2021-09-14.log

- S0rin (QUIT: Ping timeout: 252 seconds) (~S0rin@user/s0rin)00:20
+ freakazoid333 (~matt@2603-9000-cf0f-80e3-1cde-7af7-8e9a-bd80.inf6.spectrum.com)00:21
+ S0rin (~S0rin@user/s0rin)00:23
+ freakazoid343 (~matt@2603:9000:cf0f:80e3:c410:6378:df6d:cdc7)00:25
- freakazoid333 (QUIT: Ping timeout: 268 seconds) (~matt@2603-9000-cf0f-80e3-1cde-7af7-8e9a-bd80.inf6.spectrum.com)00:29
- XgF (QUIT: Ping timeout: 252 seconds) (~quassel@2001:19f0:5001:1174:f8f1:e895:3ca6:2eb4)01:39
+ XgF (~quassel@2001:19f0:5001:1174:dca9:b399:e5ac:bcad)01:39
chartreuseI don't personally play minetest (just don't really enjoy its feel compared with minecraft, and haven't been playing that much either recently), but I could give it a try if you want02:46
chartreuseAre you still waiting on your MNT Reform?02:47
erlehmannchartreuse, i have not set it up03:00
khmis there a magic spell beyond 'screen /dev/ttyUSB0' to get early-boot console output 03:03
- freakazoid343 (QUIT: Read error: Connection reset by peer) (~matt@2603:9000:cf0f:80e3:c410:6378:df6d:cdc7)03:13
+ freakazoid343 (~matt@2603:9000:cf0f:80e3:c410:6378:df6d:cdc7)03:13
chartreuseerlehmann, just finished compiling minetest from the github source, and the issue I'm seeing right away is I'm seeing my own face infront of me03:26
erlehmannchartreuse yay03:28
chartreuseTransparent leaves look fine. 03:29
chartreuseBesides the seeing the inside of my own face in first person mode is there any other bugs I can check for?03:29
chartreuseGetting 8-10fps full screen with default settings03:29
chartreuseShortening the range quite a bit I can get it up to 15-20 depending on the range. 35 with unreasonably close fog03:31
chartreuseTurning off smooth lighting, 3d clouds, and shader support I get around the same fps03:33
chartreuseSurprisingly dropping the window size down doesn't affect FPS, nor does going fully full screen03:34
chartreuseSwapped over to the one included with Reform, and getting basically the same performance but not seeing the inside of my face03:38
chartreuseMaybe slightly better but that's likely because the biome I spawned this time was different even with the same seed03:38
chartreuseDefinitely playable though03:39
khmah, got it.  now to figure out how the hell u-boot works.03:42
chartreusekhm what did you do to get the early boot messages? Just have a serial terminal on another PC?03:44
chartreuseForgot how crazy caves are in minetest, just followed one down from the surface down to -26003:45
khmchartreuse: yes, at first I had TX and RX swapped, so I got some trash.  'screen /dev/ttyUSB0 115200 8N1' got me a serial console once I had the wires uncrossed03:48
khmthe result looks like this http://okturing.com/src/12062/body03:48
sbatesthere's no way to see u-boot output without connecting a separate serial device, is that right?03:48
chartreuse8N1 in that shouldn't be needed, that's the default03:49
chartreuseThe screen isn't yet initialized so not really03:49
khmchartreuse: I never trust GNU software to do the right thing.03:49
chartreuseThough bluerise is working on that IIRC03:49
khmsbates: so far, no.  a couple people have worked toward getting u-boot output on the inbuilt screen, but haven't got there yet.03:49
chartreuseYou could use minicom instead of screen if you like03:49
chartreuseIt's a more proper serial terminal emulator03:50
sbatesI think my USB to serial adapter might be in my desk in work which I haven't been to in like a year and a half03:50
