What happened to increased communication?

[whatsEJstandfor]

1 hour ago, Scarecrow71 said:

As I've posted a zillion times out here:

• AMD Ryzen 9 3900 12 core
• GeForce RTX 2060 Super
• 32 GB RAM

A 2060 Super was pretty underperformant when it was new, but you should still be getting better framerates than what you listed (I'm assuming you're playing at 1920x1080). This is a dumb question but are you sure that your GPU is plugged into a x16 PCIe slot?

[Icegrx]

2 hours ago, Scarecrow71 said:

As I've posted a zillion times out here:

• AMD Ryzen 9 3900 12 core
• GeForce RTX 2060 Super
• 32 GB RAM

It shouldn’t be an issue, but for KSP2’s sake it’s the processor holding you back mainly here. The way the cache is split between the two CCDs on that architecture hamstrings the performance of games that cannot utilize all 12 cores. KSP further limits that down to a rumored 1 core. So your single highest performing core can only utilize about half the processors available cache. 

[NexusHelium]

2 hours ago, Scarecrow71 said:

• AMD Ryzen 9 3900 12 core
• GeForce RTX 2060 Super
• 32 GB RAM

Oh yeah that’s way worse than mine…

I need a new PC 

[calabus2]

3 minutes ago, Icegrx said:

It shouldn’t be an issue, but for KSP2’s sake it’s the processor holding you back mainly here. The way the cache is split between the two CCDs on that architecture hamstrings the performance of games that cannot utilize all 12 cores. KSP further limits that down to a rumored 1 core. So your single highest performing core can only utilize about half the processors available cache. 

Unity strikes again 

[Stephensan]

22 minutes ago, Icegrx said:

It shouldn’t be an issue, but for KSP2’s sake it’s the processor holding you back mainly here. The way the cache is split between the two CCDs on that architecture hamstrings the performance of games that cannot utilize all 12 cores. KSP further limits that down to a rumored 1 core. So your single highest performing core can only utilize about half the processors available cache. 

from what i test ksp 2 uses 6 cores 12 threads for me 

could be pinging back and forth or whatever but.

[steveman0]

3 hours ago, calabus2 said:

Unity strikes again 

Is this based on actual knowledge of KSP2 specifically? Unity games are perfectly capable of leveraging multiple threads just as nearly any other engine when designed as such. Can you elaborate on the KSP2 design of their threading approach as to why it's an issue?

[chefsbrian]

21 minutes ago, steveman0 said:

Is this based on actual knowledge of KSP2 specifically? Unity games are perfectly capable of leveraging multiple threads just as nearly any other engine when designed as such. Can you elaborate on the KSP2 design of their threading approach as to why it's an issue?

Its not KSP2 or Unity specific, its physics specific. You can't really do parallel physics calculations, since you need the results of the last physics frame to start computing the future physics frames. In theory, you can parallelize calculations for groups of entities that are guaranteed to not interact in a frame, but in practice doing that without a bajillion bugs is near impossible. Best you get is KSP's attempts to on-rails stuff with predictable outcomes.

So while its inaccurate to say KSP2 is single-threaded, KSP2 is only able to effectively utilize a single thread for what is probably the overwhelmingly most expensive part of its CPU work.

[steveman0]

21 hours ago, chefsbrian said:

Its not KSP2 or Unity specific, its physics specific. You can't really do parallel physics calculations, since you need the results of the last physics frame to start computing the future physics frames. In theory, you can parallelize calculations for groups of entities that are guaranteed to not interact in a frame, but in practice doing that without a bajillion bugs is near impossible. Best you get is KSP's attempts to on-rails stuff with predictable outcomes.

So while its inaccurate to say KSP2 is single-threaded, KSP2 is only able to effectively utilize a single thread for what is probably the overwhelmingly most expensive part of its CPU work.

Unity is not really central to the situation. It is possible to solve these issues either through multithreading, as difficult as this would be, or through batching parts to reduce simulation load. The choice of engine is irrelevant to this though as any good solution here isn't going to rely on the simplified toolset of any off-the-shelf engine.

[cocoscacao]

5 hours ago, steveman0 said:

It is possible to solve these issues either through multithreading, as difficult as this would be, or through batching parts

Maybe just theoretically, and I would put that under a giant maybe. I don't think writing your own engine would help much either. 

[PDCWolf]

4 hours ago, cocoscacao said:

Maybe just theoretically, and I would put that under a giant maybe. I don't think writing your own engine would help much either. 

