Print

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

[cochramd]

(removed)

[Orange Wizard]

~~~

Please don't bicker

[Fayrik]

I'm sure someone will correct me if I've forgotten something here, but I'm fairly certain that there are only four areas of Dwarf Fortress that cause considerable FPS drops.

• Creature Pathing and Decision Making
• Fluid Dynamics
• Heat Calculation
• Weather

I can't think of any other parts of the game that can come close in comparison to the amount of time these parts of the simulation use. They're the heavy numbers behind everything, after all.
So, if you limit your pop-cap to 20, ensure there's no large water bodies, turn off all the cavern layers and underground features, and switch off weather for good measure... You'll find a fort like this would take much much longer to slow down at all.
But even at this hyper optimized super fast fort, it would eventually slow down, as you modify the environment and accrue dead creatures and such.
Trying to make a single game last forever is never going to work, due to the nature of forever.

Basically Dwarf Fortress can either be played with everything switched on for a short while, or with everything switched off for a long while. Either way there is a very real upper limit the length of time you can run a single fortress for, and that's not in any way a fault of the game, but rather fault of the universe our computers exist in.

In fact, even the most theoretically ideal multi-threaded environment could be brought to a very painful halt when the entirety of the goblin race decides to invade.
(Although, in practice it would likely suffer a more boring slowdown, as hundreds of dwarves interact with a limited set of items, grinding everything to a halt in a hideous mess of memory locks.)

Spoiler (click to show/hide)

Alternatively skip locks and enjoy the !!fun!! of unsafe threading.

[Untrustedlife]

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.

Toady has made 2 year releases before, he can do it again too.
Actually he has a well laid out plan for the game (ON NOTECARDS believe it or not) and the 42, means its 42% done, when that hits 100% its done and he can optimize. SO yes there is an end.It might take 20 years, but there is a n end.
And even if optimizations take 2 years people will still play the game.

Optimizing the current path algorithm isn't a good investment yet actually as toady plans to implement parallel worlds/other planes and this will require a massive pathfinding rewrite anyhow.

[Orange Wizard]

Toady has made 2 year releases before, he can do it again too.

Except he doesn't want to, because he didn't like it, the fans didn't like it, and the donations dropped significantly over that period. Toady has said many times that he wants to make smaller updates with a shorter cycle.

Actually he has a well laid out plan for the game (ON NOTECARDS believe it or not) and the 42, means its 42% done

That's not how version numbers work. 0.42 means Toady has made forty-two major versions of the game thus far. It could finish at 0.85, or go on to 0.1337. It is not a percentage.

[Untrustedlife]

Toady has made 2 year releases before, he can do it again too.

Except he doesn't want to, because he didn't like it, the fans didn't like it, and the donations dropped significantly over that period. Toady has said many times that he wants to make smaller updates with a shorter cycle.

Actually he has a well laid out plan for the game (ON NOTECARDS believe it or not) and the 42, means its 42% done

That's not how version numbers work. 0.42 means Toady has made forty-two major versions of the game thus far. It could finish at 0.85, or go on to 0.1337. It is not a percentage.

You are wrong, his version scheme is based on 100 and it is also based on core components 100 core components is version 1.0 . http://dwarffortresswiki.org/index.php/DF2014:Version_number http://dwarffortresswiki.org/index.php/v0.34_Talk:Version_number
it is essentially percentages. And toady has even  stated this. which is why he was so happy with 34 he was like we are now over a third done (listen to the df talk podcasts)

also

Toady also said that they need make a new list of what needs to go in before it qualifies for v1.0.0 status so they can better estimate that "percentage" number for future versions (I think this was in one of the DFTalk podcasts).

Since this quote toady did that. and it is now in fact approximating percentages.

[Orange Wizard]

Oh, huh. Didn't expect that one.

[Libash_Thunderhead]

I am currently playing on an old 32-bit pc. I'm statisfied when fps reaches 30 after 5 years of in-game time. And I seldom play maps larger than 2x2.
When I ran the game on another pc which had a better cpu (i3-2120, not too good but still better than that Pentium D), my first reacton was "wow, everything is so fast, let me try a bigger map".

I'm curious: performance aside, how big do you think a typical df map should be? How big the world map should be? How long the history should be?

