• We have updated our Community Code of Conduct. Please read through the new rules for the forum that are an integral part of Paradox Interactive’s User Agreement.
For what it's worth, there's a new performance/gameplay overhaul mod on the scene. We're still in alpha, so not yet fearure complete, but the mod is coming along very nicely. We've just released our first major update. I've done extensive testing in this new version, and the performance gains are quite significant.

If you're interested, check out Stellaris Immortal on the workshop: https://steamcommunity.com/sharedfiles/filedetails/?id=1891758612

I read the design changes and I have to say I love everything you are trying to do. My only concern is if the AI will cope, and the fact that we will have to retool all mods around your effort.

You hit the nail on the head with reduced pop growth, job buffs, **POP SPRAWL**, reduced job numbers and other things. I'm setting up my callendar to find time to play, once its on release status. Great job, great changes!
 
My only concern is if the AI will cope
Well, I've put a ton of work into making the ai competent. In my ai only test yesterday, they trounced the end-game crisis so hard, I missed it due to being afk for 2 hours.

Of course, I haven't got around to retooling the crisis stuff yet. ;-)

Edit: so far, I'm satisfied that they can run a basic economy, and keep building stuff into end-game... There's a LOT more room for improvement, I think. But already I feel they put out quite a better presence than in vanilla...
 
Well, I've put a ton of work into making the ai competent. In my ai only test yesterday, they trounced the end-game crisis so hard, I missed it due to being afk for 2 hours.

Of course, I haven't got around to retooling the crisis stuff yet. ;-)

Edit: so far, I'm satisfied that they can run a basic economy, and keep building stuff into end-game... There's a LOT more room for improvement, I think. But already I feel they put out quite a better presence than in vanilla...

Sounds good then!!

Can you elaborate more on the design spec for the habitats? I would have expected them to have more "heavy" jobs that provide higher output but with far less pop growth, same for acrologies,ecus and ring worlds. Also is it possible to restrict how many can be build in a system somehow? One of the late game killers is the fact that everyone spams habtats and conquering a system takes hours. I think a limit of 2-3 should do it.
 
Currently habitats and ringworlds work more like their older implementations. I.e. they're like regularr planets with bespoke versions of the city and rural districts. Arcologies get 4 kinds of cities, each with 2 if the same job. This makes them very strong for highly specialized worlds, without increasing Pop density over a regular city-covered planet. Because they're less powerful than vanilla, and more tactical for specialization, we decided to make them available to everyone through a rare tech. The arcology project Ascension perk now gives +1 job per arcology district, and unlocks the tech. This amounts to 50% larger arcologies which is still a powerful benefit, but this perk is no longer the best in the game by such a huge margin...

Edit: sorry for wall of text... I'm on mobile :(
 
Well, I've put a ton of work into making the ai competent. In my ai only test yesterday, they trounced the end-game crisis so hard, I missed it due to being afk for 2 hours.

Of course, I haven't got around to retooling the crisis stuff yet. ;-)

Edit: so far, I'm satisfied that they can run a basic economy, and keep building stuff into end-game... There's a LOT more room for improvement, I think. But already I feel they put out quite a better presence than in vanilla...

Looking forward for your work, thank you !
 
Sorry if this was mentioned earlier in this thread already, but was it changed already that pop jobs get recalculated about 25% per week? Why does this not also has the additional condition that pop jobs only get calculated again if either a new pop came to the planet (growth, resettlement or whatever), one pop is gone, a building is completed or pops got modded. I think those cases are all that would a recalculation necessary at all (assuming different shuffles yield the same result). And that would also mean that in late game many planets who dont see change anymore are now completly free of job recalcs.
 
Sorry if this was mentioned earlier in this thread already, but was it changed already that pop jobs get recalculated about 25% per week? Why does this not also has the additional condition that pop jobs only get calculated again if either a new pop came to the planet (growth, resettlement or whatever), one pop is gone, a building is completed or pops got modded. I think those cases are all that would a recalculation necessary at all (assuming different shuffles yield the same result). And that would also mean that in late game many planets who dont see change anymore are now completly free of job recalcs.

From my understanding reading the thread, there's two problems:

1. Unemployed AI (and slaves, and others) don't obey this
2. Perhaps bigger, there's a mess of situations that can trigger pops to recalculate their jobs outside of the cycle, and it's a lot of trudging work that pdx should do to prevent that from happening
 
2. Perhaps bigger, there's a mess of situations that can trigger pops to recalculate their jobs outside of the cycle, and it's a lot of trudging work that pdx should do to prevent that from happening
But some things should trigger a recalc, what i mean is that unless things change that actually should make a different outcome for jobs possible like a new pop/building/trait then there should generally be no recalc afterwards, because the recalc should yield the same result (yeah there will always be fringe cases but thats life and its better than the game beeing unplayable). If anything they could make two recalc algorithms one okay algorithm for growing planets and one very great but hard to calc that gets used one time on a stagnant planet (no growth, no devastation) and only ever repeated in case there have been changes (resettlement, war whatever) but such a second improved version must also be rarely applied only and not turn to become the standard algorithm (assuming it takes a lot more cpu/memory...). Granted the game then should also be set so stagnant planets will be more likely to exist too (have seen and reported a weird case where some completly overcrowded (no housing and jobs for a lot of pops) and even bombarded planet would grow so that system needs a fix.
 
