Do you think KSP 2 will eventually end up more optimized?

[MarcAbaddon]

I think there will definitely be some optimization, but at the same kind we will have more code always running in the background and larger vessels in the future, which colonies and interstellar being a thing. So with more demand on PC resources at the same time it isn't clear that there will be a large net gain in the future. Generally, in my experience you can get some more performance out of games, but gains are rarely huge after a game has been released (even in EA). It's not just that there is a code already in place people can't easily replace, it is also that it is the same programmers working on it. I know the KSP 2 team seems to have a new person responsible for the graphics engine, so maybe it will be different here.

[Periple]

10 hours ago, MarcAbaddon said:

Generally, in my experience you can get some more performance out of games, but gains are rarely huge after a game has been released (even in EA). It's not just that there is a code already in place people can't easily replace, it is also that it is the same programmers working on it. I know the KSP 2 team seems to have a new person responsible for the graphics engine, so maybe it will be different here.

This is true a lot of the time but I'm not sure it is now. Games that get released on consoles have to meet specific performance targets to pass certification, so anything that's on a console will run pretty decently. This isn't the case here.

When a game hits alpha it commonly performs pretty badly and if optimization is only done after that, you will often see really big improvements. KSP2 EA upon release (and even now) is basically an alpha build. You don't normally see those unless you're working in or with the studio (like as a publisher). It is obviously unoptimized and some modders have dug into it to point out specific low-hanging fruit that will improve performance a lot. 

The hard part is going to make it scale to dozens or hundreds of ships with hundreds of parts. Right now that's not even where the bottlenecks are.

[RaBDawG]

On 6/10/2023 at 1:39 AM, RandomGrape said:

I think it probably will, but not for a while. Quite a lot of modern games have come out like this.

 

Edited June 29, 2023 by RaBDawG

[Ruedii]

They have a surprising number of engineers on their team, due to the nature of the game allowing them to attract them. 

I don't think they will have any problem optimizing the game itself.  The question is if they can do so without substantial quality loss.

There is a massive amount of A/B testing and smoke testing to do to determine if something creates reliable performance gains while not having notable quality loss. 

Depending on the gains, it may be worth adding to the list of performance optimizations to put for low end systems, provided it provides the same performance gains on those older systems.

[Wheehaw Kerman]

Patch 1.3 has brought some optimizations in.  Running noticeably more smoothly for me.  I expect things will only get better.

[Rubenio]

Frankly, I am still disappointed in performance.  Installed 1.3 when released and to me there is still something wrong with this game.  I'm getting 16-17 FPS and it doesn't matter if i run low res or high res.  doesn't really make sense.

Here are specs of my pc:

Processor:  Intel(R) Core(TM) i7-10700F CPU @ 2.90GHz

RAM : 48 GB

GPU:   NVIDIA GeForce RTX3060

Any thoughts on what performance I should expect?

[Spicat]

42 minutes ago, Rubenio said:

Frankly, I am still disappointed in performance.  Installed 1.3 when released and to me there is still something wrong with this game.  I'm getting 16-17 FPS and it doesn't matter if i run low res or high res.  doesn't really make sense.

Here are specs of my pc:

Processor:  Intel(R) Core(TM) i7-10700F CPU @ 2.90GHz

RAM : 48 GB

GPU:   NVIDIA GeForce RTX3060

Any thoughts on what performance I should expect?

There is definitely something weird here, I have the same CPU and an RTX 2070 (so a bit less performant than you) and I got like 45 fps with a ~100 parts ship and 60 fps in space.

Have you played on a new save? With what craft did you play? You might take a look because it's not normal that I have three times more fps than you with approximately the same specs.

[Royalswissarmyknife]

Considering how I got 100 FPS in 1.3.0 

Yes

Spoiler

Please note that I was on low settings

 

Edited June 24, 2023 by Royalswissarmyknife

[dr.phees]

There will be, well, has to be a lot of optimization. However, as athmospheric heating already seems too much to be done without loosing too much performance, I don't think KSP 2 will really become the same cult like the original. KSP 2 is not really offering any new things, nothing creative and no feeling of playing an advanced KSP. Just a polished up copy of it. I doubt the solution for interstellar will be any more engaging from a gemplay standpoint to pull many players in.

