| - mjw (QUIT: Ping timeout: 252 seconds) (~mjw@gnu.wildebeest.org) | 00:21 | |
| - chomwitt1 (QUIT: Ping timeout: 260 seconds) (~chomwitt@2a02:587:7a13:5400:1ac0:4dff:fedb:a3f1) | 00:28 | |
| - mtm (QUIT: Ping timeout: 244 seconds) (~textual@47.202.75.129) | 01:02 | |
| + mtm (~textual@47.202.75.129) | 01:06 | |
| - vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:50) | 01:48 | |
| - mtm (QUIT: Ping timeout: 252 seconds) (~textual@47.202.75.129) | 02:06 | |
| + mtm (~textual@47.202.75.129) | 02:08 | |
| - L29Ah (QUIT: Read error: Connection reset by peer) (~L29Ah@wikipedia/L29Ah) | 03:00 | |
| + L29Ah (~L29Ah@wikipedia/L29Ah) | 03:03 | |
| - cobra (QUIT: Ping timeout: 268 seconds) (~cobra@user/Cobra) | 03:04 | |
| - paperManu (QUIT: Ping timeout: 272 seconds) (~paperManu@107.159.71.33) | 03:19 | |
| - L29Ah (QUIT: Read error: Connection reset by peer) (~L29Ah@wikipedia/L29Ah) | 03:26 | |
| - nsc (QUIT: Ping timeout: 265 seconds) (~nicolas@i5C74DFD7.versanet.de) | 03:28 | |
| + nsc (~nicolas@137-97-142-46.pool.kielnet.net) | 03:29 | |
| + cobra (~cobra@user/Cobra) | 03:59 | |
| - mtm (QUIT: Ping timeout: 265 seconds) (~textual@47.202.75.129) | 04:04 | |
| + mtm (~textual@47.202.75.129) | 04:06 | |
| wickedshell | minute in talk of the workstation type stuff, is the server stuff still a back burner item? | 04:22 |
|---|---|---|
| - mtm (QUIT: Ping timeout: 246 seconds) (~textual@47.202.75.129) | 04:23 | |
| - wickedshell (QUIT: Read error: Connection reset by peer) (~wickedshe@2601:8c0:800:4baa:d52d:9d68:8254:652) | 04:30 | |
| + chomwitt1 (~chomwitt@2a02:587:7a13:5400:1ac0:4dff:fedb:a3f1) | 06:37 | |
| + wickedshell (~wickedshe@2601:8c0:800:4baa:d52d:9d68:8254:652) | 06:39 | |
| - colinsane (QUIT: Quit: bye) (~colinunin@97-113-92-101.tukw.qwest.net) | 07:12 | |
| + colinsane (~colinunin@97-113-92-101.tukw.qwest.net) | 07:14 | |
| - colinsane (QUIT: Quit: bye) (~colinunin@97-113-92-101.tukw.qwest.net) | 07:24 | |
| + colinsane (~colinunin@97-113-92-101.tukw.qwest.net) | 07:30 | |
| - colinsane (QUIT: Ping timeout: 272 seconds) (~colinunin@97-113-92-101.tukw.qwest.net) | 07:35 | |
| - Ar|stote|is (QUIT: Ping timeout: 268 seconds) (~linx@149.210.20.235) | 07:40 | |
| + mjw (~mjw@gnu.wildebeest.org) | 08:12 | |
| - mjw (QUIT: Ping timeout: 248 seconds) (~mjw@gnu.wildebeest.org) | 08:56 | |
| [tj] | I haven't really used altra much, but I did make one do 400Gbit/s of TLS | 09:36 |
| [tj] | so pcie can't be that bad | 09:36 |
| [tj] | ampere are the cpu vendor for the switch 2, I'm not sure how that helps anything, but it might | 09:37 |
| + glu__ (~glu@user/glu) | 09:55 | |
| - glu_ (QUIT: Ping timeout: 268 seconds) (~glu@user/glu) | 09:55 | |
| gl0b | morning, is anyone able to help me troubleshoot a pocket reform (rk3588) power issue? | 10:29 |
| [tj] | sure | 10:30 |
| gl0b | thanks :) i'm finding that when my battery gets to around 70% or lower any increase in power draw will cause a hard reboot. it least I think that's what is going on as nothing gets logged. | 10:33 |
| gl0b | has anyone else experienced similar? | 10:34 |
| [tj] | maybe the training on the charge controller is bad? | 10:36 |
| minute | gl0b: oh weird, but i have a similar issue at the very low end of charge. what are the voltages of the cells in that state? | 10:37 |
