Are there any plans to try to improve the speed of the mod?

[Lethe]

Well, most limos nowadays have their most blingy features as paid-for optionals. I know that it's not your job to make it a fast mod, but if we, who don't think it would be cost-efficient to invest on a new computer just for playing Paradox Games (and this mod specifically, since all the other games run just fine), had access to the poor men's APD, it would be just great.

It could take a lot of work balancing and would be time disruptive to make two versions, so I'd really appreciate if you could give a couple of simple modding tips we could apply so we could all enjoy APD as it was meant to be enjoyed, specifically how to overhaul what you think is the biggest issue: inert pops, in case it doesn't alter the game balance too hard.

Maybe even stickying a thread where people would share tips and their mod changes to improve performance of their APD wouldn't be such a terrible idea, but that's up to us, the players. I'll probably make a thread tomorrow.

[deanwebb]

I did a little test with my system.

Got APD 3.0.1 set up just fine and started a game as Denmark. First thing, I released Iceland and played as that nation. Focused the map on Iceland and let things go. Only halts to play would be the big tan events, and I'd click through those as fast as I could.

Doing that, 1840 or so was about 105 seconds per year. By the 1850s, the years were already slowing down: 165 seconds per year by 1853. Turning off advanced water got me another 15 seconds - 150 seconds per year in 1856. All told, it took an hour to get from 1836 to 1856. I'm going to collect a few more data points as I read some history today to see if the slowing down is a linear function or if it levels off or takes off like a rocket.

Also, it's bloody boring playing Iceland, unless you're fanatic about exporting fish. The reason I chose it was so I could have a nation not likely to be attacked for the whole of the game. That way, I would not be building any armies. Any slowing down would be all AI.

[s1234567890m]

I did a little test with my system.

Got APD 3.0.1 set up just fine and started a game as Denmark. First thing, I released Iceland and played as that nation. Focused the map on Iceland and let things go. Only halts to play would be the big tan events, and I'd click through those as fast as I could.

Doing that, 1840 or so was about 105 seconds per year. By the 1850s, the years were already slowing down: 165 seconds per year by 1853. Turning off advanced water got me another 15 seconds - 150 seconds per year in 1856. All told, it took an hour to get from 1836 to 1856. I'm going to collect a few more data points as I read some history today to see if the slowing down is a linear function or if it levels off or takes off like a rocket.

Also, it's bloody boring playing Iceland, unless you're fanatic about exporting fish. The reason I chose it was so I could have a nation not likely to be attacked for the whole of the game. That way, I would not be building any armies. Any slowing down would be all AI.

indian puppet with
debug assert might be easier

[deanwebb]

indian puppet with
debug assert might be easier

I thought about the Indian puppet thing, but Iceland is totally out of the way and won't get called upon to defend the Realm.

Time now for the next part of the test... I can check to see if the years run faster after restarting the game.

Well, they didn't. Almost immediately, the game was a turtle: about 4 min/year by 1865, and I have no other programs running... I'm leaving it at 1865 and will see how fast I can get there as a vanilla Iceland... Took an hour and 20 minutes in APD.

Did 1836-1865 in 43 minutes in Vanilla, and the time to finish each year stayed pretty constant, actually dropping to 75 seconds per year in the 1860s. This was also with advanced water turned on all through the test. (Side note: USA slipped off the great power list as Mexico forced the release of Manhattan in the Mexican-American War...) So this shows that from 1836-1865, Vanilla years are constant in time as APD years increase in time length as the game progresses. By 1865, the APD year was about 3 times as long as the Vanilla year.

If the APD years continue to increase in length, that will be a serious performance hit that "buy a better box" can't really solve for me. I know there's a give and take between playability and granularity of detail, and we all have to make the call about where we draw the line.

For APD, the slowing down isn't just at the end of the game: it starts at the beginning and increases throughout, it seems.

[Robotkiller]

I had to call it quits in the late 1870's because of the slow down. Events would momentarily freeze the game before popping up and I tried to delete the massive armies of other great powers to see if it helped. I even boosted assimilation and am using Game Booster to speed things as much as I can. I'll see about reverting brigade size to 3000 or even 5000, but would I have to multiply supply consumption x3 or x5 to balance things?

Another thing that came to mind was province modifiers. Perhaps all of the modifiers in play contribute a lot to the slow down?

[flame7926]

I had to call it quits in the late 1870's because of the slow down. Events would momentarily freeze the game before popping up and I tried to delete the massive armies of other great powers to see if it helped. I even boosted assimilation and am using Game Booster to speed things as much as I can. I'll see about reverting brigade size to 3000 or even 5000, but would I have to multiply supply consumption x3 or x5 to balance things?

Another thing that came to mind was province modifiers. Perhaps all of the modifiers in play contribute a lot to the slow down?

I doubt the modifiers effect much as thy are just sitting there, not being checked by events, or are big, like flags, or changed every day, like pops.

[deanwebb]

I doubt the modifiers effect much as thy are just sitting there, not being checked by events, or are big, like flags, or changed every day, like pops.

Pops gotta be the big resource-users here... any way to get the events, reforms, dismantling in great wars, and new African nations only without the economy/pop changes? Faster factories and defines.lua changes are easy enough to hack on my own. The other stuff is more complex in getting to work right.

[Rylock]

I would suggest trying to revert the army sizes to 3,000 and see how that compares in speed to regular APD, without changing anything else.

[Kokomorari]

With a 2600k the game/mod run really fast. But that's an expensive CPU so it's not really generally used in pc's.

I actually had the mod run slower before, maybe because of bad installation or something.

