Time Nick Message 02:02 kilbith FPS is surprisingly low on Win 11: https://i.imgur.com/z05T5Ix.png 02:02 kilbith this is running at 60 FPS on Linux with the same settings 02:02 kilbith I don't remember it was so low before on Win 11 02:02 kilbith looks like a regression 13:21 proller https://github.com/minetest/minetest/pull/11843 - just types rename... 13:22 proller celeron55, say something plz 13:22 MTDiscord does it still crash when getting close to the world border? 13:58 kilbith man that elhermann is fo fucking energy-draining... 13:58 kilbith what's up with the germans arguing over and over 13:59 nrz win11 has many regression, like NVMe if used as system disk, they are 10 timers slower than expected 13:59 nrz win11 beta released ? 14:00 kilbith what the hell are you talking about 14:00 kilbith win11 is not beta for one 14:00 nrz i answer to your win11 perf ici 14:00 nrz issue 14:01 kilbith and it's actually more optimized than win10; I use it daily 14:01 nrz win11 is just not polished, they are many performance regression after upgrade for many users 14:01 nrz i'm using arch daily no upgrade issue since 12 years : 14:01 kilbith do you use it on daily basis with recent builds? 14:01 nrz https://www.reddit.com/r/Amd/comments/q2hog8/windows_11_users_suffering_from_performance/ 14:02 nrz archlinux ? yeah i use it daily since 12 years 14:02 nrz windows ? no i don't use that shit for noobs 14:02 kilbith 2 months ago it was still beta 14:03 kilbith typical non sequitur from nrz 14:03 sfan5 it might not be in beta but performance issues were in the news for sure ¯\_(ツ)_/¯ 14:04 MTDiscord ever since windows 10 released and start rolling out feature updates for it they've realized they can ship unfinished untested garbage onto their users with little to no repercussions 14:04 MTDiscord because eh, they can always fix it sometime later right? 14:05 kilbith btw, all my Steam library actually run better on Win11 than Linux 14:05 MTDiscord I would expect nerzhul to use FreeBSD XD 14:06 kilbith it's just than MT/irr is not optimized for Windows because devs don't dog-food on it 14:07 kilbith *that 14:07 nrz savilli, you are young minetest contributor ? do you know i ported MT to FreeBSD ports and i have 2 freebsd boxes at home ? ? 14:07 MTDiscord I can guess that 14:08 MTDiscord But you don't use FreeBSD as a main os? 14:08 MTDiscord That's a crime 14:08 MTDiscord kilbith: well apologies that nobody wants to have to deal with windows beyond the bare minimum 14:09 sfan5 windows is pretty okay, try using macOS 14:09 kilbith linux desktop just doesn't cut it on modern hardware 14:09 kilbith bad scaling, kernel crashes, lack of polishes 14:09 kilbith -es 14:10 kilbith and I lose -10 FPS on Factorio, Portal 2, etc. 14:11 nrz strange i run my linux 12h at least per day since 12 years without having any crash 14:11 nrz and i run cyberpunk 2077 in ultra mode at 100 fps 14:11 nrz and i run cyberpunk 2077 in ultra mode at 100 fps on steam + proton + dxvk 14:11 kilbith you'd get 150 FPS on windows 14:11 nrz i don't care about the 50 missing 14:11 nrz it's just dick 14:12 nrz 120 is sufficient, more is just energy consumption and doesn't offer more comfort ? 14:12 nrz anyway let's return back to MT: sfan5 i merged 2 PR 14:13 sfan5 sure thx 14:13 sfan5 there's a few issues that need fixes for 5.5 but none of them are simple 14:15 nrz i miss time to develop in C++ since too long time 14:15 nrz and i don't know how we can fix our code without rewriting major parts. Many parts are just card castles ? 14:16 sfan5 (#11437 and #11775) 14:16 ShadowBot https://github.com/minetest/minetest/issues/11437 -- `3d_mode` is broken 14:16 ShadowBot https://github.com/minetest/minetest/issues/11775 -- New handling of direction key bits is not fully backwards-compatible 14:19 kilbith https://www.phoronix.com/scan.php?page=article&item=alderlake-windows-linux&num=9 14:19 kilbith ouch linux 14:19 kilbith btw hecks uses windows too 14:19 kilbith john carmack too 14:19 nrz arf i don't have 3d rendering thing and i don't have joystick 14:19 kilbith that's for a reason 14:19 nrz kilbith stop that i don't care i'm not a sheep 14:20 nrz i think we have sufficient OS nowadays to make everyone happy, you want to run on NetBSD or on Windws 98, it's your problem not mine 14:20 MTDiscord windows 11 only really works best on latest intel processors because thats all they cared about in dev, everything it ends up sucking on in one shape or another 14:20 nrz windows performs well until you put a fucking antivirus which is mandatory on an os with so many holes 14:20 kilbith wrong 14:21 kilbith there's no need for that on win10/11 14:21 nrz and yeah it's just a benchmark on interl, like anybenchmark the conditions are not the whole cases 14:21 kilbith and you are brainwashed 14:21 nrz you have windows defender 14:22 nrz honestly i don't care about you anymore kilbith, the older you are the stupider you are. 14:22 MTDiscord have to agree with kilbith, win 10, 11, no need for antivirus for most users 14:22 nrz we are talking about how to release MT 5.5 here 14:22 MTDiscord previous versions you did need one 14:22 nrz yeah no A/V it's why many people are cryptolocked by noobs. 14:22 kilbith and your english still didn't improve in 7 years 14:22 kilbith tells a lot about your IQ 14:23 MTDiscord anyways, regarding 5.5, and windows, https://github.com/minetest/minetest/issues/11832#issuecomment-994316196 this issue is very annoying to windows users and would be nice to see it resolved for 5.5 14:23 nrz at least i don't need to rage like a kid on a channel because no-one cares about my PR because i want like a baby to be managed instanly 14:23 sfan5 this conversation doesn't seem very productive 14:24 nrz sfan5 i agree, i just wanted to talk a bout 5.5 before being interrupted for stupid debates 14:24 nrz is anyone we know be able to look for joystick (with some hardware) or 3d mode ? what happen if we revert the commits ? 14:26 sfan5 3d_mode is caused by the switch to irrlicht 1.9, can't be reverted 14:26 sfan5 the joystick change could be reverted but that is definitely not necessary 14:27 sfan5 @Jonathon can you post the entire debug.txt for a session where this happens? 14:28 nrz just to be sure, it's the first or the second release we do with our shiny new irrlicht ? 14:28 nrz maybe more, i can be late ? 14:29 sfan5 would be the first 14:29 sfan5 @Jonathon nvm I might not need that after all but if you have it please keep it ready 14:36 nrz such big step 14:39 MTDiscord Sfan5 ok 14:39 sfan5 pushing http://sprunge.us/YQBYkv?diff in 7m 14:42 nrz wow, this is a bit tricky, it shows a strange design in our API, how a valid object can be empty like this whereas it's a ref for something existing ? 14:56 sfan5 @Jonathon okay yes, debug.txt please 16:05 MTDiscord kilbith: Ever heard of the term "Wintel alliance"? 17:32 erlehmann kilbith is so unfrendly, can we replace with ymmv? https://irc.minetest.net/minetest-dev/2021-12-15#i_5909058 17:35 erlehmann Krock nrz rubenwardy sofar anyone of you interested in talking about minetest.find_nodes_in_area() with me? in particular about https://github.com/minetest/minetest/pull/11858 … maybe? it is a very simple fix. 17:35 erlehmann i promise to find a new crash in response! 17:35 erlehmann (give me a few minutes, i have an idea) 17:55 nrz please be concise and propose a patch ? ? 17:58 erlehmann nrz https://github.com/minetest/minetest/pull/11858 is the patch 17:58 erlehmann nrz oh sorry, wrong page. here: https://github.com/minetest/minetest/pull/11858/files 18:01 erlehmann nrz the summary is: the patch fixes the situation where a node can exist in the map but not be found by minetest.find_nodes_in_area() 18:02 erlehmann (which affects basically every use of minetest.find_nodes_in_area() that can happen near the map border) 18:03 erlehmann nrz, so are you interested? 18:17 MTDiscord sfan5: https://pastebin.com/dT0wFix3 18:18 MTDiscord debug you requested 18:19 sfan5 by "entire debug.txt" I meant the entire log, or did this happen the first time after opening the main menu? 18:19 MTDiscord no, ill delete it and replicate it, have a little bit of time now, else tonight 18:20 MTDiscord should i crank log level to verbose? 18:21 nrz erlehmann: i think it's not the good solution because it's a wide macro, and it will permit OOB in some code parts 18:23 MTDiscord oh nice, just opened and got it to do it with verbose 18:24 sfan5 first try? 18:24 MTDiscord yeah, typically doesnt happen on first try 18:25 MTDiscord https://pastebin.com/5fuSiWDv burns in 10 minutes 18:27 MTDiscord anything else you need for me? else ill be around in 4-5 ish hours 18:27 erlehmann nrz where is the OOB here? have i missed something? 18:27 nrz you extended the CLAMP range if i'm correct, it means you include more nodes by side effect in many clamp calls 18:27 nrz that can have side effects 18:28 erlehmann yes that is intended actually 18:28 erlehmann since nodes can exist outside of MAX_MAP_GENERATION_LIMIT (obviously) 18:30 erlehmann nrz what exactly do you want me to change about the patch so you will support it? 18:30 nrz if you want to avoid a problem in get_node_in_radius, fix the call inside it, can you give me the crash you fixed ? 18:31 MTDiscord anyways, afk for a while 18:32 erlehmann nrz, sfan5 fixed the crash, but used the wrong boundary here: https://github.com/minetest/minetest/pull/11770 the bug was https://github.com/minetest/minetest/issues/11769 18:33 erlehmann nrz this lead to this bug (but really, it does not matter if the nodes are visible) https://github.com/minetest/minetest/issues/11828 18:34 erlehmann basically, minetest got a new buggy area between 31000 and 32767 and i want to eliminate it to avoid more bugs 18:35 erlehmann oh yeah and it also breaks lua mapgen at world borders, which was kinda obvious at the time the change was introduced 18:36 erlehmann (only if that lua mapgen uses minetest.find_nodes_in_area() and something bad happens if it can not find stuff) 18:36 sfan5 so do lua mapgens generate further than ~30920? 18:37 sfan5 I guess that's up to them but not sure since singlenode is also involved 18:38 erlehmann sfan5 yeah, of course? basically everything generates past 30920. 18:41 sfan5 I just started up a world with a Lua mapgen that fully replaces the engine one and it does not generate past 30920, what could this mean? 18:41 sfan5 are you lying or just overconfident? 18:42 sfan5 (30927 to be exact) 18:48 sfan5 just to be sure I checked using minetestmapper and there's really nothing there 18:53 sfan5 tested mapgen with chunksize = 1 too, doesn't go any further 18:55 erlehmann sfan5 the problem is that you can emerge those areas, which is what more complex mapgens tend to do 18:56 erlehmann i mean, the problem is that you can emerge the areas and then not find the stuff you placed in it 18:56 erlehmann breaking emerge would not help here 18:56 erlehmann sfan5 with chunksize=1 i can get to x=30992, you can not? 18:57 erlehmann you may need to create a new world with that chunksize setting 18:57 sfan5 ah yes, I checked the wrong file 18:58 erlehmann regarding emerge, structure placing code does that 18:58 erlehmann so imagine placing a structure near the boundaries 18:59 sfan5 yep, chunksize=1 actually goes up to 30991 18:59 erlehmann now you can use worldedit to put nodes out there i believe 19:00 erlehmann or rip nodes out of there 19:00 erlehmann parts of worldedit probably use minetest.find_nodes_in_area() though, which is … unfortunate 19:01 erlehmann in my experiments at the mapborder i have sometimes created situation where visiting a chunk would crash the server. if a griefer can arrange that, that would be awful. 19:04 sfan5 so if you have a game that places nodes beyond the limit where mapgen is supposed to end, whose fault is that again? 19:04 erlehmann minetests fault 19:05 sfan5 for allowing the game to place nodes there, right? 19:05 erlehmann no, for having allowed it for a long time, then changing behaviour, breaking tons of stuff 19:06 sfan5 I do not consider "must be able to place nodes beyond 31000" part of Minetest's compatibility promise 19:07 MTDiscord There are issues for pushing stuff over mapgen border 19:08 MTDiscord Typically abused with pistons to hide things 19:08 erlehmann that is the visible border 19:11 erlehmann sfan5 you can of course dig in your heels and say your faulty bugfix is totally cool but that will not make the map border less buggy. we both know that a map block is 16 nodes wide. and we both know you can access the invisible outer shell mapblock in vanilla minetest. so if you get up to x=30991 you must know that there is *at least* another area out there. 19:12 erlehmann and 30991 + 16 is definitely over 31000. 19:12 erlehmann people who crash servers do not particularly care about minetest's compatibility promise lol 19:13 erlehmann and people who hide lag machines also do not care. 19:14 sfan5 erlehmann: raising the mapgen limit so it matches a full mapblock (31007) is a very good idea and should be done 19:14 sfan5 but somehow I suspect you won't be happy with that 19:14 erlehmann sfan5 it would definitely help in the short term. 19:16 erlehmann in particular, i would appreciate if mapgen limit was always snapped to full mapblocks 19:18 erlehmann sfan5 i believe making the mapgen limit match a full mapblock will eliminate an entire category of bugs, because suddenly all those functions do not stop working in the middle of easily accessible areas. 19:19 sfan5 I agree 19:20 erlehmann i it might also make it more obvious why it is a bad idea to have mapgen and functions stop working at the same boundary, but i do not want to demotivate you. 19:20 erlehmann it is a step in the right direction (less bugs) after all 19:21 MTDiscord The mood here is very tense. 19:21 nrz is it possible that we stop the mapgen & boudaries check verifying we finish on a mapblock end ? 19:25 erlehmann it is inadvisable to do so 19:26 erlehmann or maybe i misunderstand you 19:33 erlehmann nrz traditionally, the map generation limit works like this: minetest generates mapblocks up until MAX_MAP_GENERATION_LIMIT and the area beyond that contains at least one mapblock that is (normally) invisible to the client, i.e. gets never loaded, but is obviously there on the server. 19:34 erlehmann this has been carefully engineered so that mods that account for unloaded areas get sane behaviour at the map border „for free“ 20:28 kilbith alright so 20:28 kilbith https://i.imgur.com/9XLD1oC.png 20:28 kilbith there's something terribly wrong there 20:29 kilbith drawtime of 2 ms. is not 40 FPS normally, so I assume it's capped 20:29 kilbith but why/where is it capped, I have nfi 20:29 erlehmann kilbith, what's wrong with 40fps? 20:30 erlehmann is it too low? 20:31 kilbith are you seriously thinking that an Intel Iris Xe graphics cannot handle MT at 60 FPS? hell no 20:31 MTDiscord is this on a laptop? 20:31 kilbith I can run GTA 5 on 40 FPS with medium settings with that 20:31 kilbith yes, Dell XPS with Intel i5 11th gen 20:32 kilbith the very latest Dell XPS 20:32 kilbith 60 FPS equates to 16 ms. of drawtime 20:32 kilbith and I have 2 ms. 20:34 kilbith with dynamic shadows I get 4 ms. but the FPS is still capped at 40+ FPS 20:35 erlehmann kilbith i had literally no idea what hardware you have 20:36 Calinou check CPU and GPU usage, make sure either of them is not 100% used on a single core 20:37 Calinou also, remember that laptops are heavily throttled on battery by default 20:37 sfan5 also check if the driver has a settings panel and test if older MT versions are also affected 20:37 Calinou this throttling can _mostly_ be disabled in the OS' power management settings, but it will cost you a lot of battery life (even when not gaming on battery) 20:37 kilbith it's def not a hardware issue, otherwise the drawtime wouldn't be so low 20:37 kilbith it is a software issue 20:38 kilbith no matter if dynamic shadows are enabled or not, or if the draw distance is reduced to 20, it's capped at 40 FPS 20:38 rubenwardy Dell XPS has a thermal throttling problem, you'd be able to tell though 20:38 rubenwardy 2ms drawtime is odd 20:38 rubenwardy (I have a Dell XPS 13 (9370) from 2018) 20:39 kilbith I have the 9305 20:39 MTDiscord I actually have the same issue on my school laptop where when not plugged in, minetest won't go anywhere above 40fps even though it probably should be able to. plugging it into the charger fixes the issue and it goes up to the 60fps limit 20:39 Calinou check the Intel graphics driver control panel, it has power management options available 20:39 Calinou I'm fairly sure there is a default FPS limiter by now 20:40 kilbith Calinou: how do you explain it's capped at 60 FPS on Steam? 20:40 MTDiscord that was with the "full performance" profile when running off of the battery btw, no power saving 20:41 kilbith there is something with Minetest or Irrlicht 20:41 kilbith because the AAA games leave me go to 60 FPS 20:43 kilbith 2 ms. equates to 700+ FPS 20:43 kilbith normally 20:57 erlehmann kilbith which games do you have 700fps 21:05 nrz dd if=/dev/gpu of=/dev/null 21:05 MTDiscord hahaha 21:05 MTDiscord the best game 21:09 kilbith j'en connais un qui va prendre une rouste sur Paris 21:09 kilbith oops 21:10 erlehmann haha you can not even englishing 21:11 erlehmann tells a lot about your IQ 21:13 MTDiscord If we're fair, "can not even englishing" is a weird way to english. 21:14 MTDiscord It is.. irony 21:18 nrz he says a non sense thing, he thinks i'm located in paris, but no ? 21:19 erlehmann josiah_wi i was of jokings because kilbith has been rude earlier https://irc.minetest.net/minetest-dev/2021-12-15#i_5909058 21:20 erlehmann hi, i'm joe king 21:20 erlehmann (pilot for hire) 23:26 MTDiscord What the hell is happening in here today?