| darwin | when apt(-get) builds a kernel, how can I opt out of that or force it to only use one thread/job? | 08:51 |
|---|---|---|
| darwin | actually it's when I had to restart with 'dpkg --configure -a' because currently having it use all threads is overheating | 08:52 |
| darwin | it's doing dkms for some reason | 08:54 |
| gnarface | darwin: maybe you're using nvidia non-free drivers? they rely on dkms. off the top of my head i don't know where to override the build jobs though, maybe there's some environment variable... | 09:07 |
| gnarface | if you're having problems overheating on such a short build though there may be other issues at hand. make sure you have acpi installed and running, and make sure your fans are actually functional | 09:08 |
| darwin | i had ACPI automatically installed | 09:09 |
| gnarface | ok good | 09:09 |
| darwin | i will take it to local PC shop to ask if they can see what's causing overheating | 09:09 |
| darwin | fans are running | 09:10 |
| darwin | but one might be old | 09:10 |
| gnarface | if they're still spinning it's probably fine, but do a visual inspection and make sure they're not clogged up maybe | 09:10 |
| darwin | they're not | 09:10 |
| gnarface | there is the possibility of some kernel issue causing the fans to not spin fast enough, but usually the hardware knows what to do even without drivers. usually, but not always | 09:11 |
| gnarface | well, maybe there's some dkms reference somewhere that can tell you an easy way to set the job count | 09:12 |
| gnarface | it probably just sets it to the amount of cpu cores you have, but it shouldn't be hard to change | 09:13 |
| darwin | i have kernel 6.6.32 | 09:16 |
| darwin | but this has been going on since I reinstalled the system-/logic-/main-/mother-board so put new heatsink paste on | 09:17 |
| darwin | a new used one | 09:18 |
| darwin | yeah, I was using the non-free nvidia drivers because in the past nouveau had poor performance or didn't work at all | 09:29 |
| darwin | i guess I could try them again if I won't need dkms | 09:30 |
| gnarface | darwin: yea, the nouveau ones won't need dkms i'm pretty sure, but they also suck | 09:52 |
| gnarface | if you re-applied the heatsync recently maybe it's worth evaluating the methods again... using too much thermal compound is a common mistake, as is failing to fully or levelly mount the actual heatsync. on some models | 09:54 |
| gnarface | "heatsync, on some models" i meant to say | 09:55 |
| gnarface | darwin: oh, you can probably also get readouts of the fan RPMs from some or another tool, to make sure they're within expected parameters. usually i use lm-sensors. | 10:01 |
| gnarface | (after installing lm-sensors, just run "sensors-detect" as root and answer yes to all the questions, then whenever you run the "sensors" command as any user afterwards it will give you a text dump of current temperature and fan readouts, subject to the accuracy of some compiled-in database of hardware which admittedly is highly incomplete) | 10:02 |
| darwin | the temperature fluctuates about 35 degrees centigrade/celsius when doing number-crunching/compiling | 10:03 |
| darwin | within a second or two | 10:03 |
| gnarface | that does seem like a lot. | 10:03 |
| darwin | like 56 to 91 | 10:03 |
| darwin | then it stas around 90 for a while and reboots | 10:03 |
| darwin | stays | 10:04 |
| gnarface | 56 already sounds high as a starting number, though appropriate ranges do vary by device... | 10:04 |
| gnarface | you sure the heatsink is for the right modle? | 10:04 |
| gnarface | model* | 10:04 |
| darwin | when not doing much it might be 34 | 10:04 |
| darwin | yes | 10:04 |
| gnarface | and you're sure the power for the fan is hooked to the right connector? | 10:04 |
| darwin | yes--as said, they're all spinning | 10:04 |
| gnarface | hmm | 10:04 |
| gnarface | how much thermal compound did you use? | 10:05 |
| darwin | but the cooler is three years old and the PC shop says these wear out in two or three years sometimes | 10:05 |
| darwin | about pea-sized | 10:05 |
| gnarface | seems like a lot, i go for about a rice grain size, and spread it out with a credit card | 10:05 |
| darwin | i was told by them don't spread it out, just put the cooler on and let it do that | 10:05 |
