And this is why people need to stop buying DLC.
Not ready? They're just stringing their customers along at this rate.
Not ready? They're just stringing their customers along at this rate.
-
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.
If they wanted to do something or communicate about it, there would have been an official thread from the developers on the matter a long time ago and not a "megathread"...
I am just sad and jaded about the state one of my favorite game is...
I am just sad and jaded about the state one of my favorite game is...
The lack of communication on the issues of perfomance might be because they don´t have any ideas to 'fix it' if that is even possible. Scrapping everything made from 2.0 and forwards obviously isn't a viable solution and it could be as simple as the game engine not being able to cope with the vast amount of 'background calculations' thar we see in Stellaris.
Personally I'm starting to believe we might see a announcement of Stellaris 2 in a not to far future.
Personally I'm starting to believe we might see a announcement of Stellaris 2 in a not to far future.
For a kind of short-term band-aid fix, I'll suggest to PDX that you could simply switch job calculation from daily to some less frequent interval. After all, most jobs are ONLY relevant at the end of the month, when their resources are gathered. From what I've read on other posts, job calculations are overwhelmingly the driving source of lag. But, as I point out, really its only the jobs at the end of the month that actually matter.
So why re-evaluate jobs daily then?
In principle, for most jobs we could probably evaluate them a single time per month and the job-calculation would do its job, and with only 1/30 of the job-calculation related lag [as a very rough estimate]. But really, we don't have to go that extreme, and even far more modest changes would make a huge difference. Even simply making it check jobs only once every 3-5 days would cut job-calculation lag to 1/3 or 1/5 of its current value, and it sounds like, in principle, that it would massively speed up the game.
@Moah @Jamor, or whomever this would be best to direct to.... would it be possible to test this out with a beta branch that simply introduces a change like this? I don't know the internal workings of the game's code, so perhaps I'm wrong, but it seems to me that [if its programmed in a manner similarly to some codes I have worked with in the past] you might even just be able to change the definition of a single define in your code [frequency of job evaluation] and get a massive improvement in performance. If you were able to put out a beta to that effect, I'm sure that people would jump at the chance to test it out.
So why re-evaluate jobs daily then?
In principle, for most jobs we could probably evaluate them a single time per month and the job-calculation would do its job, and with only 1/30 of the job-calculation related lag [as a very rough estimate]. But really, we don't have to go that extreme, and even far more modest changes would make a huge difference. Even simply making it check jobs only once every 3-5 days would cut job-calculation lag to 1/3 or 1/5 of its current value, and it sounds like, in principle, that it would massively speed up the game.
@Moah @Jamor, or whomever this would be best to direct to.... would it be possible to test this out with a beta branch that simply introduces a change like this? I don't know the internal workings of the game's code, so perhaps I'm wrong, but it seems to me that [if its programmed in a manner similarly to some codes I have worked with in the past] you might even just be able to change the definition of a single define in your code [frequency of job evaluation] and get a massive improvement in performance. If you were able to put out a beta to that effect, I'm sure that people would jump at the chance to test it out.
I can't speak to the complexity of the codebase in Stellaris, but the slow down is likely due to ai doing ai things. There's a lot of ai. There's a lot of events and other things. Without proper management of their updates, they may all be trying to update per frame or even at a synchronised time. if that was to happen, it'd cause a bit of delay every 'tick'(insert relevant time unit into tick).
even then, the kind of slowdown I hear about suggests that something else is going on with those ticks related to the various parts of the game economy, as the slowdown increase happens as the overall game time increases. Best bet, its allllll the stuff the ai is doing as their empire gets bigger. The player manages the cognitive load by reducing it, or offloading it, or having the ai do it(pop management). The ai is probably unable to do that. Or not smart enough to do it, because it's not a mechanicified effect and so it's not accounted for in the ai map.
Also without a profiler I cannot tell you graphically what's happening. As the game time increases, there's more stuff on the map. citadels, fleets, lines drawn everywhere, blobs etc etc etc. An if they aren't handled well slowdown could also happen because frames take longer to draw due to draw calls etc etc etc.
