IRC logs for #openttd on OFTC at 2021-03-19
⏴ go to previous day
00:31:01 *** gelignite has quit IRC (Quit: Stay safe!)
00:45:18 *** Guest389 has quit IRC (Ping timeout: 480 seconds)
01:34:23 *** Wormnest has quit IRC (Quit: Leaving)
01:54:52 <DorpsGek> [OpenTTD/OpenTTD] James103 opened issue #8874: It is possible to run the `rescannewgrf` command while NewGRFs are being scanned. https://git.io/JmrJo
02:30:35 *** Wuzzy has quit IRC (Quit: Wuzzy)
02:33:11 <DorpsGek> [OpenTTD/OpenTTD] Eddi-z commented on issue #8874: It is possible to run the `rescannewgrf` command while NewGRFs are being scanned. https://git.io/JmrJo
02:44:24 *** dvim has quit IRC (Quit: Connection closed for inactivity)
02:47:42 *** didac has quit IRC (Remote host closed the connection)
03:49:18 *** D-HUND has quit IRC (Ping timeout: 480 seconds)
04:24:06 *** WormnestAndroid has quit IRC (Remote host closed the connection)
04:24:19 *** WormnestAndroid has joined #openttd
04:27:01 *** snail_UES_ is now known as Guest405
04:27:01 *** Guest405 has quit IRC (Read error: Connection reset by peer)
04:27:01 *** snail_UES_ has joined #openttd
04:34:09 *** WormnestAndroid has quit IRC (Read error: Connection reset by peer)
04:34:21 *** WormnestAndroid has joined #openttd
04:52:34 <DorpsGek> [OpenTTD/OpenTTD] perezdidac updated pull request #8794: Feature: Show rail/road/tram NewGRF name in Land Area Information window https://git.io/JthCX
04:52:54 *** Gustavo6046 has quit IRC (Ping timeout: 480 seconds)
04:57:20 <DorpsGek> [OpenTTD/OpenTTD] perezdidac updated pull request #8794: Feature: Show rail/road/tram NewGRF name in Land Area Information window https://git.io/JthCX
05:25:02 *** snail_UES_ has quit IRC (Quit: snail_UES_)
05:31:40 *** didac has quit IRC (Remote host closed the connection)
06:40:13 *** dwfreed_ has joined #openttd
06:40:13 *** dwfreed is now known as Guest411
06:40:14 *** dwfreed_ is now known as dwfreed
06:46:42 *** dwfreed_ has joined #openttd
06:46:42 *** dwfreed is now known as Guest413
06:46:43 *** dwfreed_ is now known as dwfreed
06:47:41 *** Guest413 has quit IRC (Read error: Connection reset by peer)
07:05:15 *** andythenorth has joined #openttd
07:19:09 *** sla_ro|master has joined #openttd
08:26:33 *** HerzogDeXtEr has joined #openttd
08:37:20 *** andythenorth has quit IRC (Quit: andythenorth)
08:39:02 *** bkilm[m] has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** ad5twoknebor[m] has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** Aileen[m] has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** ket has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** Xaroth has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** menelaos[m] has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** glothit7ok[m] has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** philip[m] has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** elliot[m] has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** michi_cc has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** TrueBrain has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** EmeraldSnorlax[m] has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** milek7 has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** Alkel_U3 has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** karl[m]5 has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** patrick[m]2 has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** Hobbyboy has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** yur3shmukcik[m] has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** labs[m] has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** ircer[m] has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** nolep[m] has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** fonsinchen has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** Osai has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** Ammler has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** dihedral has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:02 *** ^Spike^ has quit IRC (reticulum.oftc.net coulomb.oftc.net)
08:39:03 *** milek7_ has joined #openttd
08:39:03 *** TrueBrain has joined #openttd
08:39:03 *** Hobbyboy has joined #openttd
08:39:05 *** michi_cc has joined #openttd
08:39:15 *** Alkel_U3 has joined #openttd
08:39:26 *** ChanServ sets mode: +v michi_cc
08:39:42 *** dihedral has joined #openttd
08:39:51 *** fonsinchen has joined #openttd
08:40:35 *** karl[m]5 has joined #openttd
08:41:21 *** ^Spike^ has joined #openttd
08:44:25 *** glothit7ok[m] has joined #openttd
09:08:43 *** ircer[m] has joined #openttd
09:13:49 *** Aileen[m] has joined #openttd
09:22:12 *** menelaos[m] has joined #openttd
09:26:19 *** elliot[m] has joined #openttd
09:37:52 *** ad5twoknebor[m] has joined #openttd
09:40:27 *** jottyfan has joined #openttd
09:41:26 *** bkilm[m] has joined #openttd
09:48:02 *** andythenorth has joined #openttd
09:51:00 *** EmeraldSnorlax[m] has joined #openttd
09:56:56 *** nolep[m] has joined #openttd
10:09:04 *** patrick[m]2 has joined #openttd
10:13:19 <Eddi|zuHause> Timberwolf: bisecting won't help you if you don't have a prior version that you were sure didn't contain the error
10:26:56 *** labs[m] has joined #openttd
10:27:42 *** yur3shmukcik[m] has joined #openttd
10:30:03 <Timberwolf> It's definitely not in 0.4, does that help? :p
10:31:15 <Timberwolf> Probably find it's down to the GRF anyway.
10:31:34 *** jottyfan has quit IRC (Quit: jottyfan)
10:31:50 <Timberwolf> The ship is 4x but the landscape is 1x, so they don't integrate properly.
10:36:51 *** philip[m] has joined #openttd
10:41:33 *** sla_ro|master has quit IRC (Ping timeout: 480 seconds)
10:41:59 <Eddi|zuHause> that's obviously a user error
10:42:18 <Eddi|zuHause> mark it as "wontfix"
10:54:15 <andythenorth> are we having any lolz today?
10:55:06 <LordAro> i'm watching some people argue about flash's EoL
10:55:40 <Eddi|zuHause> that's how many months old now?
10:56:00 <LordAro> Windows is just now rolling out the update that removes it complete, i think
10:57:40 <LordAro> damn, FreeCiv looks a lot better than it did... 10 years ago
10:57:48 <Eddi|zuHause> ah, so that's like discussing a disaster, after the first movie comes out, instead of when it's in the news
10:58:20 <LordAro> Eddi|zuHause: well mostly they're arguing about "i should be able to run it if i want to"
10:58:55 <Eddi|zuHause> which you probably can, if you really want to
10:58:57 <TrueBrain> LordAro: only their 3D mode; luckily the 2D mode still looks the same :D
10:59:22 <LordAro> the issue is distinguishing between the people who know what they're doing, and people blindly using it because their IT has deployed it or whatever
10:59:32 <LordAro> Eddi|zuHause: apparently it's quite difficult
11:00:18 <LordAro> as in, they've not found a solution
11:00:53 <LordAro> there's a few methods of downloading a standalone version of it, but if the flash app needs to be in a specific place (common with flash games), that's still not going to work
11:01:56 <LordAro> oh, apparently this is a printer with a flash-based UI
12:25:09 <peter1138> Why is rounding in JS so hard? :/
12:37:51 <andythenorth> LordAro I miss Flash :(
12:37:56 <andythenorth> Make Flash Great Again
12:55:24 <Wolf01> I don't, learn javascript :P
12:58:31 <andythenorth> that sounds....worse
12:58:43 <andythenorth> Flash was the solution to Javascript
13:03:02 <andythenorth> more coffee vicar?
13:14:14 *** Gustavo6046 has joined #openttd
14:16:53 *** snail_UES_ has joined #openttd
14:34:42 <supermop_Home> LordAro did you heard about the transit authority that was running train dispatching on flash?
14:35:36 <LordAro> Chinese one, wasn't it?
14:36:45 <supermop_Home> newDisasters in GS
14:37:16 <supermop_Home> "Your trains were all running on Flash, which is no longer supported"
14:56:29 <andythenorth> just use chrome no?
15:18:08 *** Wormnest has joined #openttd
15:53:18 *** Flygon has quit IRC (Quit: A toaster's basically a soldering iron designed to toast bread)
16:26:15 *** Progman has joined #openttd
17:41:52 *** Beerbelott has joined #openttd
17:52:50 *** schmidt_fu[m] has joined #openttd
17:58:36 <LordAro> that's presumably an eints issue?
18:06:10 <DorpsGek> - Update: Translations from eints (by translators)
18:08:45 <glx> hmm I guess it's only eints, openttd has commands to force text direction IIRC
18:09:20 <glx> maybe the issue should move to eints
18:17:58 *** Progman has quit IRC (Remote host closed the connection)
19:02:31 *** frosch123 has joined #openttd
19:04:29 *** FLHerne has joined #openttd
19:07:06 <frosch123> andythenorth: currently i have 6 items on the list you requested. but i have to check them again, whether they can be presented in an entertaining way, that is understood by more than one person
19:09:11 <TrueBrain> why do people, despite being told several times already, keep comparing performance between 1.10.3 and 1.11.0-RC1 .. it still is pointless :P
19:09:40 <andythenorth> it took you all day to notice that one?
19:09:53 <andythenorth> I can't tell if the OP is super funny, and it's just lost in translation
19:09:58 <andythenorth> or if they're just rude and aggressive
19:10:27 <andythenorth> but I can no longer respond because I'm on a forum warning
19:10:36 <TrueBrain> I read your report :P It made me laugh :)
19:11:06 <andythenorth> I have seen a certain type before and this reminds me of that type
19:11:18 <andythenorth> quite smart, lots of answers for everyone else's problems
19:11:22 <andythenorth> no skin in the game
19:11:38 <andythenorth> ridicules anything they don't agree with, as though only an idiot would try it
19:11:51 <andythenorth> and seeks higher authority for retribution when they're disagreed with
19:12:42 <andythenorth> I have employed people like this, they're an absolute plague in a team
19:12:51 <TrueBrain> well, I proza'd a nice reply, to remind yet again of not doing stupid comparisons :D
19:13:22 <frosch123> why did eints run twice today?
19:13:42 <andythenorth> TrueBrain I suspect that the numbers for the RC are so low that there's a genuine problem
19:13:53 <andythenorth> but the tone and presentation make me disinclined to think about it
19:14:03 <andythenorth> the mac build is a *fuckload* faster for all test cases I tried
19:14:27 <LordAro> that definitely does look like the usual "OpenGL is slow when scrolling for some reason"
19:15:05 <LordAro> 100ms graphics rendering is a dead giveaway, and much more than can be explained away by "preleases still have asserts"
19:15:14 <TrueBrain> frosch123: that happens if the schedule of GHA is broken enough that one time it runs 15 minutes later, the other time not :P
19:15:18 <TrueBrain> it is reallllyyyy annoying
19:16:09 *** gelignite has joined #openttd
19:19:02 *** Beerbelott has quit IRC (Quit: Leaving)
19:19:24 <TrueBrain> I clearly haven't read the forums in a while, I just found the best gem of them all
19:19:30 <TrueBrain> me explaining how a functionality works
19:19:34 <TrueBrain> me being told off I should play the game as I am wrong
19:19:56 <TrueBrain> that is just hilarious :D
19:21:30 <andythenorth> pls don't quit now TrueBrain :(
19:21:43 <TrueBrain> no worries, I had a lot of fun reporting that post :D
19:22:01 <TrueBrain> I wanted to take action, but was like: nah, I shouldn't, as it is about me. Better let another moderator deal with it :)
19:24:27 <TrueBrain> LordAro: the last 2 screens are asserts
19:24:34 <TrueBrain> the first three are GPU, I would guess :)
19:25:49 <TrueBrain> hence my "additional to" start of my comment :D
19:25:54 <TrueBrain> it is a two-for-onner
19:26:55 <LordAro> TrueBrain: which is the post where you're being told you're wrong?
19:27:49 <TrueBrain> see PM; nuff said about this, I just really needed this laugh today :)
19:34:58 <milek7_> I doubt that asserts have significant impact on perf
19:40:53 <milek7_> this is scrolling on AMD, on Windows
19:52:41 <LordAro> milek7_: it seems extremely variable which GPUs the issue affects
19:52:50 <LordAro> it's not all AMD, it's not all nvidia
20:01:05 <milek7_> as for assert impact, I checked out of curiosity
20:01:09 <LordAro> you have both AMD & nvidia in the same system?
20:01:18 <andythenorth> I have AMD and Intel in the same system
20:01:23 <andythenorth> the AMD was slower :P
20:01:27 <milek7_> wentbourne 0.6x vs 0.7x
20:01:33 <andythenorth> then I pulled last weekend and the variation had vanished
20:01:46 <milek7_> LordAro: yes, this is laptop with ryzen apu and dedicated nvidia
20:03:28 <frosch123> hmm, technically i also have an on-board amd
20:05:00 <frosch123> hmm, but i guess i cannot just p&p it...
