- v50 information can now be added to pages in the main namespace. v0.47 information can still be found in the DF2014 namespace. See here for more details on the new versioning policy.
- Use this page to report any issues related to the migration.
Difference between revisions of "40d Talk:System requirements"
m (→Benchmarks) |
(sign) |
||
(11 intermediate revisions by 7 users not shown) | |||
Line 1: | Line 1: | ||
+ | == Linux worthless, Pre-Compiled, Why? == | ||
+ | |||
+ | This is linux, where things aren't always the same. | ||
+ | Pre-compiled software doesn't work. | ||
+ | |||
+ | ARM Processors? | ||
+ | |||
+ | |||
== DF Doesn't Need a Fancy GPU == | == DF Doesn't Need a Fancy GPU == | ||
Line 59: | Line 67: | ||
It would be really useful to use one or more specific fortresses for benchmarking purposes and post them all on DFFD, rather than just including various details which don't include certain vital statistics such as the number of objects, amount of revealed tiles, mid-air Z-levels which have been "allocated" by building into them (such that using "k" actually shows them as "open space" rather than nothing at all), and actual fortress layout (and thus how much CPU time is spent pathfinding). For example, a 2.4GHz Core 2 Duo is listed as handling a minimalist 6x6 site with 95 dwarves at ~20fps, while my 2.4GHz Core 2 Quad was able to support an 8x4 site with magma + pit + underground river + HFS and 200 dwarves and get 20fps with everything turned '''on''' because I had taken care to organize the fortress in such a way as to minimize pathfinding load. --[[User:Quietust|Quietust]] 18:40, 20 January 2010 (UTC) | It would be really useful to use one or more specific fortresses for benchmarking purposes and post them all on DFFD, rather than just including various details which don't include certain vital statistics such as the number of objects, amount of revealed tiles, mid-air Z-levels which have been "allocated" by building into them (such that using "k" actually shows them as "open space" rather than nothing at all), and actual fortress layout (and thus how much CPU time is spent pathfinding). For example, a 2.4GHz Core 2 Duo is listed as handling a minimalist 6x6 site with 95 dwarves at ~20fps, while my 2.4GHz Core 2 Quad was able to support an 8x4 site with magma + pit + underground river + HFS and 200 dwarves and get 20fps with everything turned '''on''' because I had taken care to organize the fortress in such a way as to minimize pathfinding load. --[[User:Quietust|Quietust]] 18:40, 20 January 2010 (UTC) | ||
− | :I like this idea overall. I don´t think it would be possible to assume what an "average" person builds their fortress like or if they even take pathfinding into account, but a standardized fortress, set of options, DF version, and t/f on use of OpenGL "accelerator" would make this list much more insightful. Unfortunately, it would throw out all previous data collected. If someone has the will and time, I think it would benefit readers, but I think the section does already serve its purpose in giving people a very rough idea of what to expect. In the data I personally posted, I did use the same fort (which paid no attention to optimizing the layout for minimizing pathfinding calculations) but I didn´t think to upload the save and no longer have it. | + | :I like this idea overall. I don´t think it would be possible to assume what an "average" person builds their fortress like or if they even take pathfinding into account, but a standardized fortress, set of options, DF version, and t/f on use of OpenGL "accelerator" would make this list much more insightful. Unfortunately, it would throw out all previous data collected. If someone has the will and time, I think it would benefit readers, but I think the section does already serve its purpose in giving people a very rough idea of what to expect. In the data I personally posted, I did use the same fort (which paid no attention to optimizing the layout for minimizing pathfinding calculations) but I didn´t think to upload the save and no longer have it.[[User:Kludge|-K]] 03:37, 7 February 2010 (UTC) (a post-post thought -- It´d also be nice to have standardized data since the data could be merged into a pretty table!) |
== Where do I get the Linux port? == | == Where do I get the Linux port? == | ||
Well, the headline basically says it all. It's not linked on the official download page... it took quite a bit of searching until I found anything -- which I don't want to post here, as I don't know whether what I found is in any way recommendable. In a nutshell: the section about the linux port should probably start with '''where to get it'''. --[[Special:Contributions/93.104.144.228|93.104.144.228]] 16:33, 4 February 2010 (UTC) | Well, the headline basically says it all. It's not linked on the official download page... it took quite a bit of searching until I found anything -- which I don't want to post here, as I don't know whether what I found is in any way recommendable. In a nutshell: the section about the linux port should probably start with '''where to get it'''. --[[Special:Contributions/93.104.144.228|93.104.144.228]] 16:33, 4 February 2010 (UTC) | ||
+ | |||
+ | When in doubt, check the [http://www.bay12games.com/forum/index.php?topic=40349 forum announcements.] | ||
+ | Or [http://www.bay12games.com/forum/index.php?PHPSESSID=747e4a435324f30c2d12bf753bb62482&topic=44829.0 this one, which isn't announced,] if you're feeling particularly daring.[[User:Mephisto|Mephisto]] 06:04, 9 February 2010 (UTC) | ||
+ | |||
+ | == Faster on Linux? == | ||
+ | I wanted to know if anyone else can verify this. I am dual-booting Vista and Ubuntu (9.10) on my laptop. I run the same save file of 40d17 in Vista I get something like 20-30fps, I run 40d17 in Ubuntu I get 70-80fps. Is anyone else having similar results? [[User:Kenji 03|Kenji 03]] 02:35, 21 February 2010 (UTC) | ||
+ | |||
+ | Linux being less resource intensive than windows in most regards, even lacking completely capable (Though nouveux is pretty good) drivers for every piece of hardware, should generally give better performance on anything, Dwarf fortress being very resource heavy should see this as well. <small>– [[template:unsigned|unsigned]] comment by [[User:70.190.4.219|70.190.4.219]]</small> |
Latest revision as of 16:24, 24 November 2013
Linux worthless, Pre-Compiled, Why?[edit]
This is linux, where things aren't always the same. Pre-compiled software doesn't work.
ARM Processors?
DF Doesn't Need a Fancy GPU[edit]
Even considering how DF uses OpenGL for graphics, an integrated card should have absolutely no problem displaying 2000 quads on screen.
I threw together a quick and dirty test app to render 2000 dwarf faces, and it can manage 100 FPS with just 18% CPU load. Without throttling I get 600 FPS with 80% CPU. That's on an integrated (GMA x3100) card. So, at least for Intel's GMA cards going back at least two years, the graphics should perform just fine.
So that's why I don't think it should be necessary to recommend a powerful discrete GPU, especially when the screen isn't updated very often.
--Jcromartie 13:31, 9 May 2008 (EDT)
- It's specifically stated that an old GPU will work just fine. Just a bargain bin 25$ card would be sufficient for the purpose. A BIG reason for using a separate GPU is the fact that they will share resources, (Shared Memory *and* CPU Cycles,) which will quickly affect performance of the CPU, which DF is quite power-hungry for at this stage, (and many future stages as well I'm sure.) --Edward 19:39, 9 May 2008 (EDT)
- Not to mention a seperate GPU takes ALL those 2000 quads off the processor. --Darkone 20:35, 12 May 2008 (EDT)
General Performance[edit]
I may be talking out of my butt here, and as a developer I can understand the challenges of (and reasons to avoid) optimization, but might it be time to start tuning DF? It uses 99% CPU on a 2.2GHz Core 2 Duo all the time. This is even the case when displaying the main menu or when paused.
Maybe this article can shed some light on why it uses so much CPU when it doesn't need to?
--Jcromartie 13:31, 9 May 2008 (EDT)
- I can see where you're coming from, but being that the majority of the game has yet to even be programmed, it's far too early to begin real optimizations. Toady's made a few tweaks to do minor optimizations, so he's got it in mind, so far as it won't interfere with the many core and other items that still need to be added. Over half of the http://www.bay12games.com/dwarves/dev_core_1-20.html Core Components are still missing, and it's even farther from completion when looking at the http://www.bay12games.com/dwarves/dev_req_1-50.html Non-Core Required Components, a good number of which deal directly with optimization. --Edward 19:33, 9 May 2008 (EDT)
Dual Core[edit]
I believe it may be right to add back in a recommendation, but not for multiple cores but the processor itself. Someone made an erroneous edit in the history stating that a single core is usually faster then a dual- this is wrong, as the Core 2 architecture is Far more powerful then Pentium, though I do not have data for AMD processors. A Core running at a mere 1.86 gigahertz pulls in a 3000+ benchmark (cinebench) versus the lowly 1800-2000 of a 3 gigahertz pentium 4. Not to mention running DF on its own private core is probably the best thing you could do with it.--Darkone 10:45, 12 May 2008 (EDT)
Someone please check if DF might be multi-threaded after all. At least on my machine (q6600) DF quite beautifully utilizes all four cores with individual loads usually not maxing out (110 dwarves). 1 February 2010
Small Edit[edit]
I changed the Mac OsX portion of the page to a more current update.
init.txt[edit]
I think there should be some mentions of init.txt here and on Maximizing Framerate. I found that there was a considerable difference when I changed the init.txt.
I found that, by changing [SOUND], [PARTIAL_PRINT], and [PRIORITY], I was able to significantly increase performance.
While running 66 processes (DF as well), including iTunes, Firefox, Desktop Sidebar, and Last.fm, the game's default init.txt was very poor. It lagged frequently, very much so in Adventure mode. I changed the aforementioned values in init.txt and it now runs wonderfully, with the exception of world creation, entering towns, and a very small lag when placing buildings.
My FPS before was only at 100 FPS when paused, usually around 50-65 when playing. Now, I seldom see it anything other than 99 or 100.
My system is as follows:
- Windows XP SP3
- Hard Disk: 83.2 GB free, 140 GB total
- AMD Sempron 3300+ processor
- 448 MB RAM
- In short, This, bought Dec 15 2006.
Linux system Requirements?[edit]
If you run Linux over Windows, does it enhance your FPS any?Kenji 03 12:30, 2 August 2009 (UTC)
Benchmarks[edit]
It would be really useful to use one or more specific fortresses for benchmarking purposes and post them all on DFFD, rather than just including various details which don't include certain vital statistics such as the number of objects, amount of revealed tiles, mid-air Z-levels which have been "allocated" by building into them (such that using "k" actually shows them as "open space" rather than nothing at all), and actual fortress layout (and thus how much CPU time is spent pathfinding). For example, a 2.4GHz Core 2 Duo is listed as handling a minimalist 6x6 site with 95 dwarves at ~20fps, while my 2.4GHz Core 2 Quad was able to support an 8x4 site with magma + pit + underground river + HFS and 200 dwarves and get 20fps with everything turned on because I had taken care to organize the fortress in such a way as to minimize pathfinding load. --Quietust 18:40, 20 January 2010 (UTC)
- I like this idea overall. I don´t think it would be possible to assume what an "average" person builds their fortress like or if they even take pathfinding into account, but a standardized fortress, set of options, DF version, and t/f on use of OpenGL "accelerator" would make this list much more insightful. Unfortunately, it would throw out all previous data collected. If someone has the will and time, I think it would benefit readers, but I think the section does already serve its purpose in giving people a very rough idea of what to expect. In the data I personally posted, I did use the same fort (which paid no attention to optimizing the layout for minimizing pathfinding calculations) but I didn´t think to upload the save and no longer have it.-K 03:37, 7 February 2010 (UTC) (a post-post thought -- It´d also be nice to have standardized data since the data could be merged into a pretty table!)
Where do I get the Linux port?[edit]
Well, the headline basically says it all. It's not linked on the official download page... it took quite a bit of searching until I found anything -- which I don't want to post here, as I don't know whether what I found is in any way recommendable. In a nutshell: the section about the linux port should probably start with where to get it. --93.104.144.228 16:33, 4 February 2010 (UTC)
When in doubt, check the forum announcements. Or this one, which isn't announced, if you're feeling particularly daring.Mephisto 06:04, 9 February 2010 (UTC)
Faster on Linux?[edit]
I wanted to know if anyone else can verify this. I am dual-booting Vista and Ubuntu (9.10) on my laptop. I run the same save file of 40d17 in Vista I get something like 20-30fps, I run 40d17 in Ubuntu I get 70-80fps. Is anyone else having similar results? Kenji 03 02:35, 21 February 2010 (UTC)
Linux being less resource intensive than windows in most regards, even lacking completely capable (Though nouveux is pretty good) drivers for every piece of hardware, should generally give better performance on anything, Dwarf fortress being very resource heavy should see this as well. – unsigned comment by 70.190.4.219