| gl0b | i didn | 10:38 |
| minute | for me the gauge thinks we're at 15% but it's more lik 0% so i need to check how to reset the gauge | 10:38 |
| gl0b | whoops | 10:38 |
| gl0b | i didn't note voltage, but i will once it runs down further | 10:38 |
| [tj] | I wonder what the flash life time is on the 2040 | 10:41 |
| gl0b | i can force the behaviour running stress-ng and turning up the backlight, but it most often happens when i'm below 70% and load a heavy web page in firefox | 10:42 |
| [tj] | I noticed the keyboard backlight is a significant power user | 10:42 |
| minute | gl0b: if you could report the cell voltages at that point, that would be great | 10:46 |
| minute | [tj]: about altra, what was the system setup? altra as a pcie card/accelerator? or ethernet to altra? | 10:46 |
| [tj] | let me check what I can | 10:47 |
| [tj] | https://store.avantek.co.uk/ampere-altra-server-2u-mt-snow.html | 10:48 |
| [tj] | I'm not sure if it was a mt.snow or mt.jade machine | 10:48 |
| [tj] | it was in that class, it might not have been production hardware | 10:48 |
| [tj] | it was 3 years ago so its a bit fuzzy and I could only submit commands via teams | 10:49 |
| minute | [tj]: oh ok. we use a rented altra machine for building the reform system images btw... but i can't test gpu with that :D | 10:54 |
| minute | (it's at hetzner) | 10:54 |
| minute | my personal vps is also arm based at netcup, my guess is they use ampere too, but i'm not sure. i run immich on it which works great | 10:55 |
| [tj] | I can ask around | 10:55 |
| gl0b | oh it just happened while restarting firefox. 81% 3.8v on both cells, keyboard light off, display backlight at 20% | 10:59 |
| gl0b | i have an nvme ssd installed too | 11:00 |
| gl0b | idle current draw is usually in the 600-650mA range | 11:01 |
| - cow321 (QUIT: Ping timeout: 252 seconds) (~deflated8@user/meow/deflated8837) | 11:07 | |
| minute | gl0b: ok that doesn't sound normal hmmmm | 11:09 |
| hramrach | I checked nad I can also access an Altra remotely, theoretically but that does not say much about the PCIe problems | 11:10 |
| [tj] | we (freebsd) have a bunch of them | 11:11 |
| hramrach | My impression is that they tested the devices they cared about for datacenter use, and some PCIe features not used by those are broken | 11:11 |
| [tj] | they probably care a lot about gpus right now | 11:12 |
| hramrach | but not back then | 11:12 |
| [tj] | from my interactions with ampere they are very interested in doing anything to make the platform more saleable | 11:12 |
| hramrach | but saleable to who? the Arm workstation niche is pretty small | 11:13 |
| [tj] | but cloud gpu is a huge thing right now | 11:14 |
| hramrach | Also using the GPU for compute and using it for graphics is different | 11:14 |
| - chomwitt1 (QUIT: Ping timeout: 260 seconds) (~chomwitt@2a02:587:7a13:5400:1ac0:4dff:fedb:a3f1) | 11:17 | |
| minute | also, nvidia rtx are supposed to work fine on altra with nvidia's own arm64 drivers, so there must be some special sauce to make it work | 11:20 |
| minute | source https://github.com/AmpereComputing/NVIDIA-GPU-Accelerated-Linux-Desktop-on-Ampere?tab=readme-ov-file#hardware | 11:23 |
| ch | after sleeping on it, i'd vote for an MNT Network Audio player | 11:29 |
| + cow321 (~deflated8@user/meow/deflated8837) | 11:30 | |
