**** BEGIN LOGGING AT Mon Nov 28 02:59:56 2022 Nov 28 07:56:41 Morning! Nov 28 09:54:51 JaMa: Seems OSE release is hitting after all? Nov 28 09:54:58 I see updates on GG Nov 28 09:54:59 GH Nov 28 12:27:06 Morning! Nov 28 12:27:59 Herrie: my motherboard should be delivered within the hour, I'm keeping an eye on my phone :) Nov 28 12:28:24 Tofe: ah nice! Seems new notebook tomorrow here Nov 28 12:49:14 Herrie: delivered! Nov 28 12:57:49 Herrie: it was just postponed a bit Nov 28 12:59:48 JaMa: Ah OK Nov 28 13:00:00 Seems I'm having some issues with the newer enact apps Nov 28 13:00:12 But meta-webosose isn't there yet Nov 28 13:01:25 I was cleaning up my enhanced ACG so bumped those apps right away Nov 28 13:01:40 last target I've seen was first week of Dec so the meta-webosose and build-webos might be just later this week Nov 28 13:03:21 Seems it somehow doesn't pull in the required packages for enact Nov 28 13:03:35 Not at home now, can post the error a bit later Nov 28 13:04:22 Tofe: Nice! Nov 28 13:53:36 JaMa: Seems I'm missing "You can install this package by running: npm install fbjs/lib/ExecutionEnvironment", maybe needs adding here? https://github.com/webosose/meta-webosose/blob/master/meta-webos/recipes-webos/enactjs-apps/enact-framework.bb#L34-L56 Nov 28 13:57:37 JaMa: OK might have found it, testing Nov 28 14:05:28 Herrie: all enact/npm recipes in OSE need to be reworked to use npmsw:// instead of npm fetching deps in do_compile/do_install :/ Nov 28 14:06:03 JaMa: Ah OK, I see... You don't have to happen to have this pushed somewhere already? Nov 28 14:06:16 no Nov 28 14:07:09 it's not that easy, I've reworked usocket recently and later it got reverted, because it broke something else in internal build (which didn't even declare dependency on usocket) Nov 28 14:07:36 JaMa: OK, I guess I'll revert to older SRCREV for those apps then simply for now Nov 28 14:07:44 And pick it when OSE pushes meta-webosose Nov 28 14:08:28 now I was working on migrating localization-tool-native and it has its own issues as well and newer nodejs also causes all enact apps to fail due to dependency issues in them Nov 28 14:11:26 JaMa: OK I won't touch those then for now Nov 28 14:22:36 Reverted SRCREV's and build OK now Nov 28 16:25:50 Tofe: Well here's the initial stab :) https://github.com/webOS-ports/meta-webos-ports/pull/544 Nov 28 16:46:44 Tofe: Some parts from new OSE release that might be of use: https://github.com/webosose/ime-manager/commits/master Nov 28 16:51:42 https://github.com/webosose/luna-surfacemanager/commits/master Nov 28 20:38:03 Hi from my new rig ! Nov 28 20:40:22 I still have plenty to tweak, but my system boots, at least Nov 28 20:41:02 The main remaining thing being to manage to have the DDR at the right speed, currently it's only 2166Mhz Nov 28 20:42:47 cool, did you enable SPD profile in bios? it's sometimes a bit hidden even under overclocking menu Nov 28 20:43:58 and don't trust the speed shown in memtest86+ if that's where you've seen 2166 as it isn't correct in many cases (including mine showing 2666) Nov 28 20:44:47 and 2166 is a typo from 2666, right? Nov 28 20:45:16 as 2666 is the default and for 21xx it would be 2133 Nov 28 20:48:04 JaMa: well, i'll have to look further, I never tweaked the RAM before Nov 28 20:50:09 did you buy that patriot ram? PVE2464G360C0K ? Nov 28 20:50:21 yes; let me find the ref Nov 28 20:50:45 and the https://www.gigabyte.com/fr/Motherboard/B550M-DS3H-rev-1x/sp#sp board? Nov 28 20:51:20 "Patriot Memory Viper Elite II DDR4 3600 64Go" can't find the ref