News:

Want to praise Simutrans?
Your feedback is important for us ;D.

How to make Simutrans-Extended a little bit easier?

Started by Sarrus, May 24, 2018, 07:52:51 AM

Previous topic - Next topic

0 Members and 3 Guests are viewing this topic.

Sarrus

Hello,


At first I must say I love Simutrans-Extended and will never play Simutrans-Standard again. It's first transport simulation game with somehow realistic signaling system. That's really great job!


I like how complex it is, but in result, it is very hard. I changed configuration, so I cannot bankrupt. This allow me to play more than a year. After three years (1913) I am deeeep under 0 and cannot get profit on literaly any line. My local buses sometimes gets full and sometimes runs empty, just like in real life.


My question is how to set values in simuconf.tab to make it easier? I know there is beginner mode, but as far as I know it has limited features (eg factories accepts all amount of cargo). What I'm thinking is to double revenue for passenger or reduce by half maintenance costs. That will give me possibility to play around and learn how things works.

jamespetts

Hello - I am glad that you are enjoying Simutrans-Extended. I should note that the game is not fully balanced yet, but getting to the state of full balancing will take a considerable amount of time, as this will take a huge amount of work: there is a lot of complex coding that has to be done for new features before work can start on the balancing, a lot of difficult bugs to fix before work on those new features can continue, and, once the features are present, a very great number of things to balance.

In the meantime, Dr. Supergood has done some work on interim balancing up until about the 1870s.

There is no setting other than beginner mode that has a global modifier of revenue. It would be fairly easy to alter beginner mode so that it no longer act to cross-connect all industries - I wonder whether this would be useful?

I should be interested in others' views on this.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Sarrus

If you ask me - simutrans should have difficulty settings which affect revenue and/or operation costs. Now game forces me to micro-manage every single line trying to optimize it (actually without success btw :P). I would have more fun if it would be less economically challenging. If you can alter beginner mode and leave only revenue settings I say the big YES. This cross-connect industries is something preventing me from using this mode.

THLeaderH

#3
As far as I know, the most effective way to make a simutrans game financially easier is to raise the fare. The fare can be set by a modification of good.Passagiere.pak.
Once you set the fare twice, the financial balance of your company will be greatly improved  :)

Matthew

Quote from: Sarrus on May 24, 2018, 07:52:51 AM
Hello,


At first I must say I love Simutrans-Extended and will never play Simutrans-Standard again. It's first transport simulation game with somehow realistic signaling system. That's really great job!


I like how complex it is, but in result, it is very hard. I changed configuration, so I cannot bankrupt. This allow me to play more than a year. After three years (1913) I am deeeep under 0 and cannot get profit on literaly any line. My local buses sometimes gets full and sometimes runs empty, just like in real life.


My question is how to set values in simuconf.tab to make it easier? I know there is beginner mode, but as far as I know it has limited features (eg factories accepts all amount of cargo). What I'm thinking is to double revenue for passenger or reduce by half maintenance costs. That will give me possibility to play around and learn how things works.

I am one step ahead of you on the learning curve and completely agree: this is a very fun game, but has a very steep learning curve.

My main tactic is dealing with the financial issues is simply to increase the starting_money setting to 10,000,000¢. (I suspect that the 250,000¢ default setting has a hidden assumption that you are starting in 1750 by running stagecoaches and canals.) That gives you a reserve of capital that you can use in the difficult early years. I then role-play a particular level that is acceptable for initial capital investment (e.g. 1,000,000¢ for a new railway) and try not to fall below that. But if I do, it's not game over. This IMHO is more realistic, because almost all railway projects ever have exceeded their initial capital budget and required fresh capital. I also start up several companies and switch between them, so if one project goes wrong, it's not the end of the whole network. This is also what happens IRL.

Maybe I can share a few other discoveries that I made on these forums and in the game as someone else coming from Simutrans-Standard and that might be helpful. I have only played in the late 19th/early 20th centuries (like you, Sarrus) and on small maps, so these observations may be unhelpful for other circumstances. And I am still pretty new, so I might be flat wrong.


