Print

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

[PlatinumSun]

Also maybe make it so you can type a 50 maybe 100 chracter comment (excluding spaces) for each of your votes if you like most of the suggestion but not the whole thing.

Or you already break down all suggestions into there smaller components.
So if someone said lets make aquifers more realistic but also add in sprinklers and pipes. It would be split into too seperate suggestions to vote on. One for sprinklers with pipes and one for more realistic aquifers.

[Bumber]

I'm pretty sure it's abandoned. The list is flooded with old (64-bit DF / retire fort / automine) or rejected (multi-threading) ideas that can't really be removed.
Toady just reads the suggestions forums and puts stuff on his to-do list. Then he gets around to it when he reaches the proper development arc.

There was talk of creating a community-maintained version, but that never really went anywhere.

[PlatinumSun]

Oh. I didnt know. Also why cant multi threading work?

[KittyTac]

Oh. I didnt know. Also why cant multi threading work?

It would take a COMPLETE rewrite of the game. Nobody wants to wait 7 years without any real features. Just rush to mythgen.

[PlatinumSun]

Why would it take a complete rewrite?

[KittyTac]

Why would it take a complete rewrite?

The game is built from the ground up for single-threading. You remove one thing, 5 more stop working. You fix these, 15 more stop working. And so on for years on end.

[anewaname]

@PlatinumSun
Search google with ...

Spoiler (click to show/hide)

site:bay12forums.com "multithreading"

... and you will see some prior discussions regarding this topic. Some helpful, maybe some not so... but all about multi-threading.

[Miles_Umbrae]

As someone who knows a minimum about coding I can safely say that you would NOT want to even TRY to multithread this game by hand, line by line...

[Shonai_Dweller]

Oh. I didnt know. Also why cant multi threading work?

It would take a COMPLETE rewrite of the game. Nobody wants to wait 7 years without any real features. Just rush to mythgen.

I think maybe it wouldn't take 7 years. The design and logic are all in place. Maybe 2 or 3? Not much longer than a Big Wait probably.
Aurora, for example, is being rewritten from scratch in a new language and that guy has a full-time "real job". Seems like that'll only take 3 or 4 years.

The main issue is that there's a chance that it'd be a Big Wait with nothing to show for it at the end. A Multi-threaded Dwarf Fortress will theoretically run better than the current version. But what if its not noticable?

And even if it is, demonstrating how far you can go in pushing the envelope when it comes to procedurally generating fantasy lore, magic, economies, politics, etc, is of far more importance to the gaming world in general than getting a site management program to run quickly. I don't know how much Toady considers that kind of thing when prioritizing though.

Well, I imagine through his side projects Toady will eventually come to a point where he can see for certain a cost/performance benefit of taking the time to do it.

[Bumber]

Toady said the game needs optimizing way more than multi-threading.

The main issue is that DF relies on memory access a whole lot. Multi-threading helps with parallel calculations, but it won't read data any faster, especially since it has to be transferred to another core.

[thompson]

To further Bumber's point, I suspect objects aren't recorded in a particularly memory-efficient way either. This isn't a huge problem for single threaded performance as data can be prefetched by the CPU, however, for multithreading it matters as CPUs have limited internal memory and most multicore CPUs share much of this cache memory. So, if you run two memory intensive processes in parallel they could wind up writing over the cache memory of the other process, forcing more data to be fetched from the RAM. In the worst case scenario you can have calls to RAM almost every cycle, so the CPU spends more time waiting for memory than actually performing computations. I've inadvertently done that in a project of my own - mulrithreading a process slowed it down from about a second to several minutes.

There is hope, however. The map rewrite will be an opportunity for Toady to rethink how space is represented in memory and will hopefully lead to a more light-weight implementation. There could be other processes which are memory-light which could be good parallelization candidates, but I doubt the speedup would be significant without a serious memory data structure overhaul.

[KittyTac]

I don't think that, at most, a 10% increase in FPS would be worth an entire Big Wait.

[Starver]

From a naive Moore's Law expectation¹ a 10% increase in Foo (let's say that's speed, in this case) is gained in 50-100 days². I'd expect quadrupling to beyond-octupling improvements across a Big Wait, just to stand still.

(Assuming I weren't using exactly the same old hardware as I am now, naturally...  )


¹ Yes,  I know. Just running with the point for the sake of it.
² Respectively based upon "double every year" and "double every two years". Which are popular quotes, if you search, though I always used to use 18 months, which happens to be nicely between the two.

[KittyTac]

From a naive Moore's Law expectation¹ a 10% increase in Foo (let's say that's speed, in this case) is gained in 50-100 days². I'd expect quadrupling to beyond-octupling improvements across a Big Wait, just to stand still.

(Assuming I weren't using exactly the same old hardware as I am now, naturally...  )


¹ Yes,  I know. Just running with the point for the sake of it.
² Respectively based upon "double every year" and "double every two years". Which are popular quotes, if you search, though I always used to use 18 months, which happens to be nicely between the two.

It's just that DF is singlethreaded, and modern CPUs get faster by adding more cores, leaving the singlethreaded performance only slightly improved, if at all. So no.

[Starver]

¹ Yes,  I know. Just running with the point for the sake of it.