Basically. optimisation is a really hard thing. Any singular fix won't fix it. You often need a lot. An I imagine from these dev's perspective trying to explain that means you get some know it all rocking up saying "it's this, fix this" which isn't useful.
I still think paradox needs to openly focus on this. I would literally buy a DLC that was just some prettier graphics if it meant that the programming team had funding to spend several months just doing all this work. Sure, federations is good. If someone from the team like Jamor told me "buying federations would give me more optimisation budget", id do it. He seems chill, I don't think he'd be giving anyone the bum steer. Of course, at this point, I need someone to actually say it. Dead straight say the words and then deliver, because that's the problem so far, isn't it? no one says it, and then it also doesn't happen.
even then, the kind of slowdown I hear about suggests that something else is going on with those ticks related to the various parts of the game economy, as the slowdown increase happens as the overall game time increases. Best bet, its allllll the stuff the ai is doing as their empire gets bigger. The player manages the cognitive load by reducing it, or offloading it, or having the ai do it(pop management). The ai is probably unable to do that. Or not smart enough to do it, because it's not a mechanicified effect and so it's not accounted for in the ai map.
Also without a profiler I cannot tell you graphically what's happening. As the game time increases, there's more stuff on the map. citadels, fleets, lines drawn everywhere, blobs etc etc etc. An if they aren't handled well slowdown could also happen because frames take longer to draw due to draw calls etc etc etc.
Basically. optimisation is a really hard thing. Any singular fix won't fix it. You often need a lot. An I imagine from these dev's perspective trying to explain that means you get some know it all rocking up saying "it's this, fix this" which isn't useful.
I still think paradox needs to openly focus on this. I would literally buy a DLC that was just some prettier graphics if it meant that the programming team had funding to spend several months just doing all this work. Sure, federations is good. If someone from the team like Jamor told me "buying federations would give me more optimisation budget", id do it. He seems chill, I don't think he'd be giving anyone the bum steer. Of course, at this point, I need someone to actually say it. Dead straight say the words and then deliver, because that's the problem so far, isn't it? no one says it, and then it also doesn't happen.
There is, however, some evidence that vacant jobs daily checking for available pops are responsible for a lot of showdown.
In any case, thread closed in 3...2...1...
In any case, thread closed in 3...2...1...
@Jamor
Considering that, as noted, tests that community members have been doing suggests that daily job checking for pops might be the overwhelmingly driving factor for the lag, would it be feasible to put out a beta branch that simply makes jobs be checked less frequently than daily (even if its only once every 3-5 days)?
It seems like that this might offer one of the quickest and simplest possible ways to get a large performance improvement. Of course, you know more about Stellaris than we do, so maybe there is a very good technical reason why this isn't as feasible as it sounds that you know but we're unaware of. But, lacking the awareness of such a limitation on our end, it would seem to many of us that this could be a very easy and extremely impactive fix.
If you have any thoughts on the matter, that might be very helpful.
Thanks.
Considering that, as noted, tests that community members have been doing suggests that daily job checking for pops might be the overwhelmingly driving factor for the lag, would it be feasible to put out a beta branch that simply makes jobs be checked less frequently than daily (even if its only once every 3-5 days)?
It seems like that this might offer one of the quickest and simplest possible ways to get a large performance improvement. Of course, you know more about Stellaris than we do, so maybe there is a very good technical reason why this isn't as feasible as it sounds that you know but we're unaware of. But, lacking the awareness of such a limitation on our end, it would seem to many of us that this could be a very easy and extremely impactive fix.
If you have any thoughts on the matter, that might be very helpful.
Thanks.
"Lack of communication" is just the current fad honestly. For almost every game with unhappy players, you'll find those threads with the same basic structures: "I love this game, and it has issues, BUT THE BIGGER PROBLEM IS THE LACK OF COMMUNICATION", and this, no matter what the communication strategy is, or how often devs actually communicate.
The truct is that players always want more. They don't realize that we already get weekly dev diaries and dev answers. They don't realize either that what they really want isn't communication, it's promises. What you want to hear is "we're working on it and it will be fixed". But let's be pragmatic: it isn't a good idea to communicate like that until you're actually about to make huge improvements.
If things were simple, then we would already have the fixes. The devs have no reason to address the technical issues of the game.