  • This game can only really be played if you leave it unattended (e.g. overnight) or make extensive use of the Fast Forward button. TBH, this really ought to be stated explicitly at the top of the FAQs or Overview of Features because it's completely at odds with every other game I've ever played, where optimal gameplay means pausing every time you go to the toilet because you might come back to find a train has crashed, your enemy has invaded, etc., etc. The timescales in Extended are such that you need to spread your fixed costs over many hours of real-life (not game) time.

  • In Standard (I can't remember which pakset I used to play, but I think it was pak64)  and earlier releases of Extended, it was quite possible to have several bus stops in a village or a small town and run at a profit. Indeed, it was necessary, because even a railway station only had a reach of 3 tiles, so every station required a network of bus stops. In Extended, you need to look at the size of the town or village and ask yourself, 'would I use a bus to travel around this village in real life?' Railway stations can cover a whole town, so the bus networks are used differently.

  • Only passengers are reliably profitable on the default settings. Mail can be profitable, but only as a supplement to a large network. So far, I have had to create a separate 'Post Office' company that runs only loss-making routes (e.g. to connect isolated cities) in order to get profitable routes on my main networks.

  • It's difficult to make profits with freight because there is a major shortage of consumer industries: the game only creates one or two bakeries for each grain mill and one or two pubs for each brewery, but the production industries (e.g. grain mill) are actually designed to feed many more consumers. I create a new consumer industry every three months using the Public Service player, using some role-playing rules (e.g. I place it in the town with the highest City Growth and with the highest number of stops).

  • Industrial connections will break if they are not served at least once every game month. For this reason, you cannot use the 'Wait until maximum load' feature; you must only use time scheduling. If freight journeys take more than one month, you will need more than one convoy, which may often mean taking losses on part of the route.

  • Commuters do not want to travel more than 2 hours. This is a magic cut-off point. So we have to pay attention to it when we are scheduling services.

  • When building railways, pay very careful attention to per-km costs of locomotives and carriages. And think about the size of your map and route. The Great Northern Railway used its rolling stock on routes of 300km. Is your route 300km (2,400 tiles)? If not, you probably can't justify using GNR locos and carriages.

  • Make sure that prices are set to Very Low. You have to have a lot of traffic before you can starting introducing higher classes. I suspect higher classes only really work when you have different Lines for different classes, but I'm not really sure.

I may be doing this all wrong. But these strategies did eventually give me profitable companies in the end.
(Signature being tested) If you enjoy playing Simutrans, then you might also enjoy watching Japan Railway Journal
Available in English and simplified Chinese
如果您喜欢玩Simutrans的话,那么说不定就想看《日本铁路之旅》(英语也有简体中文字幕)。

Sarrus

Quote from: THLeaderH on May 24, 2018, 11:56:33 AMAs far as I know, the most effective way to make a simutrans game financially easier is to raise the fare. The fare can be set by a modification of good.Passagiere.pak.

My short research lead me to this file: https://github.com/jamespetts/simutrans-pak128.britain/blob/master/goods/goods-128.dat. I need to make modification and build it somehow with makeobj right? If am I right, can you tell me what exactly I need to change?

@Matthew thank you for sharing your experience. It's a long post, so I need more time to reply to it ;).

THLeaderH

Quote from: Sarrus on May 24, 2018, 08:17:19 PM
My short research lead me to this file: https://github.com/jamespetts/simutrans-pak128.britain/blob/master/goods/goods-128.dat. I need to make modification and build it somehow with makeobj right? If am I right, can you tell me what exactly I need to change?
It seems to be true. L95 of the file represents the passenger fare.
https://github.com/jamespetts/simutrans-pak128.britain/blob/master/goods/goods-128.dat#L95
I can see the value parameters for specified distances below. These parameter cannot be seen in standard as far as I know.

jamespetts

Matthew - thank you for your detailed analysis - that is most useful. Some of the things to which you refer are intentional (such as the pace of the game - designed to be played in large online games over many months of real time, albeit fast forward can be used in single player mode and only being able to have a profitable 'bus network in a town of a size that in reality would sustain a profitable 'bus network); some of them are as a result of known issues (the lack of consumer industries) or the game not being fully balanced yet; but an issue I am not aware of and I should be grateful if you could elaborate upon them. The issue is this:

Quote
Industrial connections will break if they are not served at least once every game month.

May I ask what you mean by "break" here?
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

DrSuperGood

As mentioned the pakst is only roughly balanced until 1870 at the moment. Even with modifying everything in a spreadsheet it still takes a while to determine fixed cost values as I have to make an estimation on how many employees are needed.

It is worth noting that some vehicles will never be balanced. For example the SS Great Eastern never particularly made profit, a big driving factor for its early scrappage. One certainly cannot fill it with the current passenger generation model in extended, which would be the optimum revenue model. Since such vehicles were never mass produced, it is possible they require special mechanics limiting their number and availability in exchange for other perks.



jamespetts

Quote from: DrSuperGood on May 25, 2018, 11:15:24 AM
As mentioned the pakst is only roughly balanced until 1870 at the moment. Even with modifying everything in a spreadsheet it still takes a while to determine fixed cost values as I have to make an estimation on how many employees are needed.

It is worth noting that some vehicles will never be balanced. For example the SS Great Eastern never particularly made profit, a big driving factor for its early scrappage. One certainly cannot fill it with the current passenger generation model in extended, which would be the optimum revenue model. Since such vehicles were never mass produced, it is possible they require special mechanics limiting their number and availability in exchange for other perks.

I do not think that such idiosyncratic mechanics really make sense in the context of a game that is intended to be a realistic economic simulation. There were no extrinsic factors preventing more than one ship to the design of the SS Great Eastern being built - no doubt, if that design had been a great success, more would have been built, even if not identical in every detail, close enough to be represented by the same vehicle type for Simutrans purposes.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

DrSuperGood

#10
QuoteThere were no extrinsic factors preventing more than one ship to the design of the SS Great Eastern being built - no doubt
Next to economics. The only ship of its kind built was never a big profit turner.

It was designed for non-refueling trips from UK to Australia and back. Except by the time Brunel built it coal was found in Australia and the ship was too big to fit through the Suez Canal, as well as Brunel's bad health. It made no economic sense to run on any route. It had limited use laying cables for the first UK to US telegraphs due to its size, but that really is it and certainly not something simulated in Simutrans. This is why it was scrapped so quickly.

With your argument you are justifying its removal from Pak 128 Britain as a whole...

jamespetts

The word "extrinsic" was intended specifically to exclude economics, as economics are simulated: we do not need to have a mechanism that simulates nothing that exists in reality just to force the same result as can be obtained by what actually occurred. It makes no sense to change the fundamentals of the game to accommodate specific vehicles: either the SS Great Eastern will, when the game is properly balanced and international destinations are introduced, find a niche, albeit perhaps a small one, in which it is a useful vessel, or this ship will not be useful to those players who wish to run a profitable company, but might still be of interest to sandbox players.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Ranran(retired)

In pak128.britain-Ex, IMHO, starting money 250,000¢ may be a small amount of money depending on the starting from modern era.
That would be too much money to buy a carriage, but you can only buy one latest hovercraft at that money.

However, if you set the parameter of simuconf.tab as follows, you can change your starting money automatically by the age you start.

starting_money[0]=1750,25000000,1
starting_money[1]=1900,30000000,1
starting_money[2]=2000,40000000,1

It was added at 110.0 of the standard. It works even with Extended.
ひめしという日本人が開発者達の助言を無視して自分好みの機能をextendedに"強引に"実装し、
コードをぐちゃぐちゃにしてメンテナンスを困難にし(とりわけ道路と建物関連)、
挙句にバグを大量に埋め込み、それを知らんぷりして放置し(隠居するなどと言って)別のところに逃げ隠れて自分のフォーク(OTRP)は開発を続けている
その事実と彼の無責任さに日本人プレイヤーは目を向けるべき。らんらんはそれでやる気をなくした(´・ω・`)
他人の振り見て我が振り直せ。ひめしのようにならないために、らんらんが生み出したバグや問題は自分で修正しなくちゃね(´・ω・`)

jamespetts

Thank you for that. I have incorporated these figures into the pakset, although they are subject to revision once balancing is eventually completed.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Matthew

Quote from: jamespetts on May 25, 2018, 10:23:41 AM
Matthew - thank you for your detailed analysis - that is most useful. Some of the things to which you refer are intentional (such as the pace of the game - designed to be played in large online games over many months of real time, albeit fast forward can be used in single player mode and only being able to have a profitable 'bus network in a town of a size that in reality would sustain a profitable 'bus network); some of them are as a result of known issues (the lack of consumer industries) or the game not being fully balanced yet;

I'm so sorry if the post came across as criticism - that wasn't my intention at all. I understand that the game is made in people's free time and in early access, so all kinds of incompleteness and imbalance is part of the deal. As a new player, I've found it hard to make a profit & had several games that went nowhere, so I thought it worth sharing some strategies that work for me in the interim.

On the specific point of "designed to be played in large online games over many months of real time": I'd strongly suggest that this should be flagged up in the download thread or some other place where fellow-newbies are likely to encounter it at an early stage. If people think they're getting a single-player real-time game (similar to Railroad Tycoon and Transport Tycoon) and they find out that they're getting an multiplayer online-only game (vaguely like SimCity 2013 or World of Warcraft), then they might well be frustrated and disappointed. I have played for several years and hundreds of hours without realizing this. ::'( TBH my brain's not working well at the moment, but a few years ago it was fine and I still didn't realize a fundamental part of the game's architecture. You've designed the game according to the playstyle you enjoy and what you've made is great fun. I'm not saying the online-first multi-month approach right or wrong, just very different from what I expected.

EDIT: The FAQ says that Sim-Ex "can" be played online. But that mislead me. Lots of games can be played online multi-player, but are mostly played offline single-player (IIRC Paradox have said only about 20% of their players ever connect to a server). Sim-Ex is designed to be played online over many months, which is something quite different from most "tycoon" genre games.

Quotebut an issue I am not aware of and I should be grateful if you could elaborate upon them. The issue is this:

May I ask what you mean by "break" here?

This looks like a topic where I'm flat wrong! I was trying (and obviously failed) to describe the phenomenon discussed by Vladki in another thread. Because factories' in-transit limits don't align with the size of freight wagons (and there's no reason they should), and consumer industries are so scarce, it's very difficult to line up freight routes in such a way that factories keep running when you use the wait-for-%-load feature. Perhaps it's not impossible, but I found it so fiddly that it took the fun out of the game. I'm sorry if this wasted your time thinking over a non-existent bug.
(Signature being tested) If you enjoy playing Simutrans, then you might also enjoy watching Japan Railway Journal
Available in English and simplified Chinese
如果您喜欢玩Simutrans的话,那么说不定就想看《日本铁路之旅》(英语也有简体中文字幕)。

ACarlotti

Quote from: Matthew on May 28, 2018, 12:23:15 AMand they find out that they're getting an multiplayer online-only game
I disagree with the implication (presumably worded more strongly than you intended) that this is a multiplayer online-only game. I played through almost 40 years of game-time (starting from 1750) on a map of my own, from about December to February, and found it quite enjoyable. I haven't run this map much since I started contributing to development - so it wasn't issues with playing that stopped me. However, I was running the game almost continuously throughout this (real) time, and was having some trouble with memory/cpu usage as well.

