sevan | not so excited about mini-itx hardware, more interested by risc-v hardware. Sipeed's new board which are going to come out soon look interesting https://sipeed.com/licheepi4a | 00:27 |
---|---|---|
sevan | we just need to cut a new notch on the Sipeed board so that it'll fit on the mnt reform? ;) | 00:28 |
bluerise | I'm less interested by RISC-V HW | 00:30 |
bluerise | minute: still waiting for that CM4 adapter! | 00:30 |
bluerise | the reason I'm not ordering the 8MP module is that OpenBSD doesn't have a driver for ath10k and with ath11k already shipping in current machines... I just don't see myself spending time on writing a driver... | 00:31 |
bluerise | ath10k and ath11k are massive :( | 00:31 |
sigrid | feeding pdf specs to chat gpt and getting a driver in c when | 00:34 |
sevan | :D | 00:34 |
sevan | maybe throw in an infiband one too for added speed | 00:35 |
bluerise | honestly, feeding the Linux driver to Chat GPT and getting an OpenBSD driver... | 00:35 |
bluerise | it's BSD licensed, so license isn't the issue | 00:35 |
bluerise | and, haha, specs for qualcomm atheros, sure :D | 00:35 |
sigrid | "openbsd remix" | 00:35 |
sigrid | the spec on that ath didn't seem as massive as the linux driver | 00:36 |
sigrid | but that's just my ignorant opinion | 00:36 |
bluerise | I remember the mvpp driver in Linux. GPL licensed. But Semihalf released a BSD licensed one for UEFI, which was basically the Linux GPL one but with some style changes. So I took the UEFI BSD one and added some sed... | 00:36 |
sigrid | heh | 00:37 |
bluerise | Talked to someone from Semihalf later, and they said "oh yeah, we basically threw the GPL one (also written by Semihalf) into sed for UEFI" | 00:37 |
bluerise | so basically it was Linux --sed--> UEFI --sed--> OpenBSD | 00:37 |
- vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:20) | 00:40 | |
minute | bluerise: the cm4 adapter works, but pi has messed up dsi | 00:47 |
minute | bluerise: so i'm still cooking up a cursed hdmi to edp adapter | 00:47 |
minute | bluerise: or do you wanna go a311d? | 00:47 |
bluerise | rk356x, but a311d would also be interesting | 00:48 |
bluerise | I have 2 rk356x modules. Radxa and ... and... shit | 00:48 |
bluerise | Ah, pine64 | 00:48 |
minute | i got a311d to boot linux on it. kind of | 00:48 |
minute | ah. the pine one didn't want to say anything | 00:48 |
minute | everybody told me that rk3568 was somehow disappointing | 00:49 |
- S0rin (QUIT: Ping timeout: 252 seconds) (~S0rin@user/s0rin) | 00:49 | |
minute | though. so i didn't bother too much with it | 00:49 |
bluerise | I'd love to have a look | 00:49 |
bluerise | so if you can build one CM4 for me, I'd happily send euros over | 00:49 |
minute | ok. we just ordered 40 of those sockets that are needed | 00:50 |
bluerise | cool | 00:50 |
bluerise | and I'm also taking the hdmi to edp adapter once that is available :D | 00:50 |
bluerise | mostly because it's easier to bootstrap :) | 00:50 |
minute | they were unavailable a few months ago, but are well stocked now | 00:50 |
minute | ok! so i have to respin that | 00:50 |
minute | it uses stdp2600 | 00:50 |
minute | slightly cursed, it has a 8086 core | 00:50 |
bluerise | Hah | 00:51 |
bluerise | my google search remembers I looked stdp2600 up, creepy | 00:51 |
+ S0rin (~S0rin@user/s0rin) | 00:51 | |
bluerise | the "good" thing on the rk356x is that OpenBSD support is pretty good as well. | 00:51 |
bluerise | don't have a MIPI-DSI driver though... but edp, heh | 00:52 |
bluerise | and they usually have a brcmfmac wifi | 00:52 |
bluerise | anyway, looking forward, let me know as soon as I can buy one, even if it's merely a prototype :) | 00:53 |