But some things should trigger a recalc, what i mean is that unless things change that actually should make a different outcome for jobs possible like a new pop/building/trait then there should generally be no recalc afterwards, because the recalc should yield the same result (yeah there will always be fringe cases but thats life and its better than the game beeing unplayable). If anything they could make two recalc algorithms one okay algorithm for growing planets and one very great but hard to calc that gets used one time on a stagnant planet (no growth, no devastation) and only ever repeated in case there have been changes (resettlement, war whatever) but such a second improved version must also be rarely applied only and not turn to become the standard algorithm (assuming it takes a lot more cpu/memory...). Granted the game then should also be set so stagnant planets will be more likely to exist too (have seen and reported a weird case where some completly overcrowded (no housing and jobs for a lot of pops) and even bombarded planet would grow so that system needs a fix.


Of course! Reclaculating is absolutely necessary.

But it also should be avoided whenever possible, and the implication from those who have seemed to put the effort in to research it is that it tends to run into cases where it's triggering far more than necessary.
 
Of course! Reclaculating is absolutely necessary.

But it also should be avoided whenever possible, and the implication from those who have seemed to put the effort in to research it is that it tends to run into cases where it's triggering far more than necessary.
Okay. Overall my point was more that there should not be a standard cycle with relativly quick recalc times at all, but instead it should recalc if this makes sense due to changes and a general recalc should happen over a very long cycle only (as it should normally not yield any different result (except after patches that changed for example a bonus). Sure completly eliminating recalcs is a bad idea but making it say a 10 year cycle when there should be no change (but for whatever reason is anyway event or other overlocked stuff...).
 
Of course! Reclaculating is absolutely necessary.

But it also should be avoided whenever possible, and the implication from those who have seemed to put the effort in to research it is that it tends to run into cases where it's triggering far more than necessary.

I disagree; the game doesn't need jobs, and it doesn't need trade to work how it does.

Without jobs to check for, pops will barely have any impact on the game; just make buildings and districts produce their own resources, rework or remove things like traits, Mining Guilds and unemployment, and limit districts by population like buildings are. Trade--and protection--is probably best off having the starbase range limited to the system it's in, which has the added benefit of promoting smaller empires, at least at first.

To my mind, that's the simplest fix beyond altering the job check and pathfinding update frequencies, which I couldn't find when I looked for them in the files.
 
I disagree; the game doesn't need jobs, and it doesn't need trade to work how it does.

Without jobs to check for, pops will barely have any impact on the game; just make buildings and districts produce their own resources, rework or remove things like traits, Mining Guilds and unemployment, and limit districts by population like buildings are. Trade--and protection--is probably best off having the starbase range limited to the system it's in, which has the added benefit of promoting smaller empires, at least at first.

To my mind, that's the simplest fix beyond altering the job check and pathfinding update frequencies, which I couldn't find when I looked for them in the files.

Yes, I did mean "Absolutely necessary [in the context of the systems we have]".

The mod Stellaris Immortal is doing a pretty neat job overhauling systems to both improve gameplay and performance. I think that's a great direction to go.

As far as what PDX may or may not do, improving the systems they've already put in place and decided on seems like the path of least resistance. I think both can happen concurrently, and neither should be dismissed.
 
The mod Stellaris Immortal is doing a pretty neat job overhauling systems to both improve gameplay and performance. I think that's a great direction to go.

It sounds great, and certainly seems to have more attention put into it than Paradox puts into Stellaris; for games that are continually updated, though, I prefer mods that make as few changes as possible, so that they're easier for the mod developers to keep updated.
 
If you love Stellaris as much as I do and don't care about money:
Get an i9-9900K. It will make Stellaris past 2400 playable, even on 1000 Star maps.

Don't buy AMD Chips for this, as they are at a GRAVE disadvantage here. No idea why, but they just are. Way more than what benchmarks suggest.
Had a 3800X and got about 12 fps in a save. Then switched my whole Platform to a 9900K and went straight to 55fps and considerable more enjoyable experience for the same savegame.
(obviously time to get past a year dropped at almost the same factor)

Spare me the fanboy crap please. If you don't want to believe it, try it yourself.
I can finally enjoy Stellaris Endgame again, without suffering the 2-3fps I had with my old 2700X or the tremendous slowdowns of the 3800X.
 
Spare me the fanboy crap please. If you don't want to believe it, try it yourself.

Fanboy crap incoming. I believe you. To suggest that you should fork out for such an over-powered rig in order to deal with performance issues in a game that are fixable, however, is bollocks.

If that's the rig needed, then that's the specification the game should recommend.

Are you an Intel salesperson? What's your cut?
 
Last edited:
That implies that it should be the multicore architecture where Ryzen is build of complexes of 4 cores that are sharing a L3 cache, and checking with the other complexes if they have it in their L3. Intel uses 2 ring busses sharing all the L3 (possible optimized for 1 cores access to its small slice?).
In effect if it hits in L3 for its own slice Intel wins, if it hits in the 4 core complex AMD wins, else Intel wins.
Similar architectures exists for many consoles, which has a Ryzen or Ryzen like architecture with 2 x 4 cores, and the performance really nose dives if your data is on the other complex.
So as there is a console version of the game PDX might be able to transfer the knowledge from there to PC's.
 
Spare me the fanboy crap please. If you don't want to believe it, try it yourself.

Sure, gona buy a $1000+ processor just because one random guy on a forum told so. No fanboy crap, just upload a video of a non modded game non laging past 2400 with 1000 stars... From all the datas the community gathered till now there's no clue stellaris gonna run smoother on a i9 than on an OC i5

I'm pretty sure 10 physical cores are gonna help just because more cores is better... like... said.... 64 bits better than 32 right ?

:rolleyes: