Jump to content

How much better will the game be with unity 5 ?


Recommended Posts

31 minutes ago, Vaporo said:

Well, the game should run much faster when flying ships with high part counts with the implementation of the new Physx, and even faster if you have a multi-core processor.

I hope so. But some people assert that the game is using O(n2) or even worse-performing algorithms in core inner loops during flight. If that's so, there may not be huge improvement: twice as many parts will mean four times as much pressure on the CPU.

Link to comment
Share on other sites

1 hour ago, Berlin said:

I just really hope the garbage collection issue is fixed

That will be interesting. Maybe someone knows, but I'm not sure how much better it might be. Maybe better because the work can be handed off to another core, but maybe not because it still has to happen.

Link to comment
Share on other sites

1 hour ago, Berlin said:

I just really hope the garbage collection issue is fixed :( 

GC must happen, and by nature it must stop executing managed code in the process. Offloading the work into another parallel thread will probably reduce the length of each GC run, but the better solution is to reduce the need for it to run in the first place, in this respect, it appears there's still plenty of optimisation that could be done on Squads side - for example: the resource calls walk the entire craft, often more than once per frame. Here's an obvious example of  inefficient code at work, exacerbating the Unity GC issues.

U5 will be nice I'm sure, but it's no silver bullet - slow code paths and unnecessary repeated operations still abound, and U5 won't fix that.

 

Link to comment
Share on other sites

1 hour ago, steve_v said:

U5 will be nice I'm sure, but it's no silver bullet - slow code paths and unnecessary repeated operations still abound, and U5 won't fix that.

Sadly this is true.  
There is a lot of optimisation still to be done.  However I have hope that the interface upgrade will yield improvements as a number of the memory leak problems seem to stem from aspects of the current interface. 

But overall I expect that the first release on U5 will actually cause more heartache and will break a load of stuff (and there will be a number of community riots). It's a massive change that will impact on so many aspects of the code and I don't think Squad follow TDD (Test Driven Development), even if they did and had near 100% test coverage there will still be things that get missed. 

Link to comment
Share on other sites

1 hour ago, 322997am said:

I'm going to build a pc. What is worth it for u5?

*Rummages about*
*Rummages about some more, empties desk drawers*

Nope, sorry, crystal ball gone missing again. Must be those damn gremlins.
As nobody knows how much of an improvement U5 will make, nobody is likely to know what kind of PC is "worth it" either - whatever that means.
 

Link to comment
Share on other sites

On 12/11/2015, 11:09:50, manaiaK said:

I hope so. But some people assert that the game is using O(n2) or even worse-performing algorithms in core inner loops during flight. If that's so, there may not be huge improvement: twice as many parts will mean four times as much pressure on the CPU.

That doesn't mean they coded it to limit parts count does it?I keep worrying this turns out to be the case and they are purposely not allowing more out of some idea of balance or design... 8(

Edited by Arugela
Link to comment
Share on other sites

13 minutes ago, Arugela said:

That doesn't mean they coded it to limit parts count does it?I keep worrying this turns out to be the case and they are purposely not allowing more out of some idea of balance or design... 8(

Of course not, there's no benefit to making the game perform worse with higher part counts. That's not to say that there isn't room for optimization (some of which is coming in 1.1 afaik), but it's not written to deliberately punish part count.

Of course, more parts will always perform worse, no matter how much optimization is done. Even if they could get it as good as O(log n) there would still be a performance penalty for increasing part count.

Link to comment
Share on other sites

2 minutes ago, Red Iron Crown said:

Of course not, there's no benefit to making the game perform worse with higher part counts. That's not to say that there isn't room for optimization (some of which is coming in 1.1 afaik), but it's not written to deliberately punish part count.

Of course, more parts will always perform worse, no matter how much optimization is done. Even if they could get it as good as O(log n) there would still be a performance penalty for increasing part count.

Well that gives a bit more hope if that is the case.

Link to comment
Share on other sites

I'm switching my application to Unity 5 next sprint. The benefits:

  • The standard shader is much easier to set up and have good performance than was the legacy shader setup. For me, this means my art team can improve the visuals a lot. Alternately I could have improved the performance. This might help KSP as well.
  • Lower memory consumption. I don't know why, but we're using 30% less memory than Unity 4 on iOS (I haven't profiled on OS X or Windows since we don't really care on those). Since I don't know why, I can't predict the effect on KSP.
  • On iOS, the il2cpp backend is much faster than Mono AOT. Totally irrelevant to KSP.
  • Faster development speed because I can ditch a couple of plugins I was using that now have built-in equivalents, so now I don't have to wait for them to recompile and reimport their assets. I don't know if this helps KSP.

The cost:

  • Some old plugins don't work anymore. In particular, iOS -- we yanked some old features because we couldn't port them.
  • Some of my code doesn't quite work anymore either.
  • Had to redo all the shaders. I wanted to anyway, but it's still a cost.

Switching has taken a couple of months, and I've probably got an order of magnitude less complicated a system than KSP.

Link to comment
Share on other sites

Im hoping we at a minimum get out pre 1.0 performance back (the new heating and aero have really harmed FPS with larger vessels), with teh ability to have one core per vessel at a minimum allowing say 3-4 500 part ships loaded in the same area with total part count exceeding say 1500 parts and not much lag.  Ive hear that only one core works per vessel, but that would be fine if well, each core can at a minimum support 1 500 part ship (back in 0.90 i could easily have 1 500 part ship before lagfest kills my enjoyment of the game).  RIght now, anything above 300 parts starts to get some stutter/lagging, and anything above 500 parts becomes laggy enough to make the game not very enjoyable.  Honestly if i wasnt dependent on many of the new features id prolly revert to 0.90 and play that one as the lag was way less and not as much of an issue there, but ive taken too much a liking to the new bugfixes in 1.0.5, so despite the fact that i actually liked the 0.90 aero more (i mean common, sci-fi stuff worked so much better back when jets had orbital velocity, and perfectly flat flying saucers flew fine sideways, ect ect ect), i think ill stick to 1.0.5 since many of the really agrevating bugs were fixed (claw is useable which is good as its my primary method of docking 2 ships together, water isnt a guaranteed wet grave, ships dont randomly explode due to energy flowing from cold to hot areas, ect).

Link to comment
Share on other sites

This thread is quite old. Please consider starting a new thread rather than reviving this one.

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...