I agree that there are some things that perhaps ought to be mentioned in the faq, or otherwise made a bit more obvious. Firstly, of course, the fact that in Simutrans Extended (and, to a lesser extent, in Standard) the game works best with (at least initially) a low ratio of player interaction time to game time. This can be dealt with using fast-forward, or taking my approach of running it in the background while doing other things. In conjunction with this, it would help if the -nosound flag were better publicised, especially as at present roughly half of the memory consumption of pak128.Britain is taken up by sounds (perhaps these sounds should also be more compressed).

Quote[It's] very difficult to line up freight routes in such a way that factories keep running ... I found it so fiddly that it took the fun out of the game.
It's probably worth adding to the above list that you are not expected to be able to satisfy all passenger/industry demand profitably. I think this is an issue with the player mindset - it's very easy to want to be completionist, particularly if you're playing by yourself and take the view that since noone else will supply transportation, then you have a responsibility to do so. I'm not sure how much that can be alleviated in a single player game.

I think it comes down to the issue that (as I see it) playing a single-player game feels a bit like playing with a 20th century perspective where the roads and railways are funded and supported by the state (albeit with private road traffic, including most lorries and many buses). On the other hand, a multiplayer-game feels much more like how things were in the 19th Century (or earlier), where railways were privately built and run with a lot of competition, and (perhaps at an earlier period?) turnpikes being privately or locally operated instead of the mostly public road netwrk we have now. (Note that all of this is spoken from a British perspective; other countires may differ.)

DrSuperGood

QuoteThe word "extrinsic" was intended specifically to exclude economics, as economics are simulated: we do not need to have a mechanism that simulates nothing that exists in reality just to force the same result as can be obtained by what actually occurred. It makes no sense to change the fundamentals of the game to accommodate specific vehicles: either the SS Great Eastern will, when the game is properly balanced and international destinations are introduced, find a niche, albeit perhaps a small one, in which it is a useful vessel, or this ship will not be useful to those players who wish to run a profitable company, but might still be of interest to sandbox players.
It might serve a niece once one has portals. The ship was intended for non-refuelling round the world travel pretty much (>15,000km range sort of thing).

If it is kept for sandbox usage it should be made clear that is its purpose so as to not confuse players into trying to make a profit with it

jamespetts

A clarification is necessary, I think: I did not mean that Simutrans-Extended was intended only for online play: it can be (and often is) played in single player mode, albeit players will find the fast forward feature (which is exclusively for single player mode) invaluable when playing in single player.

Rather, I mean that the paradigm use case for Simutrans-Extended is an online game. This is because many of the economic features existing or planned are only fully relevant in an environment with competition, and the AI (which exists and works in Standard, but is somewhat crude even there) has not been adapted to work with Extended, so one will be playing in a lonely world if one plays offline.

Of course, one can play single player imagining that one is in a place where all transport is controlled by the state, or one can play as different companies simultaneously in a single player game - but I think that the most satisfying experience is likely to be in an online multi-player game because of the great interest that the complex emergent properties of interacting with other human players and their networks bring.

In relation to industry connexions, had you considered using the wait for % load feature combined with the maximum waiting time limit? The scarcity of consumer industries is an issue that needs to be dealt with, but this is a little way down the queue at present.



As to the Great Eastern - I am not quite sure how one would make clear that certain vehicles (which presumably also include intercontinental aircraft) are of limited use in economic play until the overseas destination feature is implemented beyond simply making clear generally that the game is not yet fully balanced and that there are features waiting to be completed.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Matthew

Apologies for the slow reply on this; I wrote a loooong post earlier in the week that was swallowed by a BSOD.

Quote from: ACarlotti on May 28, 2018, 07:03:38 AM
I disagree with the implication (presumably worded more strongly than you intended) that this is a multiplayer online-only game. I played through almost 40 years of game-time (starting from 1750) on a map of my own, from about December to February, and found it quite enjoyable. I haven't run this map much since I started contributing to development - so it wasn't issues with playing that stopped me. However, I was running the game almost continuously throughout this (real) time, and was having some trouble with memory/cpu usage as well.