The current part-fiddling (I mean, how can airbrakes be so hyped up? They are just a copy from the original!) does not give me confidence that there is enough wiggle room for better performance optimization to include all the missing systems.

Maybe they are playing for time, expecting hardware to get more performant. But based on the current slowing down of the PC market I don't expect too much myself.

The big questions are: How much money can they burn on development and can they manage to somehow make KSP 2 feel really new and exciting?

[RaBDawG]

No.

[Rubenio]

On 6/24/2023 at 10:56 AM, Spicat said:

There is definitely something weird here, I have the same CPU and an RTX 2070 (so a bit less performant than you) and I got like 45 fps with a ~100 parts ship and 60 fps in space.

Have you played on a new save? With what craft did you play? You might take a look because it's not normal that I have three times more fps than you with approximately the same specs.

I've tried a new campaign and the first flight is around 20 fps, then it drops on the second...  i am thinking of starting fresh and using a stock ship that can get into orbit and look at fps that way.  

I also dropped res down to 1080p and that made zero difference.  not sure i understand what's going on.

[Spicat]

7 hours ago, Rubenio said:

I've tried a new campaign and the first flight is around 20 fps, then it drops on the second...  i am thinking of starting fresh and using a stock ship that can get into orbit and look at fps that way.  

I also dropped res down to 1080p and that made zero difference.  not sure i understand what's going on.

So I saw ObsidianAnt video:

I have also better performance than him while he has a 4090. His PC is using only 35% of the GPU. Is this the case for you also? If yes, you should probably add a bug report about that to let them know about the problem (if they don't already know).

[Rubenio]

Thanks for the link to the video.  i've done some more homework...  

added MSI afterburner to monitor CPU and GPU - recorded some interesting events..

So on the launch page, seeing 100% GPU usage and 200FPS - i don't think we need that level of HP on the launch page...

https://imgur.com/Go3B2fT

Basically, while game is paused, FPS increases to 50% and %GPU rises to 100%.

https://imgur.com/18IdQm1

When unpausing the game, FPS goes down - as expected.  What's not expected is %GPU goes down as well - definitely not desired! %GPU goes down to 25% and FPS drops to 17!  Clearly we're not using the system's horsepower in the right places.  for static images it's overcooked and for places where you need it, it's not leveraging all the hardware.

https://imgur.com/cN44xqZ

I've also made the video to illustrate this as well.  I guess I should file this as a bug report?

And btw, the space station in the scene went off with mystery dV out of Kerbin orbit so that bug is still around in the latest update...

[Periple]

The developers will have much better profiling tools than you do so I don’t think there’s much point reporting that. Optimization is hard work and they’re clearly doing it. Have patience, more improvements are surely on the way.

[Dakitess]

It's still important for us to report that it's not running nearly as good as it should after the patch : especially because it improved the situation, sometimes with good results, it's necessary to insist here and there about perf not being enough anyway, otherwise people enjoying and bragging "amagad it's so smooth now, thanks devs, it's perfect !" are the only one represented, they're fine to say so if it's suit their requirement and expectation, totally, but about as much as saying otherwise.

Still the same things, again and again, to let the team know that it's not on par with what we expect, otherwise they would be right to only do minor upcoming improvements, if people seems to be fully happy. I'm pretty sure they would continue improving performances anyway, but maybe not as much as if other people actually continue saying they are not happy.

[RalphKerman]

I think we are at the start of a long road to optimizations.

However I am (slightly) pessimistic. I fear there are a lot of problems with many core game features at this early stage (the bug reports get more and more comical each day and IMO indicate some real trouble with the physics engine "under the hood") and with (as some users above already said) for example the delay of the heat system mechanics (which surely eat some computational resources) we are not getting much more optimization out of the current game engine state and may even lose some with more features. With prior knowledge of KSP 1 and the challenges that it brought I was really hoping the devs would have found ways to be much more prepared and efficient, especially given  the really long development time that was extended due to the pandemic (technology has advanced so much yet the code base seems to be a (poorly?) modified KSP 1 base with many errors introduced that I can't understand at all as they are not new aspects of the game: Atmosphere physics, drag, save file corruption, simple parts like decouplers or docking ports not working etc....)