| gnarface | i also make sure that there isn't any extra hanging over the sides of where the contact surfaces are, in the air gap | 10:06 |
| gnarface | yea, they sound wrong | 10:06 |
| darwin | this was several years ago already | 10:06 |
| gnarface | i mask off the contact area with tape and actually make a tape-thin coating just on the contact surface | 10:06 |
| darwin | it changed since last time I did that | 10:06 |
| darwin | most the instructions I see still say use pea-sized | 10:07 |
| darwin | and some say if you don't screw it on enough, it won't flatten it out enough, or if you tighten it too tight, it'll come out the edges | 10:08 |
| gnarface | you want enough thermal compound to fill microscopic cracks in the contact surfaces but you don't want any spilling into the air gap around the chip core like a sandwich with too much mayo... i've seen lots of instructions that are complete bullshit designed to sell more hardware replacements | 10:08 |
| darwin | i see | 10:08 |
| gnarface | i can't be sure that's your actual problem without really seeing it myself, but i've seen this done wrong enough to make things worse for sure | 10:09 |
| darwin | i used Arctic MX-6 (different company than Arctic Silver apparently) but read I should've used NT-H2 or liquid metal | 10:09 |
| darwin | there's no way to see it on mine either | 10:09 |
| darwin | it's a water cooler that clips on and it's hard to see the edges | 10:09 |
| gnarface | hmm, yea never heard of that one. the legit arctic silver brand is usually good for several degrees colder than the generic stuff though, assuming you don't put too much on | 10:09 |
| gnarface | what is the actual CPU in question here? | 10:10 |
| darwin | AMD Ryzen 9 5950X and the cooler is Corsair HX100i I think | 10:10 |
| darwin | it was running fine for maybe a year before the old system-board had a capacitor pop | 10:10 |
| darwin | it had 3950X before that | 10:11 |
| gnarface | hmm, that's pretty new, i wouldn't expect the base operating temperature to be that high on something so new, though... i should ask, do you know which cpu performance governor is set in the kernel and which "quiet and cool" profile is set in the bios? | 10:26 |
| gnarface | 56 idle sounds more like early athlon 64 numbers with a stock cooler | 10:27 |
| darwin | i don't know which CPU performance governor is set | 10:27 |
| darwin | well, it's more like low-to-mid 30s idle | 10:27 |
| gnarface | run this to check: cat /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor | 10:27 |
| gnarface | mid to low 30's sounds normal | 10:27 |
| darwin | i don't think I set 'quiet and cool' | 10:27 |
| gnarface | but i doubt it should be getting much above mid 50's at peak | 10:28 |
| darwin | i thought that would slow it down... have almost never set that | 10:28 |
| gnarface | ok, so, set it, but set it to "performance" mode | 10:28 |
| gnarface | the bios quiet and cool setting, that is | 10:28 |
| gnarface | not to be confused with the kernel cpufrequency governor also named "performance", but generally you want the former if you're using the latter | 10:28 |
| darwin | ondemand | 10:29 |
| gnarface | ondemand is a good default, leave it that way for now | 10:29 |
| gnarface | but still set the bios quiet and cool setting to "performance" | 10:29 |
| gnarface | basically "quiet and cool" is actually just the branding they're using for regular thermal threshold controls, and you want it to "performance" so it will not be conservative about how high to spin the fan. it would be more accurate if they had named it "quiet or cool" | 10:30 |
| gnarface | but i've seen it default to quiet if left alone or "off" | 10:30 |
| gnarface | and if you're having overheat problems under load, that could definitely be related | 10:31 |
| darwin | i see | 10:31 |
| gnarface | to be clear, just enabling quiet+cool itself will absolutely not limit the fan speed implicitly, it just gives the kernel control over thermal thresholds | 10:34 |
| gnarface | it's the actual profile setting you select under that, which determines what sort of fan speeds go with each thermal threshold the kernel calls | 10:34 |
| gnarface | make sense? | 10:34 |
| darwin | yes | 10:34 |
| gnarface | (i know it's misleading and unintuitive, with the labeling terminology they use, i've made this mistake before myself) | 10:35 |
