Time Nick Message 03:11 repetitivestrain sfan5: regarding https://github.com/luanti-org/luanti/pull/16265: the real performance advantage arises from avoiding garbage collection in instances where luajit cannot sink the created table 03:15 repetitivestrain i cannot offer you any benchmarks, but the jit compiler cannot compile branches that begin in blocks with large numbers of live stack slots, which applies to some of the mcl_mobs pathfinder, nor can it sink values across traces 03:15 MTDiscord Reduce, reuse, recycle 03:16 repetitivestrain Mineclonia has been extracting get_node_raw from core.get_node with debug.getupvalue for some months now and in my experimentation it consistently reduces the frequency at which the garbage collector is invoked and the duration of globalsteps that coincide with garbage collection 03:18 repetitivestrain i very much suspect kno10's benchmark is one where LuaJIT throws in the towel 03:19 MTDiscord Were these tested in singlenode? 03:22 repetitivestrain get_node_raw doesn't concern map generation but rather pathfinding 03:23 MTDiscord Wat. I'm saying for the benchmarks. Best to not have any variables when bench marking 03:24 MTDiscord Generate that one node and hammer that same memory address with no interference to see which one wins, etc 03:24 repetitivestrain those _macro_ benchmarks were conducted in the same map and at roughly the same coordinates on a real server with an essentially constant number of mobs and 2-3 players 03:26 MTDiscord I'm not talking about your benchmark, I'm talking about the original 15-20% ones 03:26 repetitivestrain You'll need to ask kno10 03:33 MTDiscord There is also another performance implication. You can reverse a lookup with a table of predefined IDs as a set and if it contains, you have hit something you're looking for. But also, you can precache a single so it's just a number to number compare. If it's okay, etc 03:34 repetitivestrain table indexing can easily account for the 6% difference observed by sfan5/SmallJoker 03:35 MTDiscord I wonder if this is actually faster to use, via precache IDs in a group via a numeric table key to numeric value to out pace checking if a node is in a group from the standard api elements 03:37 MTDiscord local gr = idt[core.get_node_raw(xyz)]; print(gr and gr > 0) 03:38 MTDiscord But that's not correct, it's a multi return, but I'm too lazy to fix it 03:40 MTDiscord So basically you'd scan the registered nodes on mods loaded and dump the group you want in via the ID so the heavier work happens at load up 03:50 repetitivestrain sfan5: https://paste.debian.net/1380007/ 04:02 repetitivestrain try this benchmark: both generate traces of the same character, but the jit compiler cannot sink the table created in core.get_node across a stich (pr:next), the trace which is stitched and its side trace, with the result that a real table is allocated and dereferenced in all three cases 07:52 sfan5 repetitivestrain: 404 07:57 repetitivestrain sfan5: it existed when i posted it, one second 07:58 repetitivestrain https://paste.debian.net/1380026/ 08:21 sfan5 get_node 715.03 ms, 0/0... 08:21 sfan5 get_node_raw 313.74 ms, 0/0... 08:35 repetitivestrain Yes 08:35 repetitivestrain that's pretty consistent with my experiences with a real lua pathfinder 10:24 MinetestBot 02[git] 04SmallJoker -> 03luanti-org/minetest_game: Beds: Add support for pova and player_monoids (#3200) 134070879 https://github.com/luanti-org/minetest_game/commit/407087977ca451863e57ca5cf591df23297a364b (152025-06-17T10:24:17Z) 14:07 [MatrxMT] hi 14:11 [MatrxMT] how do i see if someone in irc is there 14:13 sfan5 134 users here, not necessarily online 14:14 [MatrxMT] connect with an IRC client.. or !tell them something I suppose 14:42 [MatrxMT] !tell 14:43 [MatrxMT] it's !tell username message 15:05 MTDiscord Command sent from Discord by jordan4ibanez: 15:05 MTDiscord !tell jordan4ibanez hi 15:05 MinetestBot MTDiscord: yeah, yeah 15:05 MTDiscord Wow that's mean 15:05 MTDiscord Command sent from Discord by jordan4ibanez: 15:05 MTDiscord !tell jordan4ibanez jordan4ibanez 15:05 MinetestBot MTDiscord: I'll pass that on when jordan4ibanez is around 15:06 MinetestBot jordan4ibanez: Jun-17 15:05 UTC hi 15:06 MinetestBot jordan4ibanez: Jun-17 15:05 UTC jordan4ibanez 15:06 jordan4ibanez Technology 15:08 [MatrxMT] !tell bracket test 15:08 [MatrxMT] aw 15:15 MTDiscord Heh heh heh heh heh 15:15 MTDiscord Command sent from Discord by jordan4ibanez: 15:15 MTDiscord !tell Bracket hi 15:15 MinetestBot MTDiscord: yeah, yeah 15:15 MTDiscord Hm 15:16 [MatrxMT] Uhm 15:16 [MatrxMT] it did notification 15:16 MTDiscord Wow I can't see it lol 15:26 [MatrxMT] btw do you know redundantcc is back or how to remove password of uefi/bios of a pc which stores it somewhere and won't reset from removing battery? 15:27 [MatrxMT] do you know when * 18:22 [MatrxMT] looks like we have another LLM poster 18:22 [MatrxMT] https://forum.luanti.org/viewtopic.php?p=445417#p445417 18:22 [MatrxMT] https://forum.luanti.org/search.php?author_id=53258&sr=posts 18:24 [MatrxMT] we're all behind the idea of just banning posters who break rule 8 I hope https://forum.luanti.org/viewtopic.php?t=17151 18:46 MTDiscord Command sent from Discord by C: 18:46 MTDiscord !tell Bracket try using the "restore default settings" option in the bios boot screen. It should be f9 according to this old forum post. If not leave the both the cmos and laptop battery's out for a day and after hold the power button for 30s to drain every drop of juice and check to see if that help remove the power from the memory chip. 18:46 MinetestBot MTDiscord: I'll pass that on when Bracket is around 18:47 [MatrxMT] Nope that won't work 18:48 [MatrxMT] F9 is locked behind password 18:48 [MatrxMT] Instead of CMOS battery tge main battery is used, and when I do that onlyvsettings notvpassword is reset 18:48 [MatrxMT] I remember in high school I shorted two pins on one of the ICs on my school laptop motherboard. That bypassed the password requirement on the BIOS... 18:49 MTDiscord You are doomed, DOOMED 18:49 MTDiscord Idk then... your laptop might be cooked. 18:49 MTDiscord But not doom 2 because doom 1 was better 18:49 [MatrxMT] that's all board-specific though 18:49 [MatrxMT] Could one find out if you can short circuit? 18:49 [MatrxMT] idk, could you? 18:50 [MatrxMT] probaly not 18:50 MTDiscord He beat me to the joke 18:50 [MatrxMT] I wouldn't know where to search 18:50 MTDiscord HP support forums 18:52 [MatrxMT] https://h30434.www3.hp.com/ is fake right 18:53 MTDiscord Hp is a company that you're going to find a company inside of it if you search hard enough. Be careful, you might accidentally get hired at hp during this search 18:58 MTDiscord Try maybe this? https://www.youtube.com/watch?v=Mezf2fYrmcQ 19:02 [MatrxMT] https://matrix.org/_matrix/media/v1/download/toppoint.de/axyNlKBiDfTymKjvgOcAVtrQ 19:02 [MatrxMT] website not availlbe (404) 19:03 MTDiscord matrix.org is down? 19:03 MTDiscord You are doomed 19:03 MTDiscord why? 19:04 [MatrxMT] where do i get bios image 19:04 [MatrxMT] all i can find are so exes 19:04 [MatrxMT] some* 19:04 MTDiscord You're gonna have to install windows lol 19:04 MTDiscord Well, as a last resort you can try the lookout method? 19:04 MTDiscord BREAK OUT THAT SCREW DRIVER 19:04 MTDiscord https://www.youtube.com/watch?v=rvLzWW9Ot6I 19:06 [MatrxMT] what lookout method? 19:08 MTDiscord you enter the wrong password three time and get get a lockout code, then go to hp suport like in the video and they give you a password based on that code. 19:08 MTDiscord If you're going to try flassing anyway mayas well try this first. 19:09 MTDiscord *flashing 19:09 MTDiscord DOOMED 19:09 MTDiscord I'm just kidding, you should definitely try that 19:09 [MatrxMT] funny thing, if i do that it doesn't give me a code but instantly boots 19:10 MTDiscord Well the next option is to unplug that hard drive 19:10 [MatrxMT] why 19:10 MTDiscord So it doesn't boot 19:11 [MatrxMT] uhm 19:11 MTDiscord unplug both batterys and hold the power button, maybe the laptop battery was providing power for the bios? 19:11 [MatrxMT] well i kinda had that and it said "no os installed" 19:11 [MatrxMT] there was no cmos battery 19:12 [MatrxMT] the laptop battery was the subsitute 19:12 [MatrxMT] + it wont boot once the battery level goes too low 19:12 MTDiscord Why are you trying to preserve the bios? 19:12 [MatrxMT] I'm not 19:12 MTDiscord Pull that battery and the other one 19:13 [MatrxMT] I did 19:13 [MatrxMT] Bios *settings* reseted but not the password 19:13 MTDiscord Oh well then time to install windows and wipe it out with the hp bios flasher 19:14 [MatrxMT] oh no 19:14 [MatrxMT] din't i need a licence? 19:14 MTDiscord At least you get a bios upgrade in the process 19:15 MTDiscord Wat, no, you can install Windows 10®™© for free 19:15 [MatrxMT] Oh 19:15 MTDiscord Then you can wipe that drive and put a good operating system on it 19:15 [MatrxMT] can i cramp it on a 3.7GiB stick 19:16 MTDiscord I do not know, you have to read the tutorial because I am not going to walk you through that in an irc lol 19:17 [MatrxMT] Hm yea windows I can do alone 20:28 ireallyhateirc @luatic, take a look at the gltf morph target PR when you get time. I believe that it won't work with the usual Blender workflow without the "KHR_texture_transform" extension 21:49 MTDiscord yeah i saw your comments. frankly i'm a bit surprised that blender expects this extension. i'll see. 22:18 ireallyhateirc @luatic, afaik blender shape keys are only meant for shape changes and it appears that texture/UV animation is done mostly/only by redoing the logic on the game engine side 22:18 ireallyhateirc so my request here is non-standard and would make Luanti special if implemented 22:22 MTDiscord i know :> 22:23 MTDiscord i also have an ulterior motive for getting this into the engine: i need a sequel to "bad apple but it's a b3d", and what would be more suitable than "bad apple but it's a gltf" for that? 22:24 ireallyhateirc I don't get it but I approve 22:24 ireallyhateirc worst case scenario I will hide 10 face variants into the head 22:25 ireallyhateirc and control them with shape keys or bones