| sknebel | ch: there are the OtterCast devices as open designs, but I dont think anyone is currently selling those premade... | 11:32 |
| ch | huh | 11:34 |
| hramrach | MNT NAP :) | 11:35 |
| ch | sknebel: do you also happen to know an hdmi switch with audio breakout and ip? :) | 11:35 |
| gl0b | minute: is there anything else i should be checking? i did open it up and double check that my power cables were properly connected a couple of days ago. | 11:35 |
| sknebel | ch: hm.... dont think so | 11:35 |
| minute | gl0b: maybe complete cpu module reseat? i need to drive now and will get back to you | 11:37 |
| gl0b | minute: thanks, i'll give that a go in the meantime | 11:39 |
| - wytch1 (QUIT: Ping timeout: 265 seconds) (~wytch@user/wytch) | 11:48 | |
| gordon1 | the state of suspend on rk3588 is very sad, looks like dram suspend isn't even supported in TF-A, i also tried to do test_resume with suspend to disk, it did restore but lost nvme (and probably all pcie) somewhere in process, logs are here https://0x0.st/8mo0.txt | 12:00 |
| gordon1 | minute: ^ tested on rk3588 rock5b+ | 12:00 |
| + L29Ah (~L29Ah@wikipedia/L29Ah) | 12:02 | |
| Zaba | [tj]: Nvidia is the SoC vendor for the Switch 2, same as the original Switch. “Ampere” also happens to be an nvidia GPU microarchitecture name | 12:03 |
| gordon1 | with collabora tf-a and u-boot and mainline kernel | 12:05 |
| gordon1 | oops, i can't read, it said "failed to recover" | 12:11 |
| gordon1 | *load | 12:11 |
| gordon1 | with nvme disconnected it looks like it sorta succeeded https://0x0.st/8mHs.txt | 12:16 |
| * Guest9808 -> mjw | 12:22 | |
| gordon1 | did anyone measured reform consumption in s2idle btw? | 12:40 |
| + paperManu (~paperManu@107.159.71.33) | 12:50 | |
| - cow321 (QUIT: Read error: Connection reset by peer) (~deflated8@user/meow/deflated8837) | 12:52 | |
| [tj] | Zaba: well that was confusing, thanks for the correctin | 12:53 |
| + gustav28 (~gustav@c-78-82-52-87.bbcust.telenor.se) | 13:02 | |
| + wytch (~wytch@user/wytch) | 13:03 | |
| gl0b | minute: ok! crisis averted. turns out my power cables were not properly seated at the motherboard end despite me having checked previously. it took a good bit more force than i expected to get them completely in there. i'm now on 17% 3.3v and can run stress-ng without it rebooting. | 13:14 |
| hramrach | and was it charging with the cable not seated correctly? | 13:18 |
| hramrach | I guess it's hard to tell with the bug that sometimes causes sysctl to report as charging when the cable is removed :/ | 13:19 |
| hramrach | I really like the PSUs that tell you if the device negotiated power delivery, and how much it's drawing | 13:20 |
| L29Ah | hramrach: can you advise one? | 13:26 |
| L29Ah | i only seen that on "power bank" controllers | 13:26 |
| gordon1 | pine64 desktop usb shows that | 13:33 |
| gordon1 | https://pine64.com/product/pinepower-120w-desktop-power-supply-us-version/ that one | 13:33 |
| gordon1 | but it has only one usb-c | 13:34 |
| + chomwitt1 (~chomwitt@2a02:587:7a13:5400:1ac0:4dff:fedb:a3f1) | 13:34 | |
| + cow321 (~deflated8@user/meow/deflated8837) | 13:34 | |
| gordon1 | i would love to have similar thing but more portable | 13:35 |
