite mod setup (My LP series). What I don't agree with is thinking forward to 1.0 when they're still telling us things they're doing in 1.0 that we didn't even know about.

And (to reiterate) I also think waiting for 1.1 to optimize is a terrible idea.

[Kerbart]

1.1 should be optimization? No, 1.0 should be optimization. They're essentially releasing the game as "production-ready" when it can't manage its memory. They're even adding new parts and systems to this "production-ready" software that can't manage its memory.

Although bugs seemed semi-acceptable when the game was in beta, the proven mantra "fix bugs first, then add features" never seemed to have been part of the development strategy.

I always thought the number of features added should be an asymptotic curve nearing 100% as you get closer to Gold. Instead it seems to be an exponential curve with KSP. Every version comes with an ever expanding number of features. I stepped in around .20 I think, looking at the wiki the features ever since were:

• .21: more parts, interface improvements
  
• .22: science, career mode, biomes
  
• .23: tweakables, and tweaks to science
  
• .23.5: asteroids, new parts
  
• .24: new parts, contracts, budgets, interface improvements
  
• .25: new parts, explosions, career mode enhancements, navball enhancements, destructable buildings
  
• .90: upgradeable building, crew experience, enhanced career mode, fine print, overhauled mk3 parts, biomes on every planet, editor overhaul
  
• 1.0: revamped atmospheric modeling, re-entry heat, procedural fairings, rebalanced parts, update to engineering

It seems to me the amount of “large features†seems to increase with every release instead of decrease. Personally I'd think that atmospheric modelling alone would be worth the 0.91 release, engineering and other gameplay features a 0.92 release, then rebalancing the parts in 0.93 through 0.95 (because rebalancing is hard and you'll need feedback from the community and a couple of tries to get it perfect) and then and only then I'd start to think about release candidates to squash the last bugs before issueing version 1.0

I almost get the feeling that Squad feels they're running out of time and are trying to rush the product out.

[bartekkru99]

I also think, that Squad feels uncomfortable with few years old game still in early access, old players will stay for a long, time, but won't buy the game second time.

And back to the main topic, I've noticed, that KSP is getting more unstable with each update. I was using Astronomer's visual pack since it's early versions, without serious performance issues, but now in 0.90 even with standard EVE my game runs on about 10fps, I also got used to B9 and KW, because game was crashing every few scenery changes (from VAB to space center, or from the tracking station to the vessel), despite the fact, I'm using forced OpenGL and ATM. oh vand what do you think, is using strange model formats (.mu/.mbm) make the game slower? because seriously, nobody uses extensions like those, after searching .mu in google, there were only KSP forum threads...

[Renegrade]

I always thought the number of features added should be an asymptotic curve nearing 100% as you get closer to Gold.

Strongly agree. Classically, there's usually a feature freeze long before anything is published to the public at large, so that the focus can be entirely on bugfixing. New features means new regressions - not something you want in a shiny '1.0'-styled release. :/

Personally I'd think that atmospheric modelling alone would be worth the 0.91 release, engineering and other gameplay features a 0.92 release, then rebalancing the parts in 0.93 through 0.95 (because rebalancing is hard and you'll need feedback from the community and a couple of tries to get it perfect) and then and only then I'd start to think about release candidates to squash the last bugs before issueing version 1.0

That's pretty much how I'd do it (assuming I did a release at all - I'd probably do something more like the "LTS" packages that Linux distros do these days). Not only would it help isolate the source of any new bugs, but it would also step up the release pace. And feedback on balance changes are critical, as sometimes unexpected and serious side effects occur (such as the rise of ion landers after the ion buff).

I almost get the feeling that Squad feels they're running out of time and are trying to rush the product out.

There is definitely some sort of 'rushed' feeling here. It boggles my mind that there was an entire, single, solo 'beta' release..

[juanml82]

I know very little about game design, but if every game used Squad' system for managing textures we wouldn't have the graphics we have today. There is something regarding memory that every single game developer is doing and Squad is not. Most likely, not loading every texture into RAM at the beginning of the game and keeping them there, whether they are used or not. And we have nothing from the devs in this regard.

Even more, there is no reason why a 500+ parts ship which isn't under any force, either external or internal, should be calculating physics for every part every frame. If it's in orbit and all engines are off, the results of the physics check in frame 1 will be the same than in frame 2, 3, 4 and frame 8239047892387. And with something exerting some force on the vessel, there is likely a lot of optimization possible: if the ship won't bend because you fire the engines, why check what's happening to every single part there? Just temporarily weld the parts and be done with it.

[edrobotguy]

It seems to me the amount of “large features†seems to increase with every release instead of decrease. Personally I'd think that atmospheric modelling alone would be worth the 0.91 release, engineering and other gameplay features a 0.92 release, then rebalancing the parts in 0.93 through 0.95 (because rebalancing is hard and you'll need feedback from the community and a couple of tries to get it perfect) and then and only then I'd start to think about release candidates to squash the last bugs before issueing version 1.0

I almost get the feeling that Squad feels they're running out of time and are trying to rush the product out.