right now Nov 28 20:51:37 No, I took Asus TUF Gaming B550M Wifi II Nov 28 20:52:00 ok, yes, it's PVE2464G360C0K Nov 28 20:52:39 There is a D.C.O.P. profile to activate in the BIOS, but that resulted in a hardware bootloop Nov 28 20:53:30 https://www.vortez.net/articles_thumb/44911_asus_b550m-plus_wifi_bios_ai_tweaker.jpg Nov 28 20:53:46 yes, that's SPD/XMP profile Nov 28 20:54:27 did you go directly to 3600? or e.g. 3200? Nov 28 20:54:38 There only one 3600 profile Nov 28 20:54:43 there's Nov 28 20:55:34 as a comment on amazon I see "Direkt wieder in's Bios, XMP aktivieren und die Einstellungen für die neuen Riegel eingeben, bei mir 3600 Mhz 20-25-25-46 und speichern! Neustarten und fertig" Nov 28 20:55:58 ah, then I would try to change just freq to 3200 while increasing ram voltage a bit (or to 1.35 like your DCOP probably does) Nov 28 20:57:24 and again the boot might take long time due to memory training Nov 28 20:58:11 well, currently it just reboots 3 times and then go into safe mode in bios Nov 28 20:58:35 ok, that might indicate failed training Nov 28 20:59:41 if I read a bit the forums it seems easier to have 3200Mhz running Nov 28 21:01:33 ok I'll try out the 2 steps: 1. select DOCP #1; 2. lower frequency to 3200 Nov 28 21:03:39 Tofe: Good it boots, 3200 should be fine in general Nov 28 21:04:40 well, that worked, I think Nov 28 21:04:47 looks like terrible training times I remember were with 12th gen Intel with DDR5 https://www.youtube.com/watch?v=uu9U7TVNImI Nov 28 21:08:21 at least lshw-gtk is showing 3200Mhz Nov 28 21:08:27 you can also test DOCP for 3600 and check if you can lower FCLK, ideally it should be 1:1 with MEMCLK (1800), but it's possible that your CPU doesn't manage it (and DOCP 3600 would work for mem) Nov 28 21:09:24 I'll test that a bit later; right now I'd like to see if my PSU is holding under high usage Nov 28 21:09:35 because, well, 300W... Nov 28 21:09:45 if need be I'll upgrade it too Nov 28 21:09:57 power total: 200.00 W Nov 28 21:09:58 power +12v: 166.00 W Nov 28 21:10:01 is my idle :) Nov 28 21:10:16 where do you see that? Nov 28 21:10:42 in sensors (from lm-sensors), but your older PSU is not likely supported Nov 28 21:10:54 ah, exact Nov 28 21:11:07 ok, let's rebuild nodejs Nov 28 21:11:29 don't forget to increase -j Nov 28 21:11:40 ah, right ! Nov 28 21:11:52 eh, I didn't upgrade for nothing :) Nov 28 21:12:16 are you using stock cooler? Nov 28 21:12:37 no, a Noctua cooler (NH-L12S) Nov 28 21:12:47 there's no stock cooler for most of AMD cpus Nov 28 21:13:14 ah, small case Nov 28 21:14:05 yes, but I could still put it in high profile Nov 28 21:14:40 ok, let's go Nov 28 21:14:57 CPU is now all-in Nov 28 21:15:17 I'm keeping an eye on the cpu temp too Nov 28 21:16:21 mmh glances tells me I'm only at 2.2Ghz Nov 28 21:16:31 governor issue? Nov 28 21:17:51 "glances"? don't know this app, what does cat /proc/cpuinfo | grep ^cpu.MHz show? Nov 28 21:17:59 it says the same Nov 28 21:18:13 I'm now trying cpupower-gui to tweak the governor Nov 28 21:18:49 ah, putting it to "performance" did the trick Nov 28 21:19:02 was it on powersave before? Nov 28 21:19:06 yes Nov 28 21:19:09 as ondemand should work relatively OK Nov 28 21:19:47 well, nodejs is a bit deceiving anyway, right now it's only using 2 cores Nov 28 21:20:06 true and takes long time even on my builder Nov 28 21:20:40 but it was using all cores for at least some parts of the build, right? Nov 28 21:20:54 at the beginning yes, for a minute or so Nov 28 21:21:18 ah, now it's back at full pace Nov 28 21:21:44 it should also eat all our memory (almost) Nov 28 21:22:51 67°C, 3.6 to 4.4Ghz, all nice Nov 28 21:25:04 67 is a bit high for 65W cpu, but maybe it's just the smaller cooler (or rather the profile on it) as I have 67 while eating over 280W Nov 28 21:25:26 it's 75W under high load Nov 28 21:25:44 and my ventilation is far from perfect, too Nov 28 21:26:19 now it's touching 70° here and there Nov 28 21:26:55 similar to what I had with the i5-6600 anyway Nov 28 21:27:28 yeah, I'm a bit spoiled with my 5x12cm, 5x10cm, 1x8cm noctua fans in case :) Nov 28 21:29:15 it's only eating 14Gb Nov 28 21:29:19 of ram Nov 28 21:29:48 that's weird Nov 28 21:30:05 with -j 16? Nov 28 21:30:14 yes yes, all cores are at max Nov 28 21:30:39 well check it near the end of do_compile when it's linking Nov 28 21:30:56 for me it's over 2GB per thread Nov 28 21:30:57 "make -j 16 CC.host=aarch64-webos-linux-gcc [etc...]" Nov 28 21:31:35 for me they're all under 500Mb Nov 28 21:31:47 and not for long, might not catch it in htop if you're not watching graphs somewhere Nov 28 21:31:58 I usually notice it only when it causes OOMK :) Nov 28 21:32:03 hehe Nov 28 21:32:14 I've started a graph now Nov 28 21:33:01 cpu t° seems stabilized at 68-70°, it was one of my main concerns Nov 28 21:33:18 No PSU suicide either Nov 28 21:34:30 and, for all I can see, no RAM failure (but maybe the binaries are all corrupt :) ) Nov 28 21:34:54 better to check with memtest86+ Nov 28 21:35:31 sure Nov 28 21:36:03 damn, it's been 20min ! should I upgrade my CPU ? :D Nov 28 21:36:21 (still building) Nov 28 21:37:52 366.06 seconds nodejs-native-16.14.2-r0/do_compile Nov 28 21:39:46 anyway I didn't buy a high-cpu either Nov 28 21:40:20 s/-cpu/-end cpu/ Nov 28 21:41:14 and I'm not yet using Herrie's NVMe SSD Nov 28 21:50:44 ok, do_compile, ended Nov 28 21:51:21 I didn't look closely, but around 30min I'd say. And nothing over 17Gb of ram Nov 28 21:52:38 do you have buildstats enabled? you can see exact do_compile time there Nov 28 21:59:38 not sure Nov 28 22:00:26 but I see: Nov 28 22:00:28 Nov 28 21:14 log.do_configure.4996 Nov 28 22:00:32 Nov 28 21:49 log.do_compile.5033 Nov 28 22:00:44 which gives are pretty good idea Nov 28 22:01:23 I've started the target nodejs build as well to flex a bit :0 Nov 28 22:04:54 I think that nodejs took hours, before my upgrade Nov 28 22:06:51 3h30 yes, looking at some logs Nov 28 22:07:45 7 times faster is nice :) Nov 28 22:08:41 yes, quite so ! and with only 4 times more cores Nov 28 22:14:08 this is hopefully some bug:) CPU: 162.00 A Nov 28 22:14:43 vrm temp: +64.0°C (crit = +70.0°C) Nov 28 22:14:43 case temp: +46.5°C (crit = +70.0°C) Nov 28 22:14:43 power total: 426.00 W Nov 28 22:14:43 power +12v: 386.00 W Nov 28 22:14:43 power +5v: 28.50 W Nov 28 22:14:46 power +3.3v: 13.00 W Nov 28 22:14:48 curr +12v: 32.25 A (crit max = +100.00 A) Nov 28 22:14:51 curr +5v: 5.69 A (crit max = +40.00 A) Nov 28 22:14:53 curr +3.3v: 4.00 A (crit max = +40.00 A) Nov 28 22:16:15 grep Elapsed tmp-glibc/buildstats/20221128220705/nodejs-16.14.2-r0/do_compile Nov 28 22:16:18 Elapsed time: 491.83 seconds Nov 28 22:18:47 nice Nov 28 22:26:08 webruntime build will be more interesting as it can take advantage of all the cores for more significant portion of do_compile Nov 28 22:28:06 yes, that'll be for tomorrow, maybe Nov 28 22:28:31 or whole image build once you have NVME as well Nov 28 22:41:25 exactly Nov 28 22:41:44 but right now, my pc (and me) will go to sleep :) Nov 28 22:57:07 * JaMa and mine as well **** ENDING LOGGING AT Tue Nov 29 02:59:56 2022