[Robotkiller]

I successfully made brigades to be 5000 men each, with appropriate adjustments for supply consumption. Not much better with speed, and some nations can't cope with it. They are either overrun by rebels, like Prussia, or consumed by a stronger power. It's only my opinion, but I think new resources are the major cause of slow down. I just wish I could build a supercomputer if I had the money.

[Naselus]

yeah, the unit number increases aren't a big factor, as I've said before. Here's a simple test, tho - just delete all the events and decisions in APD and see how it runs like that. I suspect you'll see a huge increase in speed.

[deanwebb]

yeah, the unit number increases aren't a big factor, as I've said before. Here's a simple test, tho - just delete all the events and decisions in APD and see how it runs like that. I suspect you'll see a huge increase in speed.

I did that with my Iceland game for the year 1869. No events fired, but it still ran about 4 minutes/year, pretty much where it was before I deleted the events and decisions.

[windock]

I have a slightly different problem - game freezes around once in 1.5 months. For a minute or so, with only mouse active. Is specific for me, or it is just how things are supposed to be?

[personduder]

I have a slightly different problem - game freezes around once in 1.5 months. For a minute or so, with only mouse active. Is specific for me, or it is just how things are supposed to be?

Well that definitely isn't normal... Try a clean reinstall maybe?

[videonfan]

For me its kinda slow but playable

[Don_Quigleone]

I've recently come back to playing PDM, now APD, and I can definetely confirm that APD is significantly slower then PDM was in vanilla. In vanilla I had no problem reaching the 1900s, and even finished a game or two, whereas in APD, By 1870 or 1880 the game becomes intolerably slow, with a single day taking 3 seconds(or more...). It's even worse if you save and reload, leading me to think there may be a memory leak somewhere (though that may be paradox's business).

I think it might be worthwhile to study things a bit and see what components of the mod cause the most slowdowns. If one or two particular components are particularly egregious, it may be worthwhile to release them as optional components instead.

I can understand Naselus's view that speed is not a primary concern, and that features should come first, but on the flip side, if the speed continues to decline it's only going to drive players, and potential contributers and testers away from an otherwise great mod.

If it's not possible to easily bring the speed back up, it may be worthwhile for interested players to make an "APD Lite", which would serve both groups.

There may also be imaginative ways to reduce cpu load. There may be ways to reduce the number of pops. For instance currently governments with "residency" have -10% to assimlation, this is likely causing an explosion of different pop types in the americas that immigrate there, and don't subsequently assimilate. By 1880 I have seen provinces where half the population is made up of a rainbow of different nationalities. This may have been a poor decision perfomance wise. I think many would be willing to sacrifice that rule (and others) in order to bump up the speed of their games. That or do not allow conditions to exist where countries can attract large numbers of immigrants, but not be able to assimilate them.

[Naselus]

A little head's up on this - I've got exams coming up next month, so naturally in an effort to avoid doing any revision at all I'll probably be rescripting the ideology, issue and migration tables, which all contain a fair amount of dead weight. Ideology in particular is based on some fairly creaky code from 1.2, and could do with an overhaul anyway. Reducing and streamlining these areas should reduce overhead a bit (I'm not gonna guess how much, but it could be fairly considerable) without compromising content much.

[Don_Quigleone]

Good to hear. I'd also consider examining anything that might cause the creation of inordinate quantities of pops. While having a 50 small ethnic comunities in Stockholm is cool, it doesn't really add to gameplay in any meaningful way.

Either way, we all understand you putting things a bit on hiatus due to exams. It's nothing too urgent.

I might try and do a bit of testing myself, to see how the game is affected by removing particular files.

Seeing as how it takes a long time to run a meaningful test (you really want to see what the speed is in the 1870s and 80s), it may be useful to create a list of tests that people can perform seperately. What's needed is hard data. It's a bit much to expect just Naselus, or another core member to run all these tedious tests, when they're already donating so much of their time as is.

Off the top of my head:

1. Comparison of speed with all events and decisions removed
2. Comparison of speed with all pop tables replaced with vanilla tables
3. Comparison of speed with all PDM specific goods/factories removed.
4. Comparison of speed with all superfluous countries removed.

Someone with a better knowledge of the game might be able to further elaborate on that.

With events/decisions it may be useful to further subdivide things into categories. All tests would consist of timing the length of a month in 1836, 1850, 1860, 1870, 1880 and 1890, while perhaps playing one specific country. Once we see which of these causes the most slowdown, we can home in further and see if there's something specific in those categories. Everyone doing the tests would also need to a benchmark with vanilla APD, and maybe Vicky 2: AHD, for the sake of comparison.

Some of those tests would require extra care though, like removing countries, as it's easy to cause conflicts due to events or history files referencing missing countries. It's also possible that such conflicts may already exist, such bugs may have slipped through, I do no such conflicts don't necessarily cause crashes.

[Naselus]

The main problem with that is, a lot of files tie heavily into others - so removing all the goods will cause a lot of events to crash (and also requires you to weed out the provinces that have been changed); using the vanilla POP tables will cause market chaos and probably also cause crashes from some history files and issue files.

Basically, it's not one thing that causes big slowdowns, it's a combination of EVERYTHING. To cure it, you need to slim all the features a little bit, which is in itself probably more work than releasing a version of PDM at all - you're combing through 18 month's worth of heavily-interconnected work trying to pick bits out here and there. This is why I usually ignore any suggestion of making a 'PDM lite' - it'd be a massive job to do, far outstripping the amount of work needed to produce a new version, since you'd be rebalancing the whole economy, rewriting dozens of events and decisions, filtering out country references from a dozen places... it'd make the job of converting PDM over to AHD look like a picnic, frankly.

[Robotkiller]

I'll probably get a supercomputer anyway. Saw some nice ones that were affordable on Newegg, right on the first page too! (not an ad )