chartreuseIt needs to be a USB to ttl serial adapter and not a 9 pin one03:50
chartreuseThe reform uses 3.3v TTL serial on those pins rather than RS232's +/- 12v03:50
khmmy ttl-usb converter was too slow so I ordered some  cable off adafruit03:50
sbates I have a little Adafruit FTDI board03:51
chartreuseYeah the adafruit one is very suitable for this as it has individual female dupont connects already03:51
khmI love that u-boot tells you how many bytes it read, but not, say, which file it was reading03:51
chartreuseThe FTDI boards work just fine as well, just need a female to female dupont cable03:51
erlehmann<chartreuse> Besides the seeing the inside of my own face in first person mode is there any other bugs I can check for?03:54
erlehmann<chartreuse> Getting 8-10fps full screen with default settings03:54
erlehmannchartreuse that sounds unreasonably low, i have 27fps on a thinkpad T6003:54
erlehmannchartreuse can you bisect the “seeing the inside of my face” issue maybe? that is unusual, they might have broken the renderer03:55
chartreuse8-10 was at full view distance with default settings like 3d clouds and smooth lighting03:57
erlehmannwhat is full view distance?03:57
khmerlehmann: is that unexpected?  I get something like 5x the flops from a c2d chip than I do from a coretex a5303:57
erlehmanndo you have shaders turned on?03:57
sbatesoh, I think I might have a Bus Pirate somewhere, which I think should also provide UART over USB03:57
chartreuseI had them on at first (but none enabled) but turning them off made no difference03:57
erlehmannkhm having 10fps for minetest is absurdly slow03:57
chartreuseI'm getting 11fps with view range: all in some mountainous terrain03:59
chartreuseDroping to 150 gives 12, and down to 80 gives 1703:59
chartreuseThis is with shaders off, smooth lighting off, particles on, 3d clouds off, opaque water off, connected glass off, "node outlining", fancy leaves, no texture filtering or mipmaps and no AA04:00
chartreuseMaking water opaque and turning off node outlining doesn't really gain any04:01
+ freakazoid12345 (~matt@2603:9000:cf0f:80e3:d8e1:9267:d4e3:a072)04:02
chartreuseCPU usage standing still and spinning around is around 105-110%04:02
erlehmannchartreuse that sounds bad, maybe you can ask in #minetest if you are interested why it is the case04:02
erlehmannlol04:02
chartreuseEh, I don't personally care all too much. I don't play it04:03
erlehmannah ok04:03
chartreuseWhat do people normally get on a raspberry pi 3?04:03
erlehmanncan you post an issue about the inside of your face on github?04:03
erlehmannno idea, but it is playable04:04
chartreuseI suspect these framerates are normal for mobile gpus like these04:04
erlehmanni doubt it04:04
erlehmanni am writing you from a thinkpad t6004:04
chartreusehttps://forum.minetest.net/viewtopic.php?t=1718704:04
erlehmannit has opengl 1.404:04
sbatesfor connecting to the serial output, is it the 3 pins beside the battery connectors?04:05
- freakazoid343 (QUIT: Ping timeout: 252 seconds) (~matt@2603:9000:cf0f:80e3:c410:6378:df6d:cdc7)04:05
sbatessorry, serial console04:06
chartreuseFunny enough turning on all the shaders, smooth lighting and mipmaps/aniso filtering only drops performance by 2 or so fps04:12
chartreuseThough it feels a bit worse, probably higher jitter04:12
chartreuseCould actually be cpu limited in some ways as it is showing over 100% on the main application and if all of that is in one thread it could be the bottleneck04:14
erlehmannchartreuse, hmm are you using sotware rendering??04:14