Fortress map: 4x4, 8x8, 16x16, or even bigger?
World map: medium, large, very large or ...?
History: 200 years, 500 years, 1000 years, or ...?

[Killgoth]

This technology will come to processing speed in no time:  http://www.pcgamer.com/experimental-5d-data-storage-could-store-360tb-of-games-for-138-billion-years/

[Arbinire]

Alpha game is Alpha, other than a general bit here and there to keep the game running relatively smooth, optimization is not a concern at all right now, Mr. Escaped Lunatic.  And no, these posts don't need to be brought up frequently, it is not helpful, altruistic, or for the good of the game, and no it is not doomed, technically or otherwise.

This post is just as asinine as the UI and Graphics posts claiming that if DF doesn't make these adjustments right this minute then the game is doomed to fail, and those posts have been popping up for years now, yet the game is still going strong in a functioning manner despite the cries that the sky is falling.

[Col_Jessep]

This technology will come to processing speed in no time:  http://www.pcgamer.com/experimental-5d-data-storage-could-store-360tb-of-games-for-138-billion-years/

That's only for data storage, think of it as a high-capacity DVD/Blu-Ray. It won't speed up the processing of the data. It means that we might be able to store an entire galaxy of generated DF worlds with millions of years of history onto one single disk. Enjoy the lag! =)

[Kirkegaard]

I think it is a mistake to look at it as "multi threading or no changes at all" there are many other ways to optimize performance. It is my impression that most of the basic code is +10 years old, and with 10 years more experience I would find it profoundly odd if there was not rather large gains in both speed and general code structure to be gained by having a look at it and "cleaning the code".

A entire other option is to remove some of the features designed over the years, like the dynamic world, in principle it is a very cool feature, but when playing a fort. Do one really care if the rest of the world is changing? As I remember it, that feature did cost something around 20% of the fps. 

[MDFification]

I don't care if improvements come from the most brilliant implementation of multithreading in gaming history or some other means.  I just know something dramatic must be done, and that doesn't involve waiting for hardware to improve in a magical way that makes an inefficient engine behave otherwise.  I hope the options, however daunting, are being considered.

The problem isn't that DF is inefficient (though it could stand to be more efficient, although optimizing early is problematic), the problem is that it's setting out to do a ludicrously large, complex simulation. It's now looking like versions in the far future won't be able to run on 32bit. As the complexity of DF continues to escalate as planned, you are not going to see dramatic improvements in performance without better hardware, even if Toady decides optimizing a partial alpha build is worth making his job more difficult.

The major causes of FPS death that the community is aware of are pathing, fluid dynamics and changes to an item's state (contaminants, temperature changes etc). Of these, Toady is unsatisfied with fluid dynamics already, and finding a more efficient way to do the other two isn't easy. This isn't in the sense that only an expert coder could do it; there's just only so many methods available, each with their own benefits and drawbacks, and very few of them are suitable for DF.

[Witty]

... "this is toady's baby, he'll inevitably work on fps issues as it starts bothering them". Does this mean that maybe he'll spend a whole year rewriting fps blockers? Probably.

This is pretty much the only answer on the subject, imo. The Great FPS Death is coming, but it'll get fixed. Even if it takes a two or three year rewrite to do so.

Is that ideal? Of course not. But that's almost certainly what's going to happen. And I don't really think the community will collapse because of it. Maybe some will quibble a bit - but that's always been the case, and there's nothing wrong with that.

And as others have already said - neither 64-bit or even multithreading will have a huge impact on improving DF's current FPS issues. Fayrik already listed the major FPS killers. Once those get some work - long term FPS will improve dramatically.

[Willfor]

A entire other option is to remove some of the features designed over the years, like the dynamic world, in principle it is a very cool feature, but when playing a fort. Do one really care if the rest of the world is changing? As I remember it, that feature did cost something around 20% of the fps.

Hey, yeah, why don't we remove the primary feature that makes DF so great? The thing that none of the DF clones actually even try to copy? The part of the game the Toady actually really, truly enjoys making, and part of what makes it a game worth featuring in the museum of modern art? That'll really help development along.

:V