| gordon1 | alternatively you can use one of these bypass usb pd power monitors if you need it for debug https://m.media-amazon.com/images/I/515TDOXglEL._AC_.jpg | 13:37 |
| hramrach | I only have a local noname one. Somebody linked an anker one earlier, and that's also quite big. It's portable in the sense you can lift it without a forklift | 13:40 |
| hramrach | https://www.alza.de/alzapower-m500-display-multi-charge-power-delivery-200w-mit-display-d7377619.htm | 13:43 |
| minute | gordon1: super interesting @ resume from disk, maybe feasible if OS is on emmc? | 13:44 |
| gordon1 | minute: i think whole pcie is borked after resume, so if you have anything there it won't work | 13:44 |
| hramrach | you can get PD cables that light up when PD is negotiated from the same supplier | 13:45 |
| gordon1 | maybe it will work if you unload whole pcie module and load it after resue? | 13:45 |
| gordon1 | *resume | 13:45 |
| gordon1 | also there is this irq 47 hiccup but not 100% sure if that will cause any problems, fb console brings itself back after resume but i am testing it inside the initrd so don't have wayland or xorg running for obvious reasons | 13:47 |
| hramrach | this is the anker one posted earlier https://www.anker.com/uk/products/a2345-anker-prime-charger-250w-6-ports-ganprime?variant=46089442001060 | 13:47 |
| josch | gordon1: somebody claims that they have working suspend/resume with rk3588 on the rock 5b+ board: https://fedi.imu.li/notice/ArB2GY3qVMAFwjnbZg | 13:47 |
| gordon1 | >it’s on radxa’s debian image | 13:48 |
| gordon1 | well, that's not an option for me, but thanks for the info | 13:48 |
| ch | but something to investigate | 13:48 |
| gordon1 | radxa kernel sources are absolutely dreadful btw | 13:48 |
| gordon1 | and aren't even rebased on mainline | 13:49 |
| hramrach | aut of the three the pinepower onle looks biggest and lamest so maybe worth looking at the other two | 13:49 |
| gordon1 | at least that was true for pcie drivers | 13:49 |
| gordon1 | i'm pretty sure major part of suspend is just handled in their closed-source tf-a | 13:50 |
| hramrach | and that's closed source why exactly /o\ | 13:55 |
| gordon1 | because they can | 13:56 |
| gordon1 | i mean, not them exactly, i think it's from rockchip | 13:57 |
| gordon1 | that thing https://github.com/rockchip-linux/rkbin/tree/master/bin/rk35 | 13:57 |
| hramrach | but what's in it for rockchip to release crap like this? | 14:03 |
| hramrach | they could as well release the source | 14:03 |
| hramrach | if somebody really wanted to steal their secret sauce they can disassemble it | 14:03 |
| gordon1 | after checking their code i think the obvious reason is they just too ashamed to release it | 14:04 |
| gordon1 | i certainly would be | 14:04 |
| ch | less work for rk i guess | 14:04 |
| hramrach | it might be that it's more difficult to get legal clearance, somebody has to review that it does not contain anything that should not be released | 14:07 |
| hramrach | the tf-a is quite small and very specific so it's unlikely but anyway | 14:07 |
| josch | in a similar context, i was unable to contact the right person at NXP responsible to look into the licensing situation which makes the blobs for imx8mq unredistributable if you are not MNT | 14:13 |
| gordon1 | i checked the code for rk3299 and it's about 1k LOC so it is not massive to implement it from scratch, i do not know how PITA is the actual debugging and testing it tho | 14:59 |
