I dont understand why you cannot lower the frequency of the job - worker checks, even if that decreases economic efficiancy. What about checking it once every two months?
-
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.
This isn't something we can do with modding, but the improvements in Stellaris Immortal are already enough to mitigate pop job selection as a performance bottleneck. The late-game slowdown you still see with the mod comes from military ai during wars, and from rendering the galaxy map. If you're running SI at 2400+, zoom into an empty part of a star system during a period of galactic peace, and watch the days fly by almost as fast as in 2200...
If anyone hasn't watched it yet, it's worth sitting through the PDX talk with John Wordsworth (Technical Director) and Mark Dickie (Engine Team Lead), it covers a lot of PDS history, the current situation and covers the anatomy of a frame, what each thread contains, engine dev team sizes and much more.
https://twitter.com/JohnWordsworth/status/1194910120506998784
https://twitter.com/JohnWordsworth/status/1194910120506998784
Wonderful source. But how does that translate to needing a new engine?
From what I can see of the game files every pop runs through a calculation of potential job candidacy.
For example:
Head Researcher.
If the pop has an owner (is part of an empire), isn’t enslaved, being purged or assimilated they are eligible for the job.
Start with 1000 (because it’s a ruler job, it’s also where the 1000 below comes from)
Add 0.2 x 1000 if they are non mechanical and aren’t a citizen.
Add 2 x 1000 if they have the living standard Academic Privilege.
Add 3 x 1000 if they are erudite.
Add 2 x 1000 if they have logic engines or intelligent trait.
Add 1.5 x 1000 if they have natural engineers,physicist or sociologist.
Add 5 x 1000 if they have the job already.
Add 0.1 x 1000 if they have enigmatic intelligence failed.
Add 2 x 1000 if they have enigmatic intelligence standard or poor.
Add 2 x 1000 if they have presapient natural intellectuals.
Add 2 x 1000 for brain slugs
Add 1.5 x 1000 if they are robust
Add 1.25 x 1000 if they are charismatic or domestic protocols
Add 0.75 x 1000 if they are repugnant
Add 0.01 x 1000 if their controller is swarm or contingency
So... if this scale of calculation is being done for every job on a planet for every pop (so on a 50 pop planet with 50 jobs that’s 2500 calculations) how often is it being done?
Theoretically it only needs to be done just prior to economic production. Upon production of a new pop or migration of a pop, Job creation, suspension or prioritisation It could be suspended until the 28th day.
Would a better way to be internalise these calculation in engine? I’m not a Clausewitz expert but some engines run scripting at a slower rate than native calculations. Rather than having a separate modifier chain for job allocation surely just calculating the production output of the job for the given pop would give a more accurate figure as you’re eliminating the risk of a modifier not being factored due to accidental programmer omission (not to mention making it easier and cleaner for modders). You could also (yes memory usage would need optimising) cache the results, each pop will have the same job output for each job month on month until a tech, building, trait or other modifier is added.
For example:
Head Researcher.
If the pop has an owner (is part of an empire), isn’t enslaved, being purged or assimilated they are eligible for the job.
Start with 1000 (because it’s a ruler job, it’s also where the 1000 below comes from)
Add 0.2 x 1000 if they are non mechanical and aren’t a citizen.
Add 2 x 1000 if they have the living standard Academic Privilege.
Add 3 x 1000 if they are erudite.
Add 2 x 1000 if they have logic engines or intelligent trait.
Add 1.5 x 1000 if they have natural engineers,physicist or sociologist.
Add 5 x 1000 if they have the job already.
Add 0.1 x 1000 if they have enigmatic intelligence failed.
Add 2 x 1000 if they have enigmatic intelligence standard or poor.
Add 2 x 1000 if they have presapient natural intellectuals.
Add 2 x 1000 for brain slugs
Add 1.5 x 1000 if they are robust
Add 1.25 x 1000 if they are charismatic or domestic protocols
Add 0.75 x 1000 if they are repugnant
Add 0.01 x 1000 if their controller is swarm or contingency
So... if this scale of calculation is being done for every job on a planet for every pop (so on a 50 pop planet with 50 jobs that’s 2500 calculations) how often is it being done?
Theoretically it only needs to be done just prior to economic production. Upon production of a new pop or migration of a pop, Job creation, suspension or prioritisation It could be suspended until the 28th day.
Would a better way to be internalise these calculation in engine? I’m not a Clausewitz expert but some engines run scripting at a slower rate than native calculations. Rather than having a separate modifier chain for job allocation surely just calculating the production output of the job for the given pop would give a more accurate figure as you’re eliminating the risk of a modifier not being factored due to accidental programmer omission (not to mention making it easier and cleaner for modders). You could also (yes memory usage would need optimising) cache the results, each pop will have the same job output for each job month on month until a tech, building, trait or other modifier is added.
Check out "Stellaris Immortal" on the workshop. It has many gameplay improvements, ai improvements, and performance improvements.
Nobody at Paradox seems to be talking about switching to a new engine. Just forum randos.
Quite true. I guess I don't care that much. It would just be nice if they fixed a few of the worst bugs someday...
Well in fact, it doesn't, and I'm not in favor of an engine switch. But that rendering thing is still a sign of the engine being old, and it would need some updating on that front at least, as there seem to be quite some room for improvement there, and that would give more leeway for computationally intensive game logic, that PDox seems fond of.
Go easy on the coders. I'm sure some of them *are* amateurs. And as for the amateur modders - the devs working on Stellaris Immortal have years of game design and programming experience.
Also, the Omnimessiah hath blessed SI with an imbuement of the Machine Spirit, for it giveth machines the opportunity to embrace the supernatural.
That's precisely my point....
From what I can read, they must have pleased the Omnimessiah by their faithful deeds.
Last edited:
Aspec's performance circa 2350 is way, way worse than what I get. I wonder what the difference is? I have a PC of a similar age and power to his, with fairly standard hardware (intel, nvidia, etc).
My performance doesn't start to really deteriorate until 2400, and even then it usually stays this side of playable.
Is there any work on fixing the performance on stellaris? its a great game, but without major performance improvements nobody will ever play this game again. With the game slowing down so insanely fast and after just 100years into the game, the game is unplayable at the moment. Would be highly appreciated if you improved this so i can play this amazing game again.
I do agree with Aspec and what he says. Performance is a very subjective thing. The footage of him in the backround would be totally unplayable for me. I have better perfomance too but it still destroys my immersion and fun to play the lategame. It stutters, the symbols are not moving smooth, the windows are lagging: the overall control of the game is slow and unfun. The gameplay is slow and unfun and gets very passive later on.
Stellaris is still a decent game but i can not recommend it as Aspec said. The game has to compete with other games for my playtime and its not doing well atm.
He's got better performance than I do, I'll tell you that. By 2350 it takes 2 seconds for 1 game day to pass for me on the fastest settings without mods. Do bear in mind I only play 1k star galaxies with 24 other AI plus all 5 FEs and 3 Marauders, so I don't expect the same performance on a small galaxy with 5 AI empires, but let's put that 2s/day into perspective.
At 2 seconds per day, 1 month takes 60 seconds, or 1 minute, 1 year takes 12 minutes, and 50 years takes 10 hours. This doesn't include the time where I have the game paused to manage all my planets, the large lag spikes caused by a large exchange of territory after a war, or the extra lag caused by having a heavily populated planet or the species menu being open.
By 2400 however, things get worse. By 2400, the time it takes for 1 day to pass has increased to 3 seconds.
At 3 seconds per day, 1 month takes 90 seconds, or 1 and 1/2 minutes, 1 year takes 18 minutes, and 50 years takes 15 hours. Again, this doesn't include the time where I have the game paused to manage all my planets, the large lag spikes caused by a large exchange of territory after a war, or the extra lag caused by having a heavily populated planet or the species menu being open.
I've only made it to the victory screen once since 2.2 launched because my will to complete a game is invariably crushed by the knowledge that it will take me days of chugging along to reach the end. This is why I just can't get excited about Federations. Why should I care if the fed types will shake up the late game? I'll never get to see it anyway.
I've got an Intel Core i7-6700, an Nvidia GeForce GTX 980, and 16 gigs of Ram. Not a shiny new super computer, but not a pushover either.
Last edited:
He is playing with mods in that video so that's probably a factor.
I put together a new computer recently with a new CPU but the performance still crumbles after 2400 for me as well.
Arrive here who hasn't heard if it yet, please check out Stellaris Immortal on the workshop. It's a massive performance increase, and it improves gameplay in a bunch of ways.
Stellaris Immortal improves performance by, among other things, reducing the population by at least 66% (and much more than this for large empires, because there's a diminishing returns effect for additional planets) and the number of non-capital building slots by 80%. It's an interesting way to alleviate the symptoms, and some people may well prefer the mod's minimalistic approach, but only the Paradox devs can actually cure the underlying disease, which is that the basic system for assigning pops to jobs wastes absurd amounts of computational resources.
Last edited:
Acouple of days ago i would have said that there is little hope for a performance fix but there is some movement building outside of this forum, like the reddit megathread and the video made by Aspec (isn`t he a ex Paradox employee?)... I think public pressure is the only way to make Paradox do something about this.