| 10 May 2025 |
Michael | * did you included esos.elf in your initfs? (without it, it hangsm though my log looked different) Maybe for reference: https://github.com/cwt-opi-rv2/linux-cwt-orangepi-ky | 15:48:02 |
Michael | * did you included
esos.elf in your initfs? (without it, it hangs, though my log looked different)
Maybe for reference: https://github.com/cwt-opi-rv2/linux-cwt-orangepi-ky | 17:57:17 |
| 11 May 2025 |
misuzu | Thanks! I've managed to make it work: https://gitlab.com/misuzu/hydra-riscv64/-/commit/f0074904a768816ed826dc59644736b4a6729e1c https://hydra.ztier.in/build/10852 | 09:53:38 |
Tristan Ross | Had a nice little discussion with the founder of DeepComputing. I asked about community distro's like NixOS supporting their hardware and he mentioned things will be open sourced soon. No idea on what exactly it'll take for getting everything to work but sounds promising. | 16:55:04 |
Tristan Ross | My hope is to get that new Framework RISC-V system with the 8 Sifive P550 cores and use that as a secondary dev machine but mainly for RISC-V stuff. | 16:55:52 |
| 12 May 2025 |
Tristan Ross | It took almost 20 hours to compile gfortran on my VisionFive2 lol | 16:24:17 |
| 13 May 2025 |
vixea | Sheesh | 00:27:58 |
Tristan Ross | 855 / 1737 derivations built for my RISC-V system. | 14:54:32 |
Alyssa Ross | Not sure if this will interest anybody but me, but I figured out why pkgsStatic Rust builds don't work on riscv64 :) https://github.com/NixOS/nixpkgs/pull/406797 | 17:10:47 |
Alyssa Ross | * | 17:12:15 |
| 14 May 2025 |
Alyssa Ross | Is the VisionFive2 the RISC-V computer to get, or is there anything out there with more RAM? | 08:23:10 |
fgaz | In reply to @qyliss:fairydust.space Is the VisionFive2 the RISC-V computer to get, or is there anything out there with more RAM? there's the milk-v pioneer, but it's pricey. alternatively, EIC7700X boards (milk-v megrez, pine64 starpro64) go up to 32G but they are all out of stock | 08:35:56 |
Steven Keuchel | The p550 cores (e.g. EIC7700X) are currently the fastest commercially available ones. | 08:41:07 |
Steven Keuchel | Much better experience than the VF2. | 08:41:23 |
fgaz | and apparently spacemit m1 based stuff also goes up to 16G but it isn't faster than vf2 despite the higher core count | 08:41:45 |
Steven Keuchel | Pioneer has more cores, but the cores themselves are slower. | 08:41:47 |
Alyssa Ross | Pioneer seems unavailable anyway | 08:48:23 |
Alyssa Ross | In reply to @fgaz:matrix.org there's the milk-v pioneer, but it's pricey. alternatively, EIC7700X boards (milk-v megrez, pine64 starpro64) go up to 32G but they are all out of stock Any idea how mainline support is for the EIC7700X boards? | 08:49:45 |
Steven Keuchel | No mainline support yet. Vendor kernel needs some custom CONFIG values to compile on Nixos. | 08:57:23 |
Alyssa Ross | yikes | 08:57:39 |
Steven Keuchel | At least the vendor uboot has distro boot enabled by default. | 08:57:57 |
Alyssa Ross | maybe visionfive is the way to go then | 08:58:04 |
Steven Keuchel | Depends on what you want to do. | 08:59:02 |
Alyssa Ross | mostly I want to test generic OS images I'm building, so only systems with good mainline support are really interesting | 09:00:33 |
fgaz | then vf2 is 100% the way to go at the moment | 09:01:42 |
Steven Keuchel | And you build those through cross compilation or do you want to build them on the device? | 09:02:25 |
Alyssa Ross | Native | 09:03:01 |
fgaz | * then vf2 is 100% the way to go at the moment ah right native compilation | 09:04:13 |
Alyssa Ross | (I know it will be very slow, that's fine) | 09:05:06 |
fgaz | the "low" amount of RAM is also a problem for some packages, but a bit of swap is enough to compensate in my experience | 09:07:51 |