20:05:05 <TrueBrain> milek7_: 10% is on average what I notice on games, asserts vs non-asserts :) So 0.6 vs 0.7 sounds about right
20:05:09 <TrueBrain> it is mostly in train ticks
20:05:30 <TrueBrain> one could wonder if we cannot remove a few of those asserts to reduce the impact of them
20:06:57 <milek7_> funnily enough, I'd expect integrated amd to be faster, as it shares system memory (and nvidia needs to push through PCIe)
20:12:15 <milek7_> I guess we need d3d11 driver ;p
20:12:53 <TrueBrain> someone should really debug these GPU issues (as in, with OpenGL tooling) .. possibly it is just something really silly in our OpenGL code :D
20:14:20 <milek7_> I don't think there's proper opengl profiler available for amd
20:14:47 <milek7_> amd have but only for d3d12/vulkan
20:14:54 <TrueBrain> just we have nobody that knows how to profile and has this issue on NVidia, I believe :P
20:16:55 <TrueBrain> what are the odds I can find the assert that has the most impact on performance by enabling/disabling some?
20:17:48 <glx> check the most called functions in profiler, and disable assert in them ?
20:18:21 <glx> random guess, asserts in game loop
20:18:29 <TrueBrain> I think it is the PF
20:19:58 <TrueBrain> but yeah, flamegraph is best way to go about this
20:21:02 <TrueBrain> 1800+ asserts .. lol
20:25:56 <DorpsGek> TrueBrain: 0.9800000000000001
20:27:13 <TrueBrain> less than I expected :D
20:28:00 <glx> oh yeah should be complex enough
20:28:27 *** Progman has joined #openttd
20:28:59 <TrueBrain> 5s for 100 ticks in a normal game .. 4.9 with PF asserts off .. 4.1 with all off
20:31:21 <TrueBrain> we define assert_msg, but seem to never use it
20:32:09 <TrueBrain> owh, the assert handler for RC1 and beta1 is different from debug builds
20:33:46 <glx> because "normal" assert are disabled in non debug
20:41:10 <Kitrana1> is there a grf that makes it so helicopters don't circle airports like a weird wingless plane
20:41:27 *** Kitrana1 is now known as Kitrana
20:44:35 *** jottyfan has joined #openttd
20:53:26 <LordAro> airport patterns are very fixed
20:54:09 <milek7_> maybe it's autogyro :D
21:13:47 <milek7_> but still, 25ms for what should be memcpy..
21:16:15 <milek7_> (removed pixel unpack buffer altogether and just passing host memory pointers to glTexSubImage2D)
21:40:38 <milek7_> 'I deleted blitting code and it still works' type of confused :P
21:44:08 *** ericnoan has joined #openttd
21:55:39 *** Samu has quit IRC (Quit: Leaving)
22:14:11 <milek7_> michi_cc: I might just need sleep, but..
22:14:34 <milek7_> why with this, it still renders anything?
22:15:12 <michi_cc> 8bpp sprites? Look at ReleaseAnimBuffer instead.
22:26:17 <michi_cc> Some speculation: Scrolling is *the* major operation that reads back from the video buffer. The video buffer is created as GL_DYNAMIC_DRAW and mapped GL_READ_WRITE exactly because of that. You would normally expect a buffer of that kind to reside on DMA-accessible CPU memory.
22:27:51 <michi_cc> As scrolling is observed to be slow, the first theory would be that for some unknown reason the AMD drivers places the buffer in the wrong kind of memory region which has very slow readback. Why that is and how the driver can be influenced to change it is an open question.
22:28:26 <LordAro> but not all AMD drivers?
22:29:41 <michi_cc> Interestingly enough, it was reported that re-enabling persistent mapping for AMD is not faster, even though the persistent mapping API allows to better specify buffer usage.
22:30:10 <michi_cc> LordAro: All drivers that are slow, how ever few or many that might be.
22:31:16 <michi_cc> I think that except for Vulkan (and maybe DirectX 12), you do not get any APIs to tell the driver exactly where to place your buffer, just suggestions.
22:32:23 <LordAro> i can't imagine this isn't a solved problem
22:32:43 <andythenorth> does the mac build use OpenGL at all (in any circumstance)?
22:32:58 <michi_cc> Hit the hardware acceleration switch.
22:34:06 <andythenorth> which causes the AMD 5500M to be selected :P
22:34:22 <milek7_> huh, it switches gpu globally?
22:34:36 <andythenorth> unclear on the mac
22:34:45 <andythenorth> the apps report which GPU they're using
22:34:57 <andythenorth> but I think it's one GPU per screen display
22:35:25 <andythenorth> in theory on the mac, if you delegate some stuff to GPU, that can do tasks, then pipe the output back to the screen on another GPU
22:36:22 <milek7_> as before, "(removed pixel unpack buffer altogether and just passing host memory pointers to glTexSubImage2D)"
22:36:32 <milek7_> I just forgot about anim buffer previously
22:37:45 <michi_cc> I implemented that first in the OpenGL commits, and for nVidia and Intel where I can test on, using a buffer object was always better than not using one. It still feels like the driver does something stupid.