I hope we can  stabilize at or above the current 0.1.3.1 level of performance with many game-breaking bugs eliminated and many features added. If the performance keeps stable and I can run a Duna lander and base mission in KSP 2 with science, heating and no Kraken attacks every 5 min I would be happy for some time till the long-term refinements and cool mods become available. And I guess a lot of the community would be happy, too, and maybe even new players would join the journey.

But at the current stage of the game I just don't see this happen anytime soon and within a financially stable time for the publisher. KSP 2 is no lovely Indie title as KSP 1 was that had time to evolve and grow slowy. My biggest fear is that the high bar the devs have set for themselves and the high expectations of an impatient community (which I am guilty of as well, I want KSP 2 feature-complete rather sooner than later) will lead to economical problems and someday long before all that optimization is being able to be implemented the money will stop flowing and the title will be abandoned in a very poor state.

I want my prior paragraph to be 100% false and really wish the game gets all the time and resources it needs to become what we were promised. I still want to believe KSP 2 will be a great great game someday.....

Over the weekend I have completely re-installed KSP 1 after watching several new videos and I set up a nice 82-mods heavy install with a huge tech-tree, stunning visual mods and some non-base-game stuff like life support and new planets (no RO/RSS, just Kerbin system, I am not a real life NASA guy, hehe) that I have never tried before and I am having a blast. It runs on a PC from 2016 (it was a monster back then, now it's "old and outdated but still decent") with good FPS and in the 16 hours I played since Saturday I had 0 crashes or Krakens. So my desire for sending green little guys to space is satisfied for now. I will hang on and follow the development of KSP 2 and hope for the best.

[KUAR]

7 hours ago, RalphKerman said:

I think we are at the start of a lon.g road to optimizations.

However I am (slightly) pessimistic. I fear there are a lot of problems with many core game features at this early stage (the bug reports get more and more comical each day and IMO indicate some real trouble with the physics engine "under the hood")...

I understand that they are in the process of literally replacing some key parts of the engine with v0.2.x

I'm not sure of the extent of it but I do hold out hope that there's a plan there. Glass half full!

[RalphKerman]

14 minutes ago, KUAR said:

I understand that they are in the process of literally replacing some key parts of the engine with v0.2.x

I'm not sure of the extent of it but I do hold out hope that there's a plan there. Glass half full!

And that is what concerns me a bit. The game should (after this long time of development) be much further down the road in terms of getting the core engine upgraded. The current 0.1.x engine feels too much of an early prototype from which to start building a new engine before adding any new features at all. For the long time the title was in development this is very underwhelming and fuels my fear they will simply run out of time (time == money as well ...) before this switch can be completed and features become added. I also hope they have a new engine almost ready to release and were just caught off-guard with the demand for an Early Access release. If that's the case however they could maybe communicate it and say "ok, we see all the many bug reports but we have a totally new engine almost ready, give us a bit of time and we switch to that." What I am seeing instead is however the efforts to fix the current engine which leads me to believe there might not be a new engine ready and waiting to be installed. Why would they invest that much time in fixing the "interim engine" then?

[Nicrose]

I think it will be quite a bit smoother but I have a feeling a lot of what is expected will be cut by the end. I have a feeling it will go something along the lines of 20% taken out by 1.0 in the name of performance and based off the steep price and desire to raise it I wouldn't be surprised to see them put some things in a DLC. This is just my guess, I hope it is just 100 times better and everything expected and more by 1.0. 

[GradientOGames]

I have a really good feeling that optimisations are going to skyrocket once some internal transitions are complete. As a Unity dev myself, I understand what it takes to implement highly performant systems such as ECS, DOTS, and JOBS, (NEO included once multiplayer arrives). If you delve into the patch notes, the dev team have already made significant improvements by implementing these systems.

In patch 3, the devs implemented the job system for calculations with parts in liquids (making it multithreaded) which spread out water calculations from the main threads to other threads improving times for 10ms to 1.5 which is a whopping 6.6x increase in speeds (realistically it would be 3-4x as some other threads would be used for other tasks as the jobs system is implemented into other systems.)! The devs new terrain system which will be implemented within the next few months (I forgor the name) is documented to significantly improve terrain quality and framerates.