| gordon1 | and iirc user manual for rk3588 is available so there shouldn't be anything unknwon | 15:00 |
| hramrach | it's a Chinese manual | 15:00 |
| gordon1 | i'm pretty sure i saw an english one, let me check | 15:00 |
| hramrach | "shouldn't be anything unknown" hardly applies to hard quality manuals, much less to manuals for Chinese chips | 15:01 |
| gordon1 | i mean sure | 15:01 |
| gordon1 | at least we won't have any handicap | 15:02 |
| gordon1 | https://github.com/FanX-Tek/rk3588-TRM-and-Datasheet looks english to me | 15:02 |
| hramrach | that's why the sample code is provided. there is usually no way to write that from the manual alone, without consulting the people who made the chip | 15:03 |
| hramrach | it's not about the language, it's about the origin. it's a manual describing Chinese budget chip | 15:03 |
| + ndufresne (~ndufresne@apple.collaboradmins.com) | 15:34 | |
| minute | hramrach: i think this is unnecessary bashing. if you compare it with imx8mp trm, it's not that much of a quality difference | 15:38 |
| minute | gordon1: if we have ~some~ working suspend/resume we can document the bugs one by one and ask some people at collabora if they're motivated to help | 15:39 |
| + mark_ (~mjw@gnu.wildebeest.org) | 15:43 | |
| - mark_ (QUIT: Ping timeout: 248 seconds) (~mjw@gnu.wildebeest.org) | 16:26 | |
| - iank (QUIT: Quit: ZNC 1.8.2+deb2+deb11u1 - https://znc.in) (~iank@fsf/staff/iank) | 16:29 | |
| + iank (~iank@fsf/staff/iank) | 16:32 | |
| gordon1 | minute: yeah, i sorta torn now on how should i proceed because i sorta need that board running, but i certainly would sink more time into it when i get my reform | 16:59 |
| - bkeys (QUIT: Quit: With every step we take, danger will follow closely) (~Thunderbi@38-146-94-247.echocast.zone) | 17:02 | |
| + bkeys (~Thunderbi@38-146-94-247.echocast.zone) | 17:02 | |
| + bkeys1 (~Thunderbi@172.56.68.8) | 17:03 | |
| gordon1 | but hibernate looks much better than i anticipated, so hopefully there is not much there to fix | 17:05 |
| - bkeys (QUIT: Ping timeout: 252 seconds) (~Thunderbi@38-146-94-247.echocast.zone) | 17:06 | |
| * bkeys1 -> bkeys | 17:06 | |
| + bkeys1 (~Thunderbi@38-146-94-247.echocast.zone) | 17:24 | |
| - bkeys (QUIT: Ping timeout: 244 seconds) (~Thunderbi@172.56.68.8) | 17:25 | |
| * bkeys1 -> bkeys | 17:25 | |
| irimi1 | minute: i’ve experimented with the wifi on my pocket reform a bit, and it is indeed better than before. i just tested it in an unfortunate spot in my flat, it seems. | 17:28 |
| irimi1 | minute: what threw me off is that /etc/modprobe.d/reform-qcacld2.conf is owned by the 'reform-qcacld2' package, and modifications get overwritten on update (possible that i overlooked a prompt when updating it) | 17:30 |
| - bkeys (QUIT: Ping timeout: 252 seconds) (~Thunderbi@38-146-94-247.echocast.zone) | 17:30 | |
| josch | irimi1: why do you want to modify /etc/modprobe.d/reform-qcacld2.conf? | 17:30 |
| + bkeys (~Thunderbi@38-146-94-247.echocast.zone) | 17:31 | |
| josch | the reform-qcacld2 package should probably install the file into /usr/lib/modprobe.d/reform-qcacld2.conf instead | 17:32 |
| - bkeys (QUIT: Read error: Connection reset by peer) (~Thunderbi@38-146-94-247.echocast.zone) | 17:37 | |