The truct is that players always want more. They don't realize that we already get weekly dev diaries and dev answers. They don't realize either that what they really want isn't communication, it's promises. What you want to hear is "we're working on it and it will be fixed". But let's be pragmatic: it isn't a good idea to communicate like that until you're actually about to make huge improvements.
If things were simple, then we would already have the fixes. The devs have no reason to address the technical issues of the game.
It was basically a chain reaction. First, it was the disastrous release of 2.2, and then a few months after that there were not a lot of improvements. Followed by silence from PDX. And I think the final straw was the mobile game announcement. Many in the community thought that they were neglecting the PC version in favour of some cheap microtransaction-filled mobile game.
Honestly, I'm just sad about the state of one of my favourite games. I do look forward to Federations, but the performance of the game is one of my concerns about the future of Stellaris.
This game must have the worst performance of all the Paradox games I ever played, even Victoria 2 runs fluently with all the daily money/pop/promotion/work/migration calculations compared to Stellaris.
I too really look forward to Federations, but the lategame... I have some 500 hours in Stellaris, and only got to lategame twice... And that was before 2.0. I just can't stand the lagging even at the lowest possible speed, where even lasers from weaponsfire are literally stopping 'half-way', to wait for the game calculations, also sound effects sometimes finish faster than next one playing, creating silent 'gaps' in the battle, and - who the hell would spent 20 real world minutes just waiting for 1 single in-game year to pass, when you have 15 years truces... and nothing real to do, other than terrible repetetive micromanagament.
I too really look forward to Federations, but the lategame... I have some 500 hours in Stellaris, and only got to lategame twice... And that was before 2.0. I just can't stand the lagging even at the lowest possible speed, where even lasers from weaponsfire are literally stopping 'half-way', to wait for the game calculations, also sound effects sometimes finish faster than next one playing, creating silent 'gaps' in the battle, and - who the hell would spent 20 real world minutes just waiting for 1 single in-game year to pass, when you have 15 years truces... and nothing real to do, other than terrible repetetive micromanagament.
Rendering is likely the largest single source of slow down, so graphics mods can have a large effect on the performance. This might be dependent on the hardware though. Going from 40-50 seconds/month with high graphics quality to 10 seconds/month with minimal graphics is not uncommon. Unfortunately, that seems to be about as fast as the game can run even if it wasn't doing any rendering at all.
If game management doesn't have any significant issues, than yes, this reasoning is valid. However, we are dealing with a game with a multitude of mid-scale bugs that was, at a certain point, released with #TODO lines in the code instead of a important feature.
If you're interrested in the matter, there is this thread where some profiling has been done and shows some of that :
https://forum.paradoxplaza.com/foru...ellaris-a-quick-performance-analysis.1138327/
The root cause of Stellaris problems is that their has been too many differences about what the game actually is about.
Hence we have a mishmash of nods to character management, planet management, pop management, ship design and fleet combat, interstellar diplomacy, and many more, with no clear focus and none of these being properly fleshed out.
At some point, some systems need to be sacrificed in order to narrow the focus on whatever the game is supposed to be.
Personally, I would be happy to have the entire pop system reduced/scrapped/simplified in favor of reduced micro, and a stronger focus on interstellar economics and diplomacy than planetary micro.
And never to be forgotten ... a space based game is really onky about two things: Heroes and Ships.
Any game that wants to claim the label of space based 4x MUST have a strong focus on heroes, starships and space combat. Everything else is fluff.
Hence we have a mishmash of nods to character management, planet management, pop management, ship design and fleet combat, interstellar diplomacy, and many more, with no clear focus and none of these being properly fleshed out.
At some point, some systems need to be sacrificed in order to narrow the focus on whatever the game is supposed to be.
Personally, I would be happy to have the entire pop system reduced/scrapped/simplified in favor of reduced micro, and a stronger focus on interstellar economics and diplomacy than planetary micro.
And never to be forgotten ... a space based game is really onky about two things: Heroes and Ships.