Quote from: jamespetts on May 28, 2018, 10:19:03 AM
A clarification is necessary, I think: I did not mean that Simutrans-Extended was intended only for online play: it can be (and often is) played in single player mode, albeit players will find the fast forward feature (which is exclusively for single player mode) invaluable when playing in single player.

Rather, I mean that the paradigm use case for Simutrans-Extended is an online game. This is because many of the economic features existing or planned are only fully relevant in an environment with competition, and the AI (which exists and works in Standard, but is somewhat crude even there) has not been adapted to work with Extended, so one will be playing in a lonely world if one plays offline.

Of course, one can play single player imagining that one is in a place where all transport is controlled by the state, or one can play as different companies simultaneously in a single player game - but I think that the most satisfying experience is likely to be in an online multi-player game because of the great interest that the complex emergent properties of interacting with other human players and their networks bring.

I'm sure that playing online would add greater fun, competition, and realism to the game. But the only working server is Bridgewater-Brunel, which is inaccessible for a machine with only 4GB of RAM. And my Internet connection is very poor, so I suspect that I'd annoy you all with multitudes of desyncs. So no can do.  ::'( Maybe in the future I'll have the brainpower to set up my own small-map server.  ::-\

But I get lots of fun playing bursts of single-player once or twice a year, like you, @ACarlotti. And the language of "paradigm use" is indeed helpful in clarifying your intentions, @jamespetts. It's good to know that single-player has a place in your planning. It's often been pointed out that players of games like this lie on a spectrum between min-maxers (who want to calculate how to 'win', which in this case means having an efficient network) and role-players (for whom the game is a way to generate stories). Simutrans-Extended is a more role-player focused branch of Simutrans, and I'm all about story generation rather than winning, and multiple companies in single-player work well for that playstyle.

Quote from: ACarlotti on May 28, 2018, 07:03:38 AM
I agree that there are some things that perhaps ought to be mentioned in the faq, or otherwise made a bit more obvious. Firstly, of course, the fact that in Simutrans Extended (and, to a lesser extent, in Standard) the game works best with (at least initially) a low ratio of player interaction time to game time. This can be dealt with using fast-forward, or taking my approach of running it in the background while doing other things. In conjunction with this, it would help if the -nosound flag were better publicised, especially as at present roughly half of the memory consumption of pak128.Britain is taken up by sounds (perhaps these sounds should also be more compressed).

You have got us much closer to the heart of the issue here: yes, the key surprise is the "low ratio of player interaction time to game time". It's that, not playing online or multiplayer, which is the really unexpected part of Simutrans, and it applies just as much to SP as MP. Again, I think it would be really helpful if this was emphasized in the download thread or somewhere else where newbies will see it.

QuoteIt's probably worth adding to the above list that you are not expected to be able to satisfy all passenger/industry demand profitably. I think this is an issue with the player mindset - it's very easy to want to be completionist, particularly if you're playing by yourself and take the view that since noone else will supply transportation, then you have a responsibility to do so. I'm not sure how much that can be alleviated in a single player game.

I think it comes down to the issue that (as I see it) playing a single-player game feels a bit like playing with a 20th century perspective where the roads and railways are funded and supported by the state (albeit with private road traffic, including most lorries and many buses). On the other hand, a multiplayer-game feels much more like how things were in the 19th Century (or earlier), where railways were privately built and run with a lot of competition, and (perhaps at an earlier period?) turnpikes being privately or locally operated instead of the mostly public road netwrk we have now. (Note that all of this is spoken from a British perspective; other countires may differ.)