sigrid | are there plans to do a rk3588? how does one even get those | 00:55 |
bluerise | I sent the imx8mq reform2 diff to u-boot, hope someone has a look soon | 00:55 |
sigrid | nice! | 00:55 |
bluerise | I haven't had any time to make progress on pcie (since it broke after the rebase to mainline), maybe I should just port lcdif first and then look at pcie again... | 00:56 |
bluerise | sigrid: I think they are using RK3588S instead of RK3588, because the RK3588 footprint is too huge for CM4 | 00:57 |
bluerise | There's https://www.cnx-software.com/2022/09/08/radxa-cm5-rockchip-rk3588s-module-raspberry-pi-cm4/ | 00:57 |
minute | sigrid: i have icore 3588, and will make an adapter for it | 00:57 |
minute | cm5 is not shipping | 00:57 |
bluerise | minute: that's pretty cool as well | 00:57 |
minute | (i hope it will at some point, though) | 00:57 |
bluerise | The A311D seems a bit boring, only A73s... | 00:59 |
sigrid | minute: i see | 00:59 |
bluerise | minute: which one of these? https://www.firefly.store/category.php?id=3 | 01:00 |
bluerise | iCore-3588Q? | 01:01 |
- mtm (QUIT: Ping timeout: 256 seconds) (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 01:03 | |
minute | bluerise: yup | 01:03 |
bluerise | Wonder how the pinouts look like... | 01:04 |
minute | bluerise: a311d is boring but well supported, incl panfrost | 01:04 |
minute | bluerise: the pinouts are... creative | 01:04 |
minute | a mix of different connectors | 01:04 |
bluerise | How many PCIe controllers? | 01:05 |
bluerise | looks like no WiFi on the SoM, so that must go onto the board or adapter | 01:05 |
minute | dunno atm | 01:06 |
minute | gotta hit the hay! good night! | 01:06 |
bluerise | night! | 01:06 |
+ vagrantc (~vagrant@2600:3c01:e000:21:7:77:0:20) | 01:17 | |
- vagrantc (QUIT: Client Quit) (~vagrant@2600:3c01:e000:21:7:77:0:20) | 01:19 | |
- cinap_lenrek (QUIT: Ping timeout: 256 seconds) (~cinap_len@ns3076381.ip-147-135-136.eu) | 02:36 | |
+ mtm (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 03:09 | |
- Guest9779 (QUIT: Ping timeout: 260 seconds) (~nicolas@108-49-142-46.pool.kielnet.net) | 03:12 | |
+ nsc (~nicolas@234-48-142-46.pool.kielnet.net) | 03:13 | |
* nsc -> Guest824 | 03:14 | |
- Ar|stote|is (QUIT: Quit: https://quassel-irc.org - Chat comfortably. Anywhere.) (~linx@149-210-16-8.mobile.nym.cosmote.net) | 05:03 | |
+ Ar|stote|is (~linx@149-210-16-8.mobile.nym.cosmote.net) | 05:04 | |
- Ar|stote|is (QUIT: Client Quit) (~linx@149-210-16-8.mobile.nym.cosmote.net) | 05:05 | |
+ Ar|stote|is (~linx@149-210-16-8.mobile.nym.cosmote.net) | 05:05 | |
+ vagrantc (~vagrant@2600:3c01:e000:21:7:77:0:20) | 06:53 | |
+ bgs (~bgs@212-85-160-171.dynamic.telemach.net) | 07:07 | |
- bgs (QUIT: Remote host closed the connection) (~bgs@212-85-160-171.dynamic.telemach.net) | 07:49 | |
- vagrantc (QUIT: Quit: leaving) (~vagrant@2600:3c01:e000:21:7:77:0:20) | 08:46 | |
+ MajorBiscuit (~MajorBisc@c-001-032-042.client.tudelft.eduvpn.nl) | 09:26 | |
- doppler (QUIT: Ping timeout: 265 seconds) (~doppler@user/doppler) | 09:56 | |
+ bgs (~bgs@212-85-160-171.dynamic.telemach.net) | 12:01 | |
+ doppler (~doppler@user/doppler) | 12:08 | |
- bgs (QUIT: Remote host closed the connection) (~bgs@212-85-160-171.dynamic.telemach.net) | 12:55 | |
- mtm (QUIT: Ping timeout: 252 seconds) (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 13:04 | |
* wielaard -> mjw | 13:19 | |
+ bgs (~bgs@212-85-160-171.dynamic.telemach.net) | 14:24 | |
+ mtm (~mtm@c-71-228-84-213.hsd1.fl.comcast.net) | 15:09 | |
+ GNUmoon2 (~GNUmoon@gateway/tor-sasl/gnumoon) | 16:36 | |
- GNUmoon (QUIT: Ping timeout: 255 seconds) (~GNUmoon@gateway/tor-sasl/gnumoon) | 16:39 | |
