[Poll] So what are we thinking about 1000 part ships?

[RocketRockington]

I remember one of the most common pre-launch discussions centered on just how much better KSP2's foundations - the physics, the way it handles parts, etc, was going to be.   A lot of the people most anticipating KSP2 didn't care too much about the extra features (colonies, interstellar) and were mostly looking forward to better physics.  Just thought I'd do a poll to see what happened to that, what people are thinking now.

FWIW, everything I hear from people who reviewed the code is that its KSP1 in most regards, nothing rethought to allow for higher part counts.  Some things authored a little cleaner perhaps, but if anything that also means slower and less ready to be optimized.  And the core issues that played high-part-count KSP1 craft are all still present + new performance issues that KSP2 has added.  So I'm not optimistic, but I'd love to see what the community thinks.  Has the music and new shininess made the physics foundation of the game less relevant?

Edited March 24, 2023 by RocketRockington

[Dakitess]

I don't see how it's going to be possible, while it was quite a promise from dev, in the very mouth of Nate himself IIRC and not just one time. Optimization and "large vessel" were key words during the communication, not so much for the 6 last monthes before release as they probably notice it won't run at all.

Sure, optimization will be done and performance will improve, just like Patch1 did it by 30 to 100% more framerate for most people. But this is about 10 to 15 times lower than expected, and this is a HUGE gap that I don't feel can be call "optimization", more like a complete revamp that won't happen.

Still hoping i'm wrong, of course, but... well, that would be a premiere OR the actual release and first patch are all on a game that is complete raw devs versions, and I don't it's the case, else it's not even Early Access but very Alpha. 

[twich22]

The game is clearly much more capable than KSP 1 ever was. When KSP 1 was at the stage KSP 2 is at now, which was a long long time ago, the size of ships you could build was much smaller. Most of that had to do with CPU power. From what I can tell, KSP 2 draws far less on CPU for much much larger craft. Once they fix whatever disaster is causing such an unnecessary draw on the GPU the game will be amazing. 

[pandaman]

Although the current state doesn't inspire confidence, I do tend to share @twich22's optimism that things will improve a lot.  

[Dakitess]

Uh ? No, I play KSP since 0.13 and the actual version of KSP2 is supposed to be about 0.24 or so, which allowed to build multi-hundreds parts. Sure, it ran at 10 FPS at most, but according to the PC specs at that time, about 2-4 times less performant than nowadays rigs, it's important to not forget about this. I had a 500 parts multi-modules station, in Jan 2013, that ran about 3-6 FPS. But it's a Laptop from... 2009, that was 650€ at this time so not a gamer one. See the gap ? About the same performance or better (actually probably way better haha), while we are speaking of KSP1 10 years ago, on a machine 14 years old compare to now, game being dev by a single guy at this time. And it's facing a 4 year Professionnal dev game, in 2023, using 2000€ machines of the same year. 

No, KSP2 is BY FAR lower in performance and Parts Counts than KSP1. But it will improve, of course. By how much ? I won't bet on 10-15 times optimization unfortunately.

Edited March 24, 2023 by Dakitess

[Scarecrow71]

I think you need to add an option to all of those polls up there:  Do you think we should give the developers some time to get more bugs out of the way and more performance enhancements into the code.

[Bej Kerman]

1 hour ago, Dakitess said:

Uh ? No, I play KSP since 0.13 and the actual version of KSP2 is supposed to be about 0.24 or so

Why? What's the thought process behind this? This is the very first release of KSP 2 - compare to the first release of KSP 1.

[LameLefty]

There’s a lot of unwarranted optimism in these polls. Maybe if KSP2 lasts as long as KSP1, in 10 years (*) PCs capable of 1,000 part ships will not be uncommon, but until/unless there’s a fundamental shift in the architecture and how physics and lighting are calculated and rendered, it’s not gonna happen.

 

(*) it was March 21, 2013 - 10 years ago - that I bought KSP and my MacBook Pro i7 struggled with 150 part ships. My current PC (i7-11700KF) can run those same ships at 120 fps. 

Edited March 24, 2023 by LameLefty

[regex]

Were 1000 part ships ever an actual thing that was promised? Why would you need 1000 parts on a single vessel?

As far as I'm concerned there are more patches to wait for and play with before I answer a poll like this.

[The Aziz]

4 minutes ago, regex said:

Were 1000 part ships ever an actual thing that was promised? Why would you need 1000 parts on a single vessel?

As far as I'm concerned there are more patches to wait for and play with before I answer a poll like this.

Note that these were shown as development footage, it wasn't just cinematic trailer. And the framerate wasn't slideshowy

[regex]

2 minutes ago, The Aziz said:

Note that these were shown as development footage, it wasn't just cinematic trailer. And the framerate wasn't slideshowy

In orbit around Jool with no real terrain in sight, there's probably one engine on that big ship docked to the platform, the platform might be using the colony system what was yanked out before release, there could be some part welding going on for orbital construction platforms (what I'm guessing this is) to accomodate the colony system... Also, framerates can be smoothed for video. Lots of questions.

[Cpt Kerbalkrunch]

I had high hopes, as I think we all did, and they were dashed in the first 10 minutes of playtime. I think anyone who's honest would have to agree. Anyone in any business knows it's always better to under promise and over deliver. Not sure what went wrong here, so I won't speculate. Where we go from here is the bigger question.

I build 1,000+ part ships in KSP1 quite a bit (manual strutting means part counts reach absurd levels quickly) and, for the most part, I get good performance. Low FPS on launch, but no stuttering, and performance climbs quickly as you drop boosters. I haven't exceeded 200 parts in KSP2 yet, so I'll have to experiment more. The performance has actually been about the same for me.

The reason why I voted "no" in this poll is that I don't think it'll ever be stable at 1,000 for the majority of players. I think it's going to be much like KSP1. If you have a high-end machine, it'll be great. If you don't, you'll stutter along at 3fps.

I was so excited this game was finally being released, I went out and bought a new, high-end PC (kinda crazy, but there ya go). At the time, I thought a 3080ti was overkill. When I saw the recommended was a 3080, I couldn't believe. My first thought was that my shiny new PC wasn't so "high-end" after all. It was barely above the recommended specs. My second thought was that they just priced out a lot of players. Especially younger players. I dreamed of Alienware for 20 years. When I could finally afford it it was garbage made by Dell and I returned it. I don't think this is the way to grow the fanbase. We need younger players to get hooked on the game. If they can't afford it, that's not gonna happen. Hopefully the switch to the new Unity version mentioned in the Dev Diaries can solve a lot of this. Sounds like a long way off, though.

[Dakitess]

1 hour ago, Bej Kerman said:

Why? What's the thought process behind this? This is the very first release of KSP 2 - compare to the first release of KSP 1.

Come on, let's be honest here, i'm even being very on the good side saying 0.24 : It's KSP2. Two. It comes after the first one being out for a decade, with all the feedback and the knowledge it brought. This is a professional team, quite a lot more staffed than KSP1 at its prime. We are in 2023. Price is 50$.

OF COURSE it's an EA, that's obvious, things will get better, etc etc. But the state of this game is definitely not a KSP1 0.13. Clearly not a KSP1 0.18. And only perhaps something like KSP1 0.24 as I proposed. It's just honesty, good faith, please don't go hyperbolic for no reason. KSP2 is advertised as an EA that will grow with us, not a random solo project from a solo Mexican passionate guy. KSP2 is sold damn expensive because they believe it's worth it (it's not, but that another story), because they consider it as well established version that can be used as EA at this price. A base for all the upcoming feature. Editors choose too rush it, probably against the opinions of Dev Team, and it was not ready, 3 weeks short of Patch1 which address quite a lot of major issues it had. But anyway : everything here leads to the conclusion that this is at least 0.24, to me. Feel free to argue differently.

And again, that's just a small part of the discussion. Even using an elderly version of KSP1, it's based on OLD computers (please, really, consider this aspect), with OLD technologies (Game development wise), by a team, money, organization that has nothing in common. And after 4 (+ ?) years of dev. Soooo yeah, obviously, what its awaited is a game that runs better than a 10 yo indie game on 12yo low end laptops, it's common sense. And since, it does not runs better or just equally bad, well, that's no good. At the same state, both game should definitely not run the same, neither twice as good, but way more. And then only, it would improve to get vastly superior, as advertised to reach 1000 parts counts crafts at decent framerate. Yeah, 20 FPS would be good enough probably. But with the actual base, the actual start, it's probably a 10-15 times gap performance to fill, does not sound possible to me, as a candid player that just played a lot with KSP1 and knowing that they apparently re-used quite a lot of the fundamental pillars if the first old opus.

 

Edited March 24, 2023 by Dakitess

[shdwlrd]

Quite honestly, it's too early to tell. With only one patch down and an unknown number to go, we can't tell the trajectory of the performance updates right now. This type of question should be asked when there is a reason for a majority of players to start assembling crafts or a collection of crafts to get close to 1000 parts. But at the moment, there's no real reason to build anything that big outside of content for creators or trying to break game.