22:38:48 <andythenorth> I don't currently see much difference on mac between the Intel UHD Graphics 630 and the AMD 5500M
22:39:15 <andythenorth> when the hardware switch first dropped in master there was more variation, but something else changed since
22:39:51 <milek7_> michi_cc: it's not that windows amd driver is known for performance :P
22:41:45 <andythenorth> ffwd fps is maybe 80% faster on mac hardware accelerated
22:41:59 <andythenorth> depending on viewport contents
22:45:54 <milek7_> >Interestingly enough, it was reported that re-enabling persistent mapping for AMD is not faster
22:46:21 <milek7_> wait, by whom? I don't remember testing that on windows
22:48:54 <DorpsGek> [OpenTTD/OpenTTD] Kuhnovic commented on pull request #8706: Feature: rail station class name filtering https://git.io/Jm1QT
22:51:22 <DorpsGek> [OpenTTD/OpenTTD] LordAro commented on pull request #8706: Feature: rail station class name filtering https://git.io/Jm1Qd
22:53:21 <DorpsGek> [OpenTTD/OpenTTD] Kuhnovic commented on pull request #8706: Feature: rail station class name filtering https://git.io/Jm17R
22:53:33 <milek7_> I'm sorry if there was confusion about that, but the issue with persistent being slow was on Linux mesa drivers
22:53:47 <milek7_> which was since fixed by CLIENT_STORAGE bit
22:54:12 <michi_cc> I think it was peter who reported persistent mapping to not improve anything.
23:01:49 <milek7_> this block should be removed, as it was fixing wrong problem: it's not that persistent mapping itself was slow, but it was missing CLIENT_STORAGE bit
23:02:45 *** andythenorth has quit IRC (Quit: andythenorth)
23:03:21 <michi_cc> I had people tell me that #define'ing GL_MAP_PERSISTENT_AMD did not make anything better. So removing it might simply mean that we trade one set of slow GPUs against another set of slow GPUs.
23:04:26 *** Wolf01 has quit IRC (Quit: Once again the world is quick to bury me.)
23:04:36 <michi_cc> It is entirely possible that their testing was invalid, but it's all anecdotes. Feel free to make a PR though, so more people can chime in.
23:05:04 <DorpsGek> [OpenTTD/OpenTTD] Kuhnovic opened issue #8875: Filter string in station window breaks flow in user interface https://git.io/Jm1FY
23:07:46 <DorpsGek> [OpenTTD/OpenTTD] stormcone commented on pull request #8706: Feature: rail station class name filtering https://git.io/Jm1FS
23:08:26 <DorpsGek> [OpenTTD/OpenTTD] stormcone commented on issue #8875: Filter string in station window breaks flow in user interface https://git.io/Jm1FY
23:17:20 <DorpsGek> [OpenTTD/OpenTTD] Kuhnovic commented on issue #8875: Filter string in station window breaks flow in user interface https://git.io/Jm1FY
23:17:55 <DorpsGek> [OpenTTD/OpenTTD] Milek7 opened pull request #8876: Change: Do not disallow persistent buffer mapping on AMD GPUs, as it is actually faster. https://git.io/Jm1Al
23:25:32 <DorpsGek> [OpenTTD/OpenTTD] michicc opened pull request #8877: Fix #8871: [OpenGL] Initialize all buffers after resize and clear back buffer. https://git.io/Jm1pf
23:28:29 <DorpsGek> [OpenTTD/OpenTTD] TrueBrain commented on pull request #8877: Fix #8871: [OpenGL] Initialize all buffers after resize and clear back buffer. https://git.io/Jm1pr
23:29:03 <TrueBrain> michi_cc: OSX has no swap buffer variant?
23:30:06 <michi_cc> OSX is "different" due to layer-backed views. The context that is active in this moment doesn't even have an associated window.
23:30:59 <DorpsGek> [OpenTTD/OpenTTD] michicc commented on pull request #8877: Fix #8871: [OpenGL] Initialize all buffers after resize and clear back buffer. https://git.io/Jm1hm
23:31:41 <michi_cc> Or put differently: Unless there's a demonstrated need for it, I'd rather not try to figure out OSX in this regard :D
23:37:35 *** gelignite has quit IRC (Quit: Stay safe!)
23:44:15 *** jottyfan has quit IRC (Quit: jottyfan)
continue to next day ⏵