- S0rin (QUIT: Ping timeout: 260 seconds) (~S0rin@user/s0rin) | 17:29 | |
+ S0rin (~S0rin@user/s0rin) | 17:30 | |
- MajorBiscuit (QUIT: Quit: WeeChat 3.6) (~MajorBisc@c-001-032-042.client.tudelft.eduvpn.nl) | 17:35 | |
bkeys | minute: My best friend says he would pre order a Reform if you made an adapter for the iCore-3588Q; (hell I'd prefer that over even cm4), is this something you are seriously gonna make? | 18:12 |
+ vagrantc (~vagrant@2600:3c01:e000:21:7:77:0:20) | 18:28 | |
minute | bkeys: yes, i will make it | 18:50 |
bkeys | How is the pocket reform launch going? | 18:52 |
+ oomono (uid328183@id-328183.tinside.irccloud.com) | 18:57 | |
minute | bkeys: good, for the video we need some final case parts that we can't get this week because of CNY | 18:58 |
minute | otherwise the device is p. much done | 18:58 |
bkeys | Do you have a loose estimate of when launch would be? My friend is really interested in it cause he likes small laptops | 18:58 |
minute | so we're prepping the US demo unit and the crowdsupply demo unit | 18:58 |
minute | bkeys: mid feb i hope | 18:58 |
minute | depends on how quickly we get the video shoot done and the review of financials | 18:59 |
bkeys | Cool beans, I'm looking forward to the Reform meetup in Seattle and I support your vision 100% man | 19:00 |
bkeys | (I know you won't be in Seattle) | 19:00 |
minute | cool! our guy bnys will be there | 19:03 |
vagrantc | ACTION will try to make it to seattle again | 19:25 |
josch | the iCore-3588Q can carry up to 32 GB of RAM?? o0 | 19:31 |
+ bkeys1 (~Thunderbi@157.201.98.1) | 19:32 | |
minute | josch: yeah, not sure if they manufactured that yet, but i shall ask | 19:34 |
minute | i think mine has 16? | 19:34 |
eery | that name is a bit unfortunate haha, when I first read it I saw "Core i3-3588Q" | 19:49 |
- mjw (QUIT: Killed (NickServ (GHOST command used by markw!~wielaard@gnu.wildebeest.org))) (~mjw_@2001:1c06:2488:1400:4fd:39a7:74ac:7bae) | 20:22 | |
+ wielaard (~mjw_@2001:1c06:2488:1400:4fd:39a7:74ac:7bae) | 20:22 | |
- bkeys1 (QUIT: Ping timeout: 248 seconds) (~Thunderbi@157.201.98.1) | 20:33 | |
josch | eery: that might be the marketing trick making you think it's just as good ;) | 20:34 |
minute | according to passmark, Intel Core i3-13100 has around double the score as rk3588 | 20:40 |
klardotsh | josch: I've never seen an RK3588 with 32GB actually sold. They were in all the marketing promo stuff but I've only ever seen 4, 8, 16GB SKUs actually sold anywhere. | 20:44 |
klardotsh | that said, a 16GB RK3588 would be an absolute monster SOM | 20:44 |
josch | it would -- that's more RAM than my old intel laptop had! | 20:45 |
eery | Definitely would be closer to fully replacing my other hardware | 20:49 |
josch | eery: because you need more ram? | 20:51 |
josch | i've done everything on my reform up to compiling linux and libreoffice without having run out of ram yet | 20:52 |
eery | I mean yeah, I regularly use a full 32 GB of memory on my laptops for $DAYJOB stuff, and the 64 GB on my desktop is nice | 20:52 |
eery | granted, that's kind of a function of core count (12/32) and parallel build jobs | 20:53 |
josch | ah indeed, that's true -- for my job i need 32+ GB of ram | 20:53 |
eery | so with fewer cores there's less need for more RAM | 20:53 |
josch | but i have a intel desktop in my office for that | 20:53 |
klardotsh | my desktop has 32GB RAM and compresses 1/4 of that with zram and I still have run out of even *swap* if I leave too much work stuff open and accidentally run like "cargo build" | 20:53 |
josch | klardotsh: you compress 1/4 or you compress it to 1/4 of its original size? | 20:54 |