Unity does come with a lot of features that are outright insufficient if you want to do a game like KSP. However, in reality, I'm 100% sure a custom engine runs contrary to what seems to be the very concept KSP2 is built upon: cash out fast by delivering a quick "modernization" of a very successful indie game. The franchise is made, the game is done, you can even reuse the off-the-shelf toolset like Unity and the same middleware, add 2 or 3 of the popular mods in a non committal way (Resources, Colonies, Multiplayer but no LS as to not force players to use any feature) and sit back to watch the cash flood in.

Meanwhile the community wanted KSP2 to be almost entirely different and to wipe away the errors of the first one... how the turn tables.

[cocoscacao]

17 minutes ago, PDCWolf said:

Unity does come with a lot of features that are outright insufficient if you want to do a game like KSP

Such as? I'm not familiar with game engines and their differences. Regardless of what we may think this game is, and how it's been developed thus far, I don't think different engine would help matters. Again, I'd need specifics on what you mean by Unity being insufficient.

Also, developing an engine isn't an easy task. It's a project for itself. Yes, earlier games had their own, but they were a lot simpler than games of today, where you're pretty much obliged to have physics simulations... So people go after off-the-shelf toolset for a reason.

[PDCWolf]

13 minutes ago, cocoscacao said:

Such as? I'm not familiar with game engines and their differences. Regardless of what we may think this game is, and how it's been developed thus far, I don't think different engine would help matters. Again, I'd need specifics on what you mean by Unity being insufficient.

Also, developing an engine isn't an easy task. It's a project for itself. Yes, earlier games had their own, but they were a lot simpler than games of today, where you're pretty much obliged to have physics simulations... So people go after off-the-shelf toolset for a reason.

Well, the joint system should be the obvious example here. Then there's the hard 2.5km limit if you want any semblance of precision (and even then we get kraken attacks), basic rigidbody physics being a mess (hello wheels), the default multithreading is unusable, updating your project's unity version will probably break it, the garbage collector is probably still broken (causing random frame spikes at times), and just you wait till we get multiplayer, we're gonna make so many amazing, fun discoveries on that netcode! Also this isn't going specifically into the limitations of the middleware.

There's definitely more, and a lot of it stems from being an engine made for the greenest of the green. It's an amazing engine but it's never been suited to big projects without severely limiting what they can do, and "spaceflight physics simulator with lots of simulated subsistems" is not a thing Unity is fit for.

[cocoscacao]

6 hours ago, PDCWolf said:

"spaceflight physics simulator with lots of simulated subsistems" is not a thing Unity is fit for.

I can't discuss the rest, cause I don't know, but I don't think any engine can handle this any better. I'm very curious what this game is gonna turn into, specifically because of it.

[calabus2]

1 hour ago, cocoscacao said:

I can't discuss the rest, cause I don't know, but I don't think any engine can handle this any better. I'm very curious what this game is gonna turn into, specifically because of it.

A forgotten failure in a few years 

[Icegrx]

On 4/18/2024 at 9:21 PM, Nerdy_Mike said:

I know the speed is not what is being asked, but next week we will have some news to share with more to come in the following weeks.

Could you tell us which day this week?

[The Aziz]

20 hours ago, PDCWolf said:

updating your project's unity version will probably break it,

10/24/2023 at 11:05 PM, Intercept Games said:

Upgraded Unity to version 2022.3.5f1

4 hours ago, Icegrx said:

Could you tell us which day this week?

Friday most likely, like everything.

[MechBFP]

3 hours ago, The Aziz said:

Friday most likely, like everything.

I would say Friday at around 5PM their local time based on past experience. 

[The Aziz]

...which is slightly annoying because a huge chunk of the playerbase (Europe) gets the news actually on Saturday. No point in hyping ourselves for Fridays anymore... It used to be around 8-10PM CET.

[Scarecrow71]

What irks me is that they say "We'll give you information next week", but then they wait until the very last minute to drop it.  Why can't "next week" mean "Tuesday(ish) in the middle of the day so we can answer questions you may have"?  Why does it always have to be late Friday (or, as The Aziz pointed out, Saturday for those not in America)?

[The Aziz]

this is the case of "why change something that works"

 

[Flush Foot]

10 minutes ago, The Aziz said:

this is the case of "why change something that works"

Ahh... the good old days!

[Presto200]

11 minutes ago, The Aziz said:

this is the case of "why change something that works"

 

I wonder if they still have these meetings internally

 

[The Aziz]

Some meetings for sure, can't run a company without presenting some progress internally. But whether Tom Vinita still screams FRIDAAAAAY remains unknown.

[Scarecrow71]

So why not have these meetings, gather the information, and then just release it the following Monday?  I mean, Mike said "Next week", right?

[Mr Shifty]

I will be very surprised if there's any news this week at all. The devs and community managers are Lucy with the football here.