| gnarface | if that turns out to have been part of the issue, i'm sure there will be an audible difference in the fan noise under load | 10:40 |
| darwin | yes | 10:40 |
| darwin | that's fine | 10:40 |
| darwin | is it only certain/newer BIOS that have this setting? I didn't see it in ones from 2010s... | 10:58 |
| darwin | my main PC BIOS doesn't have that option | 11:30 |
| darwin | it just has fan modes quiet, aggressive, and PC case with three fan settings | 11:30 |
| darwin | i'd already put it on aggressive but maybe put the case fans in middle | 11:30 |
| gnarface | darwin: do you have the amd64-microcode package from non-free installed? | 11:44 |
| gnarface | oh, as of daedalus that one is probably in non-free-firmware actually now, it's been moved | 11:45 |
| gnarface | i don't remember hearing anything specifically about temperatures in it, but i know it had bug fixes... | 11:46 |
| ted-ious | darwin: If you have a water cooler it should take many seconds for the temperature to spike not 2. | 11:48 |
| ted-ious | How long ago did you put the thermal paste on and what has happened during that time? | 11:48 |
| gnarface | hmm, suddenly i'm having deja-vu about this temperature issue with the water cooler | 13:28 |
| gnarface | some software issue with the temperatures? i forget... | 13:29 |
| ted-ious | All kinds of crazy behavior can happen when your cpu has a fever. :) | 13:54 |
| fonky | evening all | 20:46 |
| fonky | for daedalus live sources.list is oldstable instead of deadalus | 20:46 |
| fonky | with security? | 20:46 |
| Xenguy_ | fonky, fsmithred probably has a clue about that... | 20:55 |
| Xenguy_ | Maybe pastebin the sources.list file? | 20:55 |
| fonky | its the default | 20:56 |
| Xenguy_ | Sounds odd but I don't really use the live stuff myself | 20:56 |
| fonky | daedalus-security & daedalus-updates | 20:56 |
| fonky | it is displayed it is not available since 24days | 20:57 |
| fsmithred | which iso are you using? | 20:58 |
| fonky | live-desktop not minimal | 20:58 |
| fsmithred | arch? | 20:58 |
| fonky | amd64 | 20:58 |
| fsmithred | fonky, does it actually say "oldstable" or does it say "chimaera"? | 21:03 |
| fsmithred | fonky, I think you need to run 'aptitude update && aptitude safe-upgrade' and say Yes when it asks. | 21:07 |
| fonky | nono i was just asking what to change fsmithred | 21:25 |
| fsmithred | fonky, what are you trying to do? The sources.list in the daedalus isos says daedalus. | 21:26 |
| fonky | yes, but i got a message that the updates and security have been depreciated for 24days | 21:26 |
| fonky | im sorry if everything is in order im sorry for actually bothering | 21:27 |
| fsmithred | I just tried it and got a message that said testing switched to stable | 21:27 |
| fonky | it was however the first time i did it since this system is up, and used apt-get update && apt-get upgreade | 21:27 |
| fonky | upgrade* | 21:27 |
| fonky | in order to get a newer firefox-esr | 21:28 |
| fsmithred | those commands should work | 21:28 |
| fonky | they did, 0 errors | 21:28 |
| fonky | i dont have that terminal open to post a screenshot | 21:28 |
| fsmithred | oh, ok. Sometimes you'll get an error if you happen to hit a mirror in the middle of when the mirror is upgrading. | 21:29 |
| fsmithred | usually corrects in a few minutes. | 21:29 |
| fonky | if i put deb.devuan.org/merged ceres main i dont have to use the two lines daedalus-security and daedalus-updates ? | 21:32 |
| fsmithred | right. there is only one line for ceres (unstable) or excalibur (current testing) | 21:34 |
| fsmithred | If you want to test testing/unstable, I think there are excalibur installer isos. | 21:35 |
| fsmithred | Testing the isos would be helpful. | 21:35 |
| fonky | glad to be of help | 21:36 |
| fonky | just link them | 21:36 |
| fonky | well live :) | 21:36 |
| fsmithred | and upgrading to ceres from that is pretty simple - it's a week behind. | 21:36 |
| fonky | the pc i am at has a broken hd2000 intel graphics | 21:36 |
| fsmithred | no live isos for excalibur/ceres | 21:36 |
| fonky | oh :) | 21:36 |
| fsmithred | it changes to fast to be practical | 21:37 |
| fonky | ill try ceres, although i dont use much but firefox and libreoffice | 21:37 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!