Any game that wants to claim the label of space based 4x MUST have a strong focus on heroes, starships and space combat. Everything else is fluff.
that is v cool. it's not quite what I'm talking about. traditionally, most engines have inbuilt profilers for when you run the game 'in-engine'. Generally, they add a bunch of overhead to your processes, but they capture all the information about what your cpu and gpu is doing and, more importantly, the why its doing it. So instead of unknown function calls, you have very explicit information about what is and isnt causing issues.
now, the Claus?sp engine is not one I've ever worked in, but presumably, it would also have an in-engine profiler. you cant really make judgements about what is and isnt a valid use of the cpu's time until you can adequately work out what functions are calling what and how long they take. presumably, some of that is covered by the non-engine profilers, but I doubt it's got the in-depth information the in engine one should have. Additionally, if you dont do it in engine, im pretty sure you specifically can only look at stack calls, with none of that crucial information. Now, i dont unfortunately have the time to fully read that other thread, so i am admittedly guessing. An im not trying to take away from what those forum folk did. Im just saying that i reserve judgement without that extra info.
it may well be pop job switching. A few select, effective random time selections in there could conceivably make a large improvement to performance. Im not going to just say that though, because at the end of the day, as a designer, this stuff isn't my job. its a programmers, and if they aren't doing that, I would guess its because cash flow is skint.
an anyway, this thread isnt about that. its about the lack of communication. Tbh its not hard to snip actual in engine screen shots of profilerness and explain whats going on. The only logical reason that isnt whats happened yet is that the end reason for not fixing is either "we dont know" or "we cant afford to find out" or both. Id feel far happier buying and recommending dlc if someone would just say that and tell me how i can help.
edit: just as a follow on, the guy above me really illustrates why im reserving any judgement. Like. i dont think ive ever seen a worse take re performance issues and communication, an yet there it is. a post in a thread focused on them, with literally nothing to do with them except to say "i want this game to be this way".
so ill just say, i quite like what stellaris does. i just want to know what i can do to allow them to actually get it working well in late game
Last edited:
If Job checks are really the problem then then the frequency isn't the problem IMO its the technique. Use a dirty flag. i mean, why keep checking if something has changed when you can know: pop grew? migrated? building finished? gene mod done? all of the actions that can change the job and/or laborer pool of planet are already triggers in the game that cause code execution of some sort, simply make them also set the planet as dirty. every day check all planets, and re-evaluate the jobs for any dirty ones. this way you keep the speed to action of checking daily, but most days most planets simply get passed over. The application architecture might make this harder in practice, but in theory its quite simple
I suppose I did stray somewhat from the core argument. My point being that I dont know I trust that the devs know where they want to take the game, hence no clear communication about what their '1 year plan' is.
The game has just grown and grown and new mechanics added at random without a clear vision. No wonder they have no budget (as admitted by the lead designer).
Sure it would be best to do that with in-engine tools. But this piece is already strongly suggesting that game logic and rendering are stepping on each other's toes. So it seems to be a mix of all the things you brought in your earlier post, all at once. That said there's plenty of things enlighted by the community that can improve the perf side of the game without altering it mechanically. There's no reason for it to not stay the game it is while performing well.
To keep it more on topic, one other reason for the lack of fixing&comm could be that they wait to finish their 'big overhaul tour' before fixing, to avoid fixing things they would get rid of later, or destroying the earlier fixes by large changes, and that would mix up badly with willingness to not promise things that are not done yet, and player expectations. After all it was quite clear the game would go through heavy surgery as soon as they announced FTL changes, seeing that a number of game mechanics were not really working that great.
As a lot of people, I'm disappointed by the current state of things, but supposedly the next overhaul should be the last of the largest ones at least, so I'm willing to wait until after the next patch+DLC to see what their plan+communication is before really making up my mind on the devs' attitude towards this. According to the non formal roadmap given by Wiz in dev diary #50 just before 1.4, we might quite have filled what was on the menu then.
Link to the DD in question : https://forum.paradoxplaza.com/foru...llaris-dev-diary-50-the-journey-ahead.978042/
It is just shows how important performance issue is. Look at reddit - performance topics are MASSIVELY more popular compared to others. before it was full of "look at my race\fleet\setup" etc. but now it is boiling about inability to play late game. And now press articles and videos begin to surface and everything with total radio silence from the devs. no wonder people are upset.
Last edited: