Performance Megathread

[Blayze]

I'll buy Federations, certainly because I'm a dipshit but I'm still hopeful...

If the game continues to be in this poor state after it, I will not invest more money in it unless they have the brillant idea to fix it...

It's my most played game on steam and it saddens me to do so...

If you continue to give Paradox money--for anything--you're just rewarding them for pissing on you.
Find another game--a cheaper, finished, working game--and buy that on sale.

But don't give Paradox any money, whether directly or indirectly.

You're falling victim to the sunk cost fallacy, where you keep giving them money because you've already given them money, and stopping the handover of cash would be like wasting the money you've already spent.

 

[Ecervele]

If you continue to give Paradox money--for anything--you're just rewarding them for pissing on you.
Find another game--a cheaper, finished, working game--and buy that on sale.

But don't give Paradox any money, whether directly or indirectly.

You're falling victim to the sunk cost fallacy, where you keep giving them money because you've already given them money, and stopping the handover of cash would be like wasting the money you've already spent.

I'm already aware of this bias and it's may be that but I really like this game and seeing them butchering this gem makes me sad....
I've already other games but they're not other ones like it...

 

[LWE]

I just don't understand why it's so slow, I thought killing some ai empire might help, but I believe the sheer amount of planets is a big problem. I have an i-9700k, but a normal day takes about 5s. I have the tiny outliner v2 mod running and some minor simple changes, that shouldn't effect the performance. In the save game there are about 700 planets colonized in sum

Also only one core is really used

18340 pops? Now I see why people have performance issues, when I play, my population size at this date (in Large galaxies) is around 2600. The galaxy is full of systems with habitable planets in the save. Do you simply pick a large habitable planets multiplier in settings? I always keep it at the default.

In any case, this seems to confirm the need to change pop job checks to be conducted once a month.

 

[Ecervele]

18340 pops? Now I see why people have performance issues, when I play, my population size at this date is around 2600. Do you simply pick a large habitable planets multiplier in settings? I always keep it at the default.

In any case, this seems to confirm the need to change pop job checks to be conducted once a month.

May be just verify each time where a change occures on a planet and ONLY on pops who are concerned, this could solve the issue.
Because every month you'll still have a lag and may be changes will only be effective at the end of the month.

So yeah it's harder to code than checking everything each day, but optimisation should be a core value of programming

 

[Make Victoria 3]

In any case, this seems to confirm the need to change pop job checks to be conducted once a month.

It takes about 3 months to construct quick buildings (and most require at least half a year) and pop growth isnt much faster either (normally at least) so they really can cut down the calculations for pops while not sacrificing any optimization regarding pops beeing assigned to the best job that the code is going to find at least.

 

[Spartakus]

It takes about 3 months to construct quick buildings (and most require at least half a year) and pop growth isnt much faster either (normally at least) so they really can cut down the calculations for pops while not sacrificing any optimization regarding pops beeing assigned to the best job that the code is going to find at least.

I believe the problem is that the AI tries to get over resource deficits by rearranging pops and jobs (negative food income? we need farmers!) Which is IMHO a terrible idea for entirely different reasons then performance. But it's where we are.

 

[Monturiol]

I believe the problem is that the AI tries to get over resource deficits by rearranging pops and jobs (negative food income? we need farmers!) Which is IMHO a terrible idea for entirely different reasons then performance. But it's where we are.

...and those reasons are?

 

[GnoSIS]

You should give Distant Worlds: Universe a go. A quite in-depth 4X with acceptable AI (miles ahead of Stellaris' AI).

I played it and many other 4x, it was good but it never grabbed the crown from MOO. Although it had a great economy system for its time, planetary development was non existent. The tealtime battles were ok, but in general that game couldn't scale. It always gave you the feeling that you don't control anything which was a huge turnoff for many.

 

[Spartakus]

...and those reasons are?

Sending people to the farms because of a food deficit hides the underlying problem. You don't fix a food deficit by increasing the weight of the farmer jobs, you fix it by creating new farmer jobs on colonies best suited for this and/or colonizing new farming worlds. And you get over the resulting short-term food deficit by using the stockpile or buying on the market. The average joe on a world that happens to have free farming jobs is entirely unsuited to fix empire-wide problems. But by taking a farmer job when he is actually supposed to be a scientist/miner/whatever he masks the deficit with a reduced output in another resource just long enough to reach the point where no jobs are left or a second resource runs into deficit. And then we get daily job-swapping.

Seriously, I think the best solution without a complete economy redesign would be to assign pops to jobs randomly and redoing this for a small portion of pops every now and then.

 

[LWE]

Random assignment wouldn’t work for human players, though, and I do not think the economy needs a redesign. Just have the AI use the market.

 

[GnoSIS]

Sending people to the farms because of a food deficit hides the underlying problem. You don't fix a food deficit by increasing the weight of the farmer jobs, you fix it by creating new farmer jobs on colonies best suited for this and/or colonizing new farming worlds. And you get over the resulting short-term food deficit by using the stockpile or buying on the market. The average joe on a world that happens to have free farming jobs is entirely unsuited to fix empire-wide problems. But by taking a farmer job when he is actually supposed to be a scientist/miner/whatever he masks the deficit with a reduced output in another resource just long enough to reach the point where no jobs are left or a second resource runs into deficit. And then we get daily job-swapping.

Seriously, I think the best solution without a complete economy redesign would be to assign pops to jobs randomly and redoing this for a small portion of pops every now and then.

Yes, the idea of a totaly liquid workforce is a huge mistake. Its not only unrealistic, given that a person needs a lot of training even by todays standards to swich employment, but it also places a burden on performance since pops have to evaluate constantly the weights to switch jobs. Economies find an equilibrium because of sortage and surplus of goods, services, jobs, population and skills.

A workforce that won't change jobs from day to day is the solution to both problems, since it will make time for the calculations to take place and by switching jobs to smaller amounts of pops in an effor to find equilibrium.

 

[gbem]

hey guys... ive played stellaris since 2.3.3 but decided to quit due to extreme lag in the lategame... has there been any improvement whatsoever in the new patch?

 

[GnoSIS]

hey guys... ive played stellaris since 2.3.3 but decided to quit due to extreme lag in the lategame... has there been any improvement whatsoever in the new patch?

No, and if they don't do anything it will get worse with the new DLC. I'm switching to 2.1.3 once I find some of the old mods I used to run.

 

[AlknicTeos]

18340 pops? Now I see why people have performance issues, when I play, my population size at this date (in Large galaxies) is around 2600. The galaxy is full of systems with habitable planets in the save. Do you simply pick a large habitable planets multiplier in settings? I always keep it at the default.

In any case, this seems to confirm the need to change pop job checks to be conducted once a month.

I always take 5x habitable planets.
Anyway if you compare amount of colonized planets, and the amount of provinces in eu4, ck2, rome 2, imho it shouldn't be that slow

 

[GnoSIS]

I always take 5x habitable planets.
Anyway if you compare amount of colonized planets, and the amount of provinces in eu4, ck2, rome 2, imho it shouldn't be that slow

What are your system specs?

 

[LWE]

I always take 5x habitable planets.
Anyway if you compare amount of colonized planets, and the amount of provinces in eu4, ck2, rome 2, imho it shouldn't be that slow

Stellaris planet isn't really an equivalent to EU4 or even Rome province, considering their large populations and that the game calculates each pop as a separate entity, checking (rather irrationally) whether it needs to change jobs every single day... There are only c. 4000 provinces total in EU4, compared to 18000+ pops in your save.

 

[The Conmmander]

I always take 5x habitable planets.
Anyway if you compare amount of colonized planets, and the amount of provinces in eu4, ck2, rome 2, imho it shouldn't be that slow

Which year usually ends a game and do you play a genocidal empire?

 

[LWE]

Which year usually ends a game and do you play a genocidal empire?

You can look at the save AlknicTeos provided. Afaik, their empire is not genocidal and the endgame date is the standard 2500.

 

[The Conmmander]

I have an i-9700k, but a normal day takes about 5s.

Ah sry, no questions to 5x habitable planets anymore.

 

[LWE]

Having said so, since I haven't played Imperator in a long time (only one completed game shortly after release) what are the endgame population statistics there nowadays, and how is pop migration there calculated? Does each pop consider migrating to any province of the map every single day?