chartreuseopengl04:14
erlehmannllvmpipe or so04:14
erlehmannhmmm04:14
erlehmannyou can switch to ogles i think04:14
erlehmannthat might work04:14
chartreuseOne sec I'm lagging out trying to close it after going ham and turning 8x AA and full shadows on XD04:15
erlehmannshadows are broken04:17
erlehmannthey make low fps and break rendering04:17
chartreuseogles gives an invalid video driver and just changes to opengl04:18
chartreuseWith software rendering I'm getting 14 fps in a field with view range 80. OpenGL in the same position I'm getting 14, though I think the shader effects (animated grass may have been force disabled in software)04:20
chartreuseTurning shaders ofgf in the same place, opengl gets 19fps04:21
chartreuseIf I look at the ground I get up to 3204:21
chartreuseI think this is just the performance of an A53 with etnaviv graphics04:22
chartreuseA T60 with a T2500 will outperform this CPU and the ATIX1400 with 128MB of ram should also be more powerful04:25
chartreuseThe only direct comparison benchmark I'm finding is a BCM2837 (1.4GHz 4x A53) vs an E8400 Core2Duo, so lets calculate based on that04:27
chartreuseIn multicore the E8400 gets 3124 vs the Pi's 1218, adn 2106 vs 489 single core. In Geekbench 404:27
+ freakazoid343 (~matt@2603:9000:cf0f:80e3:d8e1:9267:d4e3:a072)04:28
chartreuseThe reform is at 1.5GHz, so we can add 8% or so ontop those extra04:28
chartreuseI can find geekbench 4 results for a T2500, it gives 1714 multicore or 1046 single core04:31
- freakazoid12345 (QUIT: Ping timeout: 268 seconds) (~matt@2603:9000:cf0f:80e3:d8e1:9267:d4e3:a072)04:31
khmsbates: yes04:32
khmI'm using J1804:32
chartreuseIf we add 10% to the pi3's scores there to account for the clock speed difference we get 1340 and 5376, still quite dominated by the T520004:33
chartreuseOther A53 quad cores I can find on geekbench's site is an Allwinner one at 1.46GHz, which is 1424/543 so I'd say the numbers are about right still04:34
sbateskhm: thank you, I see J1804:35
chartreuses/5376/537/04:35
- freakazoid343 (QUIT: Read error: Connection reset by peer) (~matt@2603:9000:cf0f:80e3:d8e1:9267:d4e3:a072)04:38
+ Slime19 (~Slime@c-71-206-135-127.hsd1.va.comcast.net)04:39
+ freakazoid343 (~matt@2603-9000-cf0f-80e3-d8e1-9267-d4e3-a072.inf6.spectrum.com)04:39
Slime19Hello all, I just picked up this module to experiment with once the Reform arrives: https://www.iwavesystems.com/product/i-mx-8-qm-qp-smarc-som/04:39
erlehmannchartreuse, how can a thinkpad T60 with intel onboard graphics from 2012 outperform reform04:40
erlehmannchartreuse, ogles2 is the video driver i think04:40
khmbecause it has a much faster processor with out-of-order execution04:40
Slime19It has MIPI DSI output to the SODIMM connector, I am just now wondering if this will be in any way compatible with the board/system controller in the reform, or if the board only supports the 30 pin ribbon which this module does not have 04:41
erlehmannwhat is a T2500?04:41
khman intel core 2 duo processor04:42
chartreuseBecause the reform is an embedded arm processor04:42
erlehmannah04:42
chartreuseThe T2500 is the processor in the T6004:42
chartreuseEven though it's a dual core its singlethreaded and multicore performance still tops the Reform04:43
khmT60 was available with either Yonah or Merom processors04:43
chartreuseI went with a 2GHz Yonah which is the first number I saw for it04:43
khmsorry erlehmann a T2500 is a Core Duo processor04:43
erlehmannok04:44
chartreuseI can go to the top of the line option for it which was a T7600 Merom04:44
chartreuseThat'll just increase the difference even more04:44
erlehmanni have core duo04:44
khmchartreuse: I'm saying not only is the T60 faster even with the Core Duo, erlehmann might have a C2D chip, in which case it's going to be even faster04:44
erlehmannmodel name: Intel(R) Core(TM) Duo CPU      T2400  @ 1.83GHz04:44
khmah ok04:44
chartreuseOkay so slightly slower than the 2GHz T2400 I selected, but still going to be faster than the reform04:45
chartreuseI'm seeing average geekbench 4 scores of 1690 multicore and 989 single core for that04:45
khmeither way, that processor has a lot more horsepower than the ARM chip in the Reform, and keeping VRAM full is easier with out-of-order execution04:45
chartreuseAnd if we go up to a full T7600 that seems to vary a bit, but I'mn seeing ~2300 and ~1360 for that04:46
chartreuseSo your T2400 T60 is around 26% faster in multithreaded workloads, and ~84% faster in single threaded 04:48
chartreuseBut this is the power of a 35W TDP x86 cpu vs a <4W in-order ARM CPU04:49
chartreuseIt's really all due to a lack of power in many ARM designs, since they're designed for phone/router/appliance use. And even more limited as we need one with open hardware documentation04:50
khmAh, the one I pulled testing from was a T8100.  I was using DGEMM benchmarking from the top500 test suite when I came up with the 5x number04:50
chartreuseYeah it's going to depend on the benchmarks, but it's probably going to be in the range of 20-100% difference still04:54
khmDGEMM obviously doesn't touch video output, it's more of an i/o and math test04:55
chartreuseIf you want a real comparison on "modern" processors in Geekbench for, a 4 core i7-7820HQ iin a 2017 macbook pro gets 15331 multicore and 4331 single core04:55
chartreuseI just used geekbench as it was the first number I found for a Pi 3 that I could compare against other processors yet04:56
chartreuseDoesn't say anything about the GPU either04:56
chartreuseBut I suspect even the Intel onboard is going to be faster than the Vivante04:56
khmyep, I agree04:56
khmnot to mention the software stack is going to have years more optimization for x86 and intel video04:57
khmbasically the hardware is faster AND the deck is stacked04:57
chartreuseI'd love for there to be on open ARM processor with the power of Apple's M1 soon04:57
chartreuseOr really any architecture really04:57
chartreuseRISC-V, or even a low power open x8604:58
khmmy desktop is rapidly approaching ten years old, but it's hard to justify upgrading it04:58
chartreuseMy desktop is 2017 (AMD Ryzen 1700), and I only upgraded from a Core2Duo E8500 to that. Mainly because I was gaming a fair bit and it was struggling, and I just got a job04:59
chartreuseMy laptop is a x230 and that still does basically every other task just fine with a 35W 3rd gen i505:00
khmmine's a dual E5-263v2 with 128GB of ram.  it's where I do "work" that doesn't fit on my T450s05:00
chartreuseThe Reform is definitely a downgrade in performance, but not terrible. It can be worked around05:00
khmI intend for the Reform to be my main computer, except I don't understand the boot chain so I can't get 9front running on it yet05:01
chartreuseI tried plan 9 for a little bit, but couldn't really get into it. it's not really for me05:02
khmprobably a sign of being profoundly well adjusted05:02
khmI won't hold it against you05:02
chartreuseShould be possible though, there's plan9 for the raspberry pi 3b05:02
