Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[galneon]

Dwarf Fortress has performance issues.  Everyone knows this.  Performance has been improved from time to time.  It hasn't improved lately and as complexity increases, it is likely to worsen.

Serious multi-threading is not going to happen.  64-bit DF is unlikely to significantly improve performance and some have speculated it may even hurt performance because of reliance on pointers.

Is this game not a sinking ship that is continually being added to and polished even as its bulkheads burst?

Please, give me a reason to be hopeful.  As it stands, this is a very long story with a likely tragic ending.

[AbstractTraitorHero]

Well if it really comes down to it if performance gets bad enough i can fully see the community helping make dwarf fortress run fast working together there is a tight community here and if toady needs help I can see many of the community working with him to fix it me included.

[galneon]

Well if it really comes down to it if performance gets bad enough i can fully see the community helping make dwarf fortress run fast working together there is a tight community here and if toady needs help I can see many of the community working with him to fix it me included.

This is the only realistic solution, but I cannot see this happening.  Toady either needs to reach out to the community (which would involve showing the codebase to others), or as someone else suggested, set up a separate donation fund dedicated to outsourcing major engine tuning to another development group.

But has he ever shown any inclination toward doing either of these things?

[AbstractTraitorHero]

This seems like toadys dream if it came down to it I'd imagine he'd reach out to well respected and trusted members of the community that can help then make them sign some kind of legal agreement not to share said code without permission and then give them the things they need to know to help."

[King Mir]

DF isn't getting much more performant, but it's not going to get much worse either. Toady does optimize when he adds new features that could otherwise significantly slow the game down, or when people complain about a new release that is very slow (this is usually due to a bug).

If you're hoping to play DF on the newest mobile device, don't count on it. But if you're already playing DF and you're worried about new features slowing down the game, don't sweat it.

[galneon]

I don't care about mobile gaming.  I have a 5820k at 4.4 GHz.  It's not simply that things will get worse in the future--it's that things are already bad enough now.

A successful game ends because the simulation slows to the extent that it's like watching paint dry.  What other game ends for that reason?  None I've played in ~27 years of gaming.  I'd like to retire forts when I feel like it.  Something dramatic has to be done if this is ever to change.

[Crashmaster]

A successful game ends because the simulation slows to the extent that it's like watching paint dry.  What other game ends for that reason?

Aurora - The Dwarf Fortress of 4X Games


It's pretty great too!

[vjmdhzgr]

What I've always heard is that Toady generally doesn't spend much time optimizing unless there's a significant issue or when implementing new features. He plans to save the optimization for later because a lot of that optimization will just get replaced later on. Like if he were to optimize the conversation system in adventurer mode before two years ago, it would have just been a waste of time because a completely new one was implemented.

[MonkeyHead]

Also, performance is relative. A lot of newer DF players express opinions from time to time that 20 fps is "too slow for them", where a lot of older players have learnt to put up with as low as 5 fps. I suppose this all hinges on how "fast" you feel the game should be.

[NJW2000]

Silentthunders is running at less than 1 FPS. Still pretty epic. And if FPS is a concern, you can run a small fort, with a low pop cap, 1x1 embark, etc.

FPS decreases as micromanaging and actual player input increases. And if you get a fort that can run on its own but has ridiculously low FPS, you can leave it on overnight or all day. It becomes less of a game and more of a simulation or odd pet, but still.

[Kirkegaard]

Also, performance is relative. A lot of newer DF players express opinions from time to time that 20 fps is "too slow for them", where a lot of older players have learnt to put up with as low as 5 fps. I suppose this all hinges on how "fast" you feel the game should be.

That is just excuses, from a player point of view. The game should be faster, a lot faster. Nothing is gained by playing in slow motion at 15-25 fps. If it is possible to convert all the ancient code into a more optimal running code is the question.

[FortunaDraken]

It's entirely possible, but it's also entirely the wrong time in the game's life to do so.

> Game gets theoretically optimised
> Continues to expand and get new features
> Optimisation is rendered obsolete due to either being uncompatable with new coding or outright breaking on or due to new features
> New optimisation coding is required, resulting in all the time spent on the previous optimisation a waste

Making the game run better and faster is a job to wait until all the features are in, so there's less coding clashing going on. Once everything is in the game that Toady wants, that is when people should go back and look over it, find ways to make it faster, whatever. There's simply no point wasting time on something that could quite easily be rendered obsolete so quickly.

Besides, I for one would MUCH rather Toady worked on shiny new things and getting rid of bugs then optimising it so I can have giant fortresses.

[Kirkegaard]

It's entirely possible, but it's also entirely the wrong time in the game's life to do so.
...
Besides, I for one would MUCH rather Toady worked on shiny new things and getting rid of bugs then optimising it so I can have giant fortresses.

DF is in continuous development, new features are added and will be until the end of time, there will therefore be no end point when the game is done and can be optimized.  I would rather that Toady spend a year restructuring the code base now, and then continue to expand it than the current slow walk towards total fps dead.

[MonkeyHead]

Also, performance is relative. A lot of newer DF players express opinions from time to time that 20 fps is "too slow for them", where a lot of older players have learnt to put up with as low as 5 fps. I suppose this all hinges on how "fast" you feel the game should be.

That is just excuses, from a player point of view. The game should be faster, a lot faster. Nothing is gained by playing in slow motion at 15-25 fps. If it is possible to convert all the ancient code into a more optimal running code is the question.

See, this just illustrates my point. I do not feel that 15 to 25 fps is "slow motion", and more than tolerable. How fast the game "should" be running is fairly subjective.

[MDFification]

Going multi-thread would not only be a colossal time investment on Toady's part, it's not actually a guaranteed performance boost. Multi-threading done right can be very powerful, but done wrong it can actually slow your program to a crawl if you even manage to keep it stable, and it's specifically dreaded in game programming. Multi-threading is typically only useful if your program has a lot of cache misses slowing it down, which DF probably has, but switching over is such a monumentally complicated, time-consuming and error-prone task that you could have played dozens of games over the years it could take Toady to manage making DF multithreaded without breaking the program.

If you're upset with poor performance, just let computer science advance and hold off playing the game until technology can support a faster DF without requiring unreasonably drastic changes to the program. Heck, within the next decade we could be seeing true commercial quantum computing, memristors or some other crazy invention.