Those are accurate observations, but need to be qualified by the earlier point about playstyles. Some of us are far too poor at minmaxing to be able to serve all industries. But as a role-player, the different 'companies' that I control are fiercely competitive...in my mind! You may regard that as a weak answer, and in a sense it is: I am playing a lot of Simutrans now because my health isn't up to anything more stressful. And I think this is a friendly forum that welcomes all kinds of players, so it's OK.  :)




QuoteIn relation to industry connexions, had you considered using the wait for % load feature combined with the maximum waiting time limit? The scarcity of consumer industries is an issue that needs to be dealt with, but this is a little way down the queue at present.

I have experimented with this feature this week, but I am struggling to get my head around it. Convoys seem to depart at less than the desired percentage before maximum time. But I think I need to try using it on a fresh industry chain which isn't polluted by my usual methods.
(Signature being tested) If you enjoy playing Simutrans, then you might also enjoy watching Japan Railway Journal
Available in English and simplified Chinese
如果您喜欢玩Simutrans的话,那么说不定就想看《日本铁路之旅》(英语也有简体中文字幕)。

jamespetts

Thank you for your feedback - that is most helpful. I have amended the FAQ to make the position clearer regarding single player games and the rate at which time passes.  Hopefully, this will make things clearer to new players.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Vladki

Matthew, I can host a server game for you. Just send me a save game of size the works for you.

Also try continuing the demo game. I had a lot of fun with it, and did not have to fast forward... Just turn freeplay on, and try to get all companies to black numbers.

DrSuperGood

In retrospect I think we can all agree that the massive server game might have been a bit too massive. Its size is good for testing some things, but generally for playing and testing it is a bit inconvenient, inaccessible and even hard to debug. Many cities on the map still have never been touched by transport companies. Even turning on underground mode or changing underground slice takes >10 seconds. The server tick rate has had to be slowed down at least twice already due to some players being unable to keep up with it.

For future test games I recommend maps 1/2 or 1/4 of the size. Still plenty of room but a fraction of the resources needed to run it.

jamespetts

Quote from: DrSuperGood on June 05, 2018, 10:04:00 AM
In retrospect I think we can all agree that the massive server game might have been a bit too massive. Its size is good for testing some things, but generally for playing and testing it is a bit inconvenient, inaccessible and even hard to debug. Many cities on the map still have never been touched by transport companies. Even turning on underground mode or changing underground slice takes >10 seconds. The server tick rate has had to be slowed down at least twice already due to some players being unable to keep up with it.

For future test games I recommend maps 1/2 or 1/4 of the size. Still plenty of room but a fraction of the resources needed to run it.

The size of the server game is a rather different topic - testing has shown that the thing that makes the biggest difference to performance is the number of buildings. One thing that would be likely to assist in this regard is a slower rate of growth for towns (which grow considerably too quickly) as well as fewer starting towns, without needing to make the actual size of the map (which has very little effect on performance) any smaller.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

killwater


jamespetts

Quote from: killwater on June 06, 2018, 08:02:16 PM
Would multitile buildings help then?

Only to the extent that multi-tile buildings replace single tile buildings, which is to say, to some extent. The code for multi-tile city buildings was added a few months ago, but implementing them in the pakset is not something for which I have had time (especially since the new, quick workflow for graphics is not currently available for multi-tile buildings because Tilecutter does not work with transparency; and buildings requiring Tilecutter need considerably more work than single tile buildings in any event).

However, if someone were to work on multi-tile buildings for the pakset, that would be much appreciated.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

killwater

So whole blocks of houses can be replaced with one building? For example if 3/4 of the buildings where multitiled  including 50% 2 tiled and 20% 4 tiled and 5% 6 tiled that should make a big impact. This gives us from 100 tiles of original buildings 25 one tile buildings, 25 two tile buildings, 5 four tile buildings and 1 six tile building. 56 tiles in total and 44% decrease in load if I understand correctly (and make a silly, probably erroneous assumption of linear nature of this dependency). Of course one can enforce merging as long as there are eligible tiles and thus save even more.