chartreuseThe boot process is going to be very roughly similar05:03
khmwe've got it running on all of em actually05:03
chartreuseRight now it loads a linux kernel and dtb's off the sd card, then boots than 05:03
chartreusehttps://9p.io/magic/man2html/8/booting This mentions a pi uboot image05:04
khmyeah that's richard miller's port05:05
khmI have that code somewhere05:05
chartreuseRight now the uboot configuration on the reform is "compiled in" but it can be changed and flashed05:05
chartreuseYou could enter the uboot console with the serial cable and try and manually boot a plan 9 pi image or such05:06
khmyeah, getting to the u-boot console will be my next trick, which I will attempt tomorrow05:06
chartreusehttps://source.mnt.re/reform/reform-boundary-uboot05:06
chartreuseI imagine you're going to need the serial console until you get a framebuffer driver working as well05:06
khmwell05:07
khmrather get networking up first05:07
khmI can always rcpu into the laptop to get a remote gui once I have networking up05:07
khmthat'll make the rest of the development much simpler05:08
khmboot -> wifi card support -> video -> sound is my plan of attack05:08
sbatesbeing able to read along with conversations like this makes me super excited about the Reform. It's so refreshing to have a computer that is designed to be so open and hackable05:12
sbatesI found my Bus Pirate and I think I'm going to try to connect to the serial console tomorrow05:13
+ freakazoid12345 (~matt@35.136.210.185)05:17
- freakazoid343 (QUIT: Ping timeout: 252 seconds) (~matt@2603-9000-cf0f-80e3-d8e1-9267-d4e3-a072.inf6.spectrum.com)05:19
sbatesI run Linux on my main desktop but I run an Ubuntu LTS and I boot straight into X. I hardly ever see the console and it's been years since I compiled a kernel. I feel very far away from the hardware.05:26
sbatesWith Reform I'm looking forward to getting back in touch with the hardware and lower level OS parts that I feel out of touch with. And to having a computer again that I can understand.05:27
Slime19I hope to get my nixos config working once mine comes in 06:03
- erlehmann (QUIT: Ping timeout: 268 seconds) (~erle@dynamic-046-114-039-120.46.114.pool.telefonica.de)06:16
+ erlehmann (~erle@dynamic-046-114-034-091.46.114.pool.telefonica.de)06:28
ex-parrotI need to try and get something graphical going in LFS next 😂06:58
ex-parrotand for some reason shutting down doesn’t kill power 06:58
+ freakazoid343 (~matt@2603:9000:cf0f:80e3:d8e1:9267:d4e3:a072)07:34
- freakazoid12345 (QUIT: Ping timeout: 265 seconds) (~matt@35.136.210.185)07:36
mntmnSlime19: that's an interesting module that i was also recently looking at, though totally incompatible with the current motherboard. would need a partial redesign.09:29
- freakazoid343 (QUIT: Read error: Connection reset by peer) (~matt@2603:9000:cf0f:80e3:d8e1:9267:d4e3:a072)11:34
+ freakazoid343 (~matt@2603:9000:cf0f:80e3:d8e1:9267:d4e3:a072)11:35
- freakazoid343 (QUIT: Remote host closed the connection) (~matt@2603:9000:cf0f:80e3:d8e1:9267:d4e3:a072)11:50
+ freakazoid343 (~matt@2603:9000:cf0f:80e3:d8e1:9267:d4e3:a072)11:51
+ mjw (~mjw_@2001:1c06:2487:f800:9e5c:8eff:fe8f:a440)12:13
+ freakazoid12345 (~matt@2603:9000:cf0f:80e3:a42c:2d3b:36c0:7490)13:37
- freakazoid343 (QUIT: Ping timeout: 252 seconds) (~matt@2603:9000:cf0f:80e3:d8e1:9267:d4e3:a072)13:40
- erlehmann (QUIT: Quit: Just say no, then the virus can not enter your body without your consent.) (~erle@dynamic-046-114-034-091.46.114.pool.telefonica.de)16:22