Hear, hear. There is no rush involved. KSP is damn near unique in this regard. Any other game having been in Alpha for this long would have been considered in "development hell". Instead, Squad has been making money and building a customer base and volunteer developer base for years, long before even going to Beta. In fact, if Squad is still planning on adding features then the game is still in Alpha.

DO NOT RUSH THE PRODUCT. Ubisoft rushed Assassin's Creed Unity out the door way too early, full of bugs, and has done huge damage not only to its bottom line but also its reputation.

Pushing a buggy 1.0 out the door will kill the golden goose.

Do it right, admit it's still in Alpha, and continue development of new features up to 0.95 something along the lines of what Kerbart suggests, then call 0.96 to 0.99 the Beta tests. Use those only to squash bugs and optimize, not adding anything new to the game. Then when 1.0 finally is released, it will be polished, fast, and almost entirely bug-free.

And then Squad can start making the real money. Right now, all you can buy from Squad is the game itself, T-shirts and coffee cups, a 3d printed kerbal, and that's about it. If 1.0 is a 4fps game when building a Minmus base with lots of parts, revenue will die. If instead the game runs fine even with a 1000 part base, then you're looking at KSP: the animated series and KSP: the movie and KSP: the lunchbox and KSP: the breakfast cereal and plush KSP dolls ... don't believe me? Who are Leonardo, Donatello, Michaelangelo, and Rafael?

- - - Updated - - -

Even more, there is no reason why a 500+ parts ship which isn't under any force, either external or internal, should be calculating physics for every part every frame. If it's in orbit and all engines are off, the results of the physics check in frame 1 will be the same than in frame 2, 3, 4 and frame 8239047892387. And with something exerting some force on the vessel, there is likely a lot of optimization possible: if the ship won't bend because you fire the engines, why check what's happening to every single part there? Just temporarily weld the parts and be done with it.

Agreed. If a Kerbal is in focus, then vessels in the vicinity should each just be one big collision mesh, one big display mesh and skin, and the locations and orientations of the ladders, hatches, and KAS grabbables relative to the root. If a vessel is in focus, then another vessel nearby should just be one big collision mesh, one big display mesh and skin, and the locations and orientations of the docking ports. Why treat a strut on a 500+ piece base as though it is going to move separately from the other stationary 200 tons of parts? There's no reason to do the calculations unless there is a collision, in which case the game could switch back to the full-parts description of the vessels. Doing this would increase the amount of RAM required to describe the scene, but vastly reduce the number of Physics calculations, probably by two orders of magnitude.

[Bill Phil]

Nah... 1.0 should be optimization. 0.99 should be where everything is released to the current owners of copies.

[michaelsteele3]

McPoisoned is my twin.

But yeah, this would be great

[jake9039]

the only problem I've ever had with memory is the (slow motion)SLOW (exit slow mo) startups

[Azimech]

While true, it doesn't change the fact that we don't know. 1.0 could be rock solid optimized, or it could be a mess. My money's on that it'll be somewhere in between and some people will say it's a mess and others will not understand where they're coming from.

IF it is not properly optimized, I will consider optimizations of primary import. But (like regex) I consider them of primary import NOW. I hope that Squad understands that releasing a buggy, crashy, slow game as a 1.0 release is a supremely bad move and if they don't, I hope they'll read this post and agree with me. If not, then I don't know if a 1.1 "bugfix/optimization" update will save the game.

We agree 100% on this. I installed a memory monitor mod so I could be warned when the game was in imminent danger of crashing. I can play 90-120 minutes before that happens on my current favorite mod setup (My LP series). What I don't agree with is thinking forward to 1.0 when they're still telling us things they're doing in 1.0 that we didn't even know about.

And (to reiterate) I also think waiting for 1.1 to optimize is a terrible idea.

Agreed!

[DSKY]

1.1 should be optimization? No, 1.0 should be optimization. They're essentially releasing the game as "production-ready" when it can't manage its memory. They're even adding new parts and systems to this "production-ready" software that can't manage its memory.

To be fair its recently been better than "vague notions", but I do agree that more detail on bugfixing would be better.

The memory issues with .90 have been so bad that I haven't played for two months. If things don't improve with 1.0, I won't even bother with 1.1.

It makes me literally sick to my stomach to say that about my all-time favorite computer game (and I've been playing them since the late 70s).

[basic.syntax]

Devnote Tuesday for the 17th revealed the beginning of work on memory optimization, conversion to compressed DDS image textures.

If the preliminary testing goes well, there's hope for the memory situation in 1.0.

[uglyduckling81]

I personally think that except for the lack of ram (which is a 32bit problem and not a game problem) the game runs fine even when using a lot of mods (assuming you have active texture management and a decent setup)

I agree. Apart from the constant CTD the game runs fine. It's a shame it takes me 5 mins to load the game after a CTD due to all the mods. I'm in a conundrum... do I take out mods and make the game boring to play thereby not play, or do I mod and don't play because of frustration as a result of CTD events. Oh wait I can play with mods on Linux 64 bit but then I get terrible FPS with large ships due to no SLI support. My god, my brain.... Perhaps if I could afford a better graphics card to replace the 2 aging SLI cards I would be in business. Where can I find some money?