Maybe it would even be worth to merge buildings just to save on performance (like a separate thread doing the job in the background and merging changes whenever a city size increases).
Another idea would be to have multitile buildings composed from existing buildings (like a block of different houses) made automatically with existing graphics and doing the merges in a very aggressive manner whenever possible - I think that currently there is a chance based mechanics deciding whether to merge or not. Of course a limit would have to be implemented for a block of houses to not be bigger than lets say 5 tiles in one direction to keep a reasonable amount of sources and destinations.

It would be interesting though if the quality of the simulation would not go down because of it.

ACarlotti

I think it would be better to try to improve the efficiency of the relevant code; I haven't looked at it in detail yet from the perspective of trying to make improvements, but from what I've learnt from the forum and from reading code for other reasons, I think I there are some changes that can be made to the efficiency. At the moment checkng this in detail is lower priority for me than merging stuff from Standard, but perhaps I should be looking at that sooner.

Sarrus

@Matthew thank you for you post. You wrote few things I didn't know.

I started new map with year 1950. I run railway lines througth the middle of cities where possible. I'm more than 4 millions debt but I get profit. So it's not that hard in the middle of 20th century, but 250k starting money it's not enough.

@jamespetts If you consider to implement more difficulty settings it would be great. I have many more questions about game mechanics, but I think I will post new topic instead of offtop this one.

jamespetts

Killwater: I think that I may have been somewhat optimistic with the earlier suggestion that larger buildings would help more than slightly: the thing that takes the time more than anything else is the routing of passengers through players' networks. The more passengers that there are to route, the more time that is taken in this operation. The more (or larger) buildings that there are, the more passengers that there are to route.

Having fewer, larger buildings would make the list of buildings smaller, which would have the effect of reducing the amount of time that the code takes to find a single building in the list, but this is not particularly significant in performance terms.

Andrew - I spent some time last year and the year before improving efficiency in this part of the code and others, even going so far as to multi-thread it, but if you can find further efficiency improvements, that would be most worthwhile.

Sarrus - more difficulty settings (beyond different amounts of starting money by era which has now been implemented) would take quite a lot of work, and there is a very long queue of very substantial tasks of higher priority at present.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Sarrus

Quote from: jamespetts on June 08, 2018, 07:59:55 AMSarrus - more difficulty settings (beyond different amounts of starting money by era which has now been implemented) would take quite a lot of work, and there is a very long queue of very substantial tasks of higher priority at present.
I'm aware of it. Thank you.

prissi

A multitle building is internally handled by the city as the same amount of single buildings unless experimental introduced something other than the patch. I am surprised that buildings can take time up at all. At least in Standard, the CPU is almost independent, apart from the step which iterates over the city to generate passengers. But that rather scales with city size.

Also experimental has a more efficient routing (after spending the effort of generating the connection table). That was the splitting point, Expermiental uses a static table, which needs more calculation power to update but then can easier do a more fine grained routing.

Other than that there are many things which makes Experimental slower, like the physics code, the larger catchment areas, and lots of extra book keeping of convois for the connection table. But those somewhat define Experimental.

jamespetts

Quote from: prissi on June 12, 2018, 06:10:46 AM
A multitle building is internally handled by the city as the same amount of single buildings unless experimental introduced something other than the patch. I am surprised that buildings can take time up at all. At least in Standard, the CPU is almost independent, apart from the step which iterates over the city to generate passengers. But that rather scales with city size.

The handling of multi-tile buildings in Simutrans-Extended (formerly Experimental) was changed in circa 2013 - multi-tile buildings are now treated as single buildings for the purposes of the passenger generation code.

The relationship between buildings and CPU usage in Extended is that more buildings mean more population; more population entails more runs of the passenger generation algorithm in any given amount of time, and the passenger generation algorithm is CPU-intensive. Also, more buildings mean longer global building lists, and the length of the global building lists might also affect computational intensity, although it is the number of passenger generation attempts in any given time period that is the most critical.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.