klardotsh | but that's a rare and specialty case. incredible though that my first desktop in 2008 had a whopping 1GB of RAM and I ran Gentoo on that thing with few issues. fast forward a hot minute and I can pulverize 32GB no problem... | 20:54 |
eery | on the topic of CPU performance though, I'm pleased that the imx8mq manages half the performance of my i5-6200U | 20:55 |
eery | in a random synthetic benchmark I tried | 20:55 |
josch | i can confirm -- my old cpu was a core i5-5300U | 20:56 |
klardotsh | josch: that was a somewhat inaccurate simplification on my part. zram makes an 8GB (25% of my physical RAM) swap space backed by a zstd-compressed block in physical RAM, so I likely lose more like 5-6GB of physical RAM. it's hard to tell in htop (if it even denotes the backing block at all...) | 20:56 |
eery | You can see the total size in RAM via zramctl btw | 20:57 |
klardotsh | josch: zram 1GB blocks are used on my Reform and my also-4GB-RAM Skylake-based Chromebook. modern webshit gets real finnicky without it (read: tab crashes. I actually still get those on the 4GB devices, just less of them) | 20:57 |
klardotsh | eery: whoa. mind blown. ty! compression ratio is way better than I thought, and also is using lz4, not zstd like I used to use. Disk is 7.8G, I'm storing 924.3M of data, and that compresses down to 295.5M | 20:59 |
minute | it's kind of incredible how big the performance differences are between intel core generations | 21:00 |
sigrid | i'd be fine with 4gb just as before, personally | 21:20 |
minute | yeah, i'm not so often constrained by ram, more by cpu | 21:22 |
vagrantc | anyone ever floated the idea of making a module that basically turned the mnt/reform into a portable battery/dock/monitor? | 21:24 |
vagrantc | is that feasible in the "cpu" module slot? | 21:25 |
minute | vagrantc: hmm! could be doable | 21:27 |
vagrantc | might also be interesting with the pocket reform | 21:27 |
vagrantc | minute: i know you probably have a lot of ideas on your plate already... :) | 21:28 |
vagrantc | battery/keyboard/dock/monitor :) | 21:29 |
vagrantc | kind of like a nexdock or whatever | 21:29 |
minute | it would even be possible to expose the usb hub to the outside | 21:30 |
minute | and hdmi could be an input and convert to eDP via stdp2600 | 21:30 |
minute | the building blocks are there ;) | 21:30 |
minute | (except that my stdp2600 design doesn't work yet) | 21:31 |
vagrantc | would presumably want a usb-c display ... alternate mode or whatever it's called | 21:35 |
minute | yeah, but big reform doesn't have usb-c :D | 21:35 |
minute | would need to smuggle another port in there somehow | 21:36 |
vagrantc | on an extension cable off of the module :) | 21:36 |
minute | mhmm | 21:36 |
vagrantc | :) | 21:36 |
- bgs (QUIT: Remote host closed the connection) (~bgs@212-85-160-171.dynamic.telemach.net) | 22:48 | |
- Gooberpatrol_66 (QUIT: Ping timeout: 260 seconds) (~Gooberpat@user/gooberpatrol66) | 23:07 | |
- sknebel (QUIT: Remote host closed the connection) (~quassel@v22016013254630973.happysrv.de) | 23:18 | |
- _nrb_ (QUIT: Remote host closed the connection) (~nrbnrb@xmpp01.you-host.me) | 23:19 | |
+ sknebel (~quassel@v22016013254630973.happysrv.de) | 23:19 | |
+ _nrb_ (~nrbnrb@xmpp01.you-host.me) | 23:20 | |
chartreuse | Just gotta turn one of the 3.0 ports on it into a usb-c port :P | 23:41 |
chartreuse | I see ram mentioned, 4GB has been enough on the reform though of course more is always preferred. Though of course there's not really room for a SODIMM slot or such in there | 23:45 |
eery | As an aside, USB-C ports can be 2.0/EHCI | 23:46 |
eery | cheap android phones have 2.0 C ports lol | 23:46 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!