Saturday, 2021-08-14

richardwilbur[m]I’m curious:  lkcl suggested Rockchip quad-core ARM over Intel celery (Celeron) based Chromebooks.  Would you expect the performance of the quad-core ARM to be better than the celery processor?  Do you think there is reason to believe the performance difference would be noticeable from the user interface?00:26
programmerjakeidk for sure, you could look for benchmarks01:10
programmerjakePhoronix is a good source01:11
richardwilbur[m]Thanks for the tip.  I’ll check there.01:13
programmerjakebuilding homemade chips:
lkclprogrammerjake: oooo!09:43
Las[m]richard.wilbur: If you're looking for a good cheap Chromebook, get the Acer Spin 513 Chromebook IMO09:53
Las[m]The Intel ones suck09:53
Las[m]They have either inferior performance or inferior battery life09:53
Las[m]It does support installing standard Linux, and most features work AFAIK09:54
Las[m]I've bought one myself but it hasn't arrived yet, so I haven't verified it myself yet09:54
jnaccording to the SoC is "QC-7C". huh, never heard of that10:02
jnah, qualcomm snapdragon 7c, i guess10:03
Las[m]Yeah it's a lot more powerful than the quite weak RK3399 AFAICT10:15
Las[m]I've been using an ODROID N2 as a daily driver for over a year now, and it uses an Amlogic S922X, which is double the raw compute of that, and it's not very fast.10:17
Las[m]So I can't imagine using the RK339910:17
Las[m]It would be fine for SSH-ing though10:17
jnqcom is much worse than rockchip in one thing though: openness of firmware10:18
jn(which some people here will prioritise)10:18
Las[m]I don't completely agree10:19
Las[m]The GPU side is much more open than the RK339910:19
Las[m]The Panfrost driver had to be reverse engineered10:19
Las[m]I think Freedreno development is funded by Qualcomm10:20
Las[m]Though maybe I'm wrong10:20
jntrue, qcom contributed to free GPU driver development, and ARM (for Mali) didn't10:21
Las[m]Actually, it seems it's changed recently:
jnbut my point was about all the other firmware10:22
Las[m]That is true yeah10:22
jnOh, good news, cool!10:22
jnfor a precise evaluation of the FOSS situation regarding boot firmware, i'd take a look at the coreboot implementation for both platforms10:23
jn(i haven't done this, but that's what i would do)10:24
lkclLas[m], is this it?
lkcldon't think so10:27
lkclah. with a qualcom snapdragon.10:27
jnsometimes vendors sell almost the same device as chromebook and non-chromebook, and it's reflected in the name. not sure if Acer did it here, but i've seen it with Lenovo, i think10:28
jnbut the chromebook and non-chromebook versions will probably have different boot firmware10:29
lkclLas[m]: nice10:29
Las[m]There are also some other similar projects that try to get it running standard Linux10:30
jnanyway, i highly recommend the page linked above, for navigation through the sea of chromebooks10:30
Las[m]I hope the new Framework company makes an ARM laptop10:30
Las[m]thanks for that link10:31
lkcli've done embedded linux oses, it's not hard, it's just "o dear there's no intel AMI BIOS err wtfdoido" is the normal reaction / complaint10:31
lkclthe RK3399 is pretty damn nice.  richardwilbur[m] yes the RK3399 pisses all over intel celery10:31
lkcleven the NEON-compiled ffmpeg is capable of doing 720p video decode, not even using the on-board accelerated VPU (!)10:32
lkclha, nicknamed kevin :)10:32
lkclahh that's the same codename as on that Cadmium page10:33
jnchromebook codenames are fun, and very important when looking at code10:33
lkclwoo, $499 though
lkclcompared to $200 for an RK3288 flip-top10:34
Las[m]The Acer one is the same price10:37
* lkcl got first version of svp64 sv.bc instruction working in the simulator11:46
Las[m]lkcl: Do you BTW have any ETA for when the first Libre SoC real processor will be made? 2025?11:48
lkclheck no, way earlier than that11:48
lkclsingle-core crypto-router ASIC, about 8-ish months11:49

Generated by 2.17.1 by Marius Gedminas - find it at!