Another special unity system that can easily massively reduce frame times is converting the ECS system (see ** for explanation).  I am going to remain optimistic and assume that it hasn't been implemented already. Fully implementing ecs is a monumental task, however Unity offers a way to quickly increase performance without changing any scripts which is the hybrid entities and renderer for unity. You just set it on within all objects and bam, converted to entities (not really, again, see **). I'm assuming the devs haven't already done this as it makes it harder to debug.

 

(the following may not be 100% accurate; I don't have deep technical knowledge of ecs)

**:  ECS (entity component system) is different from Unity's ordinary Game Object system in one core way, gameobjects have lots a jargon and unneeded values while entities only have required values. So Fully converting to entities significantly reduces cpu overhead as it has much less random variables to deal with. If you only convert partway by using hybrid renderer, you get all the rendering benefits (better gpu performance) while still using game objects. That's why I believe anything that can be converted to hybrid renderer, should; there's not much effort required and it works so much better. one of the reasons rendering is better with ecs and hybrid renderer is because entities give rendering calls and calls to access video memory using the unity JOB system and burst compiler (Burst compiler compiles all JOBs code to C++ to get run much faster) to significantly reduce both cpu and gpu loads.

 

My verdict is that performance will significantly increase with each major update/patch and running over 1000 objects in a performant manor.

Thanks for reading

 

P.S. This probably won't age well lmao

[Alexoff]

2 hours ago, GradientOGames said:

The devs new terrain system which will be implemented within the next few months (I forgor the name) is documented to significantly improve terrain quality and framerates.

Really? It seems that the developers did not promise anything of the kind in the coming months, much of what was said in the spring has unclear deadlines for implementation.

[Scarecrow71]

1 hour ago, Alexoff said:

Really? It seems that the developers did not promise anything of the kind in the coming months, much of what was said in the spring has unclear deadlines for implementation.

Mortoc mentioned that there will be some terrain optimizations coming in the next few patches.  His dev diary from 3/10:

I'm pretty sure @GradientOGames was talking about either CBT or HDRP, which - as you've pointed out - have no defined timelines.

[shinya]

21 minutes ago, Scarecrow71 said:

Mortoc mentioned that there will be some terrain optimizations coming in the next few patches.  His dev diary from 3/10:

I'm pretty sure @GradientOGames was talking about either CBT or HDRP, which - as you've pointed out - have no defined timelines.

I hope the next few patches will come faster than heating & reentry...

[Alexoff]

2 hours ago, Scarecrow71 said:

Mortoc mentioned that there will be some terrain optimizations coming in the next few patches.

With the addition of the grid fins, the developers have no specific timelines, although it was shown in May. Although it would seem that this is a matter of several days. And there seem to be a million such shortcomings and expected additions, it's hard to say how and when this is done. ¯\_(ツ)_/¯

[Spicat]

7 hours ago, Scarecrow71 said:

Mortoc mentioned that there will be some terrain optimizations coming in the next few patches.

If you talk about this:

On 3/10/2023 at 10:00 PM, Intercept Games said:

Coming soon in a patch, we will be able to turn off the Anti-Tile system in the terrain. In a bunch of places, the effect is negligible, but you can see the Eve surface has a repetitive texture artifact without it. This visual polish comes at the cost of accessing each texture a few more times, putting strain on the memory bandwidth of the GPU. Disabling this effect can have a small-to-medium sized effect on the framerate, depending on the GPU in question. 

Optimizations like that one are happening now and will arrive in the next few updates.

He mentions how some of the change that we'll see is disabling some graphics for low settings to allow even more people playing the game (alongside other optimization for everyone).

We can already see some examples of these changes in the previous patches:

On 6/22/2023 at 5:45 PM, Intercept Games said:

KSP2 Update v0.1.3.0 Release Notes

• Optimized tesselation factor for medium and low quality water to improve GPU performance
• Reduced low- and medium-quality cloud texture sizes for Kerbin, Duna, Eve, and Laythe  

On 4/12/2023 at 6:55 PM, Intercept Games said:

KSP2 Update v0.1.2.0 Release Notes

• Anti-tile is now disabled when low quality is  selected
• Improved low graphical setting visuals in some scenes

No ETA on other optimization though (beside the occasional ones each patch), especially the new terrain system. I hope we'll get some news on this (I think it will take several months but I would like to know if their implementation is getting good results).