- freakazoid12345 (QUIT: Remote host closed the connection) (~matt@2603:9000:cf0f:80e3:a42c:2d3b:36c0:7490)16:34
+ freakazoid12345 (~matt@2603:9000:cf0f:80e3:a42c:2d3b:36c0:7490)16:35
+ freakazoid343 (~matt@2603-9000-cf0f-80e3-1440-c013-1d06-57af.inf6.spectrum.com)16:43
- freakazoid12345 (QUIT: Ping timeout: 252 seconds) (~matt@2603:9000:cf0f:80e3:a42c:2d3b:36c0:7490)16:47
+ freakazoid12345 (~matt@2603-9000-cf0f-80e3-1440-c013-1d06-57af.inf6.spectrum.com)16:54
- freakazoid343 (QUIT: Ping timeout: 268 seconds) (~matt@2603-9000-cf0f-80e3-1440-c013-1d06-57af.inf6.spectrum.com)16:57
- freakazoid12345 (QUIT: Read error: Connection reset by peer) (~matt@2603-9000-cf0f-80e3-1440-c013-1d06-57af.inf6.spectrum.com)17:09
+ freakazoid12345 (~matt@2603:9000:cf0f:80e3:1440:c013:1d06:57af)17:10
Lewis[m]heh, schade, the (size M) MNT hoodies went out of stock before the laptops came into stock. There goes my matching my hoodie to my laptop plan17:15
+ freakazoid343 (~matt@2603-9000-cf0f-80e3-1440-c013-1d06-57af.inf6.spectrum.com)18:03
- freakazoid12345 (QUIT: Ping timeout: 268 seconds) (~matt@2603:9000:cf0f:80e3:1440:c013:1d06:57af)18:06
+ freakazoid12345 (~matt@2603:9000:cf0f:80e3:1440:c013:1d06:57af)18:08
- freakazoid343 (QUIT: Ping timeout: 268 seconds) (~matt@2603-9000-cf0f-80e3-1440-c013-1d06-57af.inf6.spectrum.com)18:11
dopplerLewis[m]: substitute Reform hoodie: https://m.media-amazon.com/images/I/41JiC3tJxWL.jpg18:16
Lewis[m]:P18:18
+ freakazoid343 (~matt@2603:9000:cf0f:80e3:1440:c013:1d06:57af)18:25
- freakazoid343 (QUIT: Read error: Connection reset by peer) (~matt@2603:9000:cf0f:80e3:1440:c013:1d06:57af)18:27
+ freakazoid343 (~matt@2603:9000:cf0f:80e3:1440:c013:1d06:57af)18:27
- freakazoid12345 (QUIT: Ping timeout: 252 seconds) (~matt@2603:9000:cf0f:80e3:1440:c013:1d06:57af)18:27
+ rasmus (~rasmus@c80-217-132-63.bredband.tele2.se)18:59
- mjw (QUIT: Quit: Leaving) (~mjw_@2001:1c06:2487:f800:9e5c:8eff:fe8f:a440)19:10
technomancy"If  the  power  management  section  is  grayed  out  in  the  xscreensaver-demo(1) window,  then that means that your X server does not support the XDPMS extension" <- any idea how to get that working on the reform with Xwayland?19:13
technomancyor like ... brightnessctl works fine; I just need a way to automatically trigger an arbitrary shell command after N minutes of idle I guess19:13
technomancyI guess a crontab which calls xprintidle?19:23
- rasmus (PART: Disconnected: closed) (~rasmus@c80-217-132-63.bredband.tele2.se)20:07
+ erlehmann (~erle@dynamic-046-114-038-041.46.114.pool.telefonica.de)20:20
+ freakazoid12345 (~matt@35.136.210.185)20:45
- freakazoid343 (QUIT: Ping timeout: 268 seconds) (~matt@2603:9000:cf0f:80e3:1440:c013:1d06:57af)20:48
+ mjw (~mark@herd.wildebeest.org)20:53
- Slime19 (QUIT: Ping timeout: 260 seconds) (~Slime@c-71-206-135-127.hsd1.va.comcast.net)21:46
- freakazoid12345 (QUIT: Read error: Connection reset by peer) (~matt@35.136.210.185)22:08
+ freakazoid333 (~matt@2603:9000:cf0f:80e3:5dd3:9b92:6096:9f9e)22:10
khmtechnomancy: swayidle does this for wayland.  you can also set logind's IdleAction to 'lock' -- xscreensaver can't properly detect idleness inside wayland.22:35
khmswayidle is probably the simplest approach here22:35
technomancyswayidle works without sway?22:36
khmyes, just pass it whatever command you want it to run instead of e.g. swaylock22:37
technomancyhm; ok. will try that; thanks22:38
- sbates (QUIT: Quit: Leaving) (~sbates@198.178.118.18)23:50

Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!