[Fullmetal Analyst]

11 hours ago, RocketRockington said:

everything I hear from people who reviewed the code is that its KSP1 in most regards, nothing rethought to allow for higher part counts.  Some things authored a little cleaner perhaps, but if anything that also means slower and less ready to be optimized.  And the core issues that played high-part-count KSP1 craft are all still present + new performance issues that KSP2 has added.

bruh, this sounds like the game is a total scam...

 

1 hour ago, regex said:

Were 1000 part ships ever an actual thing that was promised?

they actually promised multi-thousand part ships

 

57 minutes ago, The Aziz said:

Note that these were shown as development footage, it wasn't just cinematic trailer. And the framerate wasn't slideshowy

this is likely just the parts rendered without any planets or other stuff, which is never going to happen when actually playing the game

 

1 hour ago, regex said:

Why would you need 1000 parts on a single vessel?

it starts to become a problem when you are docking lots of smaller ships to a station, bring tons of satellites etc. for large scale missions

[regex]

Just now, Fullmetal Analyst said:

they actually promised multi-thousand part ships

I don't have any reason to doubt this but ... [citation needed]

[Fullmetal Analyst]

27 minutes ago, regex said:

I don't have any reason to doubt this but ... [citation needed]

it was a topic on the discord shortly before the game released

[Rosten]

Kraken-slain physics that could support large ships without cheats like autostrut was one of the biggest things I was looking forward to, as even if they had failed to deliver on everything else, I'd still be able to build the things I couldn't in KSP 1.

[regex]

1 minute ago, Fullmetal Analyst said:

it was a topic on the discord shortly before the game released

lol okay

[Cpt Kerbalkrunch]

3 minutes ago, shdwlrd said:

Quite honestly, it's too early to tell. With only one patch down and an unknown number to go, we can't tell the trajectory of the performance updates right now. This type of question should be asked when there is a reason for a majority of players to start assembling crafts or a collection of crafts to get close to 1000 parts. But at the moment, there's no real reason to build anything that big outside of content for creators or trying to break game.

You definitely have a point. The reason I haven't exceeded 200 parts is because there's no reason to. I only did it in the first place to see if I could. The fact that there's not much to do yet doesn't a great selling point, but that's where we are for now.

I still think it's a fair question, though. We've seen enough, I think, to have a good idea of future performance. High-end rigs will do well, low-spec machines will struggle. Though it took a decade, KSP1 was able to thrive like this. @Dakitesshad a good point about the march of progress. New CPUs and GPUs are released, people keep upgrading and, over time, more and more people are able to enjoy the game. If that's the case again, it's gonna be a while before you see a high player count.

[HebaruSan]

28 minutes ago, regex said:

I don't have any reason to doubt this but ... [citation needed]

I asked in another context where people thought that higher part counts were promised, and I received a link to an interview with a Youtuber, who asked how the availability of "new hardware" had affected the development process. Nate responded, in part:

Quote

Much more of our bandwidth goes to making sure that a large number of people have a positive experience, that you're not running into the same kind of CPU constraints around the simulation of rigid body arrays. I mean, we need people to be able to build very large vehicles, for example, especially with colony building, you're going to get up into high part counts. So it's important for us not to have your framerate fall to its knees, even if you're on a "normal" computer. So that's where most of our focus is these days.

... which the fan base interpreted as a commitment to dramatically improve performance for high part count craft, as can be seen in a few comments above (much like what happened with the "slay the kraken" comment, which just means "we will have a process for fixing bugs").

[regex]

1 minute ago, HebaruSan said:

... which the fan base interpreted as a commitment to dramatically improve performance for high part count craft, as can be seen in a few comments above (much like what happened with the "slay the kraken" comment, which just means "we will have a process for fixing bugs").

"high part counts", and naturally this led to some "high" expectations...

[Jarin]

On the topic of that giant station screenshot, that's almost certainly an orbital colony and isn't simulating all the individual parts physics.

[Stoup]

Just now, Jarin said:

On the topic of that giant station screenshot, that's almost certainly an orbital colony and isn't simulating all the individual parts physics.

Even if that's the case, that not every part is physically stimulated... Didn't they want colony buildings to have physical stimulation not unlike how parts are stimulated? RE: the end of that first trailer where the whole colony starts falling all over itself. I know that wasn't a gameplay trailer but I feel like I remember there being talk about how that's what they wanted.

I'd be quite surprised if that's how things actually turned out at this point, it seems more likely to me buildings and maybe also orbital assets might be static to some degree.

[Jarin]

They did say that you could "Rogue One" an orbital station with a pusher ship, even if colonies couldn't mount engines, so there's some kind of simulation. I expect it's just simplified and/or has larger component chunks. Like that hydrogen storage array there - that could have dozens or more parts if built in the VAB - could be a single Part. Likewise those identical residence rings.

Edited March 24, 2023 by Jarin