| + bkeys1 (~Thunderbi@38-146-94-247.echocast.zone) | 17:37 | |
| josch | draft mr for this: https://source.mnt.re/reform/qcacld2/-/merge_requests/8 | 17:39 |
| * bkeys1 -> bkeys | 17:40 | |
| + vagrantc (~vagrant@2600:3c01:e000:21:7:77:0:50) | 17:51 | |
| - bkeys (QUIT: Ping timeout: 260 seconds) (~Thunderbi@38-146-94-247.echocast.zone) | 18:02 | |
| + bkeys (~Thunderbi@38-146-94-247.echocast.zone) | 18:11 | |
| - bkeys (QUIT: Read error: Connection reset by peer) (~Thunderbi@38-146-94-247.echocast.zone) | 18:17 | |
| + helgoman (~helgoman@adsl-178-38-58-229.adslplus.ch) | 18:22 | |
| + bkeys (~Thunderbi@2607:fb90:7e72:3f8:f4ba:6b73:9043:b30f) | 18:27 | |
| - bkeys (QUIT: Quit: With every step we take, danger will follow closely) (~Thunderbi@2607:fb90:7e72:3f8:f4ba:6b73:9043:b30f) | 18:42 | |
| + bkeys (~Thunderbi@172.58.1.14) | 18:42 | |
| - chomwitt1 (QUIT: Ping timeout: 246 seconds) (~chomwitt@2a02:587:7a13:5400:1ac0:4dff:fedb:a3f1) | 18:55 | |
| + mark_ (~mjw@gnu.wildebeest.org) | 19:00 | |
| * mjw -> Guest9651 | 19:00 | |
| - Guest9651 (QUIT: Killed (iridium.libera.chat (Nickname regained by services))) (~mjw@2001:1c06:2486:a800:7602:5eff:dc71:a72c) | 19:00 | |
| * mark_ -> mjw | 19:00 | |
| + Guest9651 (~mjw@2001:1c06:2486:a800:7602:5eff:dc71:a72c) | 19:01 | |
| + bkeys1 (~Thunderbi@38-146-94-247.echocast.zone) | 19:41 | |
| + chomwitt1 (~chomwitt@2a02:587:7a13:5400:1ac0:4dff:fedb:a3f1) | 19:43 | |
| - bkeys (QUIT: Ping timeout: 244 seconds) (~Thunderbi@172.58.1.14) | 19:43 | |
| * bkeys1 -> bkeys | 19:43 | |
| + jacobk_ (~quassel@47-186-65-73.dlls.tx.frontiernet.net) | 19:55 | |
| jfred | x elite som in a reform would be pretty sweet | 20:07 |
| jfred | if it'd fit | 20:08 |
| jfred | minute: One possibility for a mobile device is to break out a USB connection on some pogo pins on the back. Would make it possible to add a cell modem attachment or other fun things later. Agreed that a phone would be too hard to get right | 20:14 |
| - jacobk_ (QUIT: Ping timeout: 276 seconds) (~quassel@47-186-65-73.dlls.tx.frontiernet.net) | 20:17 | |
| + jacobk (~quassel@47-186-65-73.dlls.tx.frontiernet.net) | 20:17 | |
| - b0 (QUIT: Read error: Connection reset by peer) (~b0@user/b0) | 20:24 | |
| + b0 (~b0@user/b0) | 20:27 | |
| - BoostisBetter (QUIT: Ping timeout: 260 seconds) (4a410829d7@irc.cheogram.com) | 21:15 | |
| - SavagePeanut (QUIT: Ping timeout: 260 seconds) (59eaa45ac7@irc.cheogram.com) | 21:16 | |
| + SavagePeanut (59eaa45ac7@irc.cheogram.com) | 21:19 | |
| hramrach | I wonder if it's better to put pogopins on the device, or ony contacts, and let the peripheral take care about the pogopins | 21:23 |
| - mhoye (QUIT: Ping timeout: 244 seconds) (~mhoye@li319-32.members.linode.com) | 21:43 | |
| + mhoye (~mhoye@li319-32.members.linode.com) | 21:43 | |
| + Ar|stote|is (~linx@149.210.17.8) | 21:54 | |
| - helgoman (QUIT: Quit: Konversation terminated!) (~helgoman@adsl-178-38-58-229.adslplus.ch) | 22:07 | |
| - gustav28 (QUIT: Quit: Quit) (~gustav@c-78-82-52-87.bbcust.telenor.se) | 22:15 | |
| - bkeys (QUIT: Quit: With every step we take, danger will follow closely) (~Thunderbi@38-146-94-247.echocast.zone) | 23:01 | |
| + bkeys (~Thunderbi@38-146-94-247.echocast.zone) | 23:01 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
