9 Jul 2024 |
charuto | architecture i mean. | 22:52:29 |
hexa | once plausible build hardware turns up | 23:37:17 |
10 Jul 2024 |
dramforever | is that the official answer? | 03:29:00 |
Mindavi | When there's hardware _and_ maintainers step up to maintain builds for the architecture for a long enough time to consider it stable and long lasting, probably | 06:54:59 |
Mindavi | I don't think there's an official answer, but that's at least what's needed for it to become a consideration (I think) | 06:55:32 |
11 Jul 2024 |
hexa | "turns up" is doing a lot lot of heavy lifting | 00:36:33 |
hexa | meaning … it must exist, it must be buyable/rentable for the foundation | 00:36:57 |
hexa | * meaning … it must exist, it must be buyable/rentable for the foundation, it must be able to pull some weight to build at least nixos-unstable-small in a reasonable time frame | 00:37:15 |
hexa | it should be rackable, if it needs to be housed somewhere | 00:37:34 |
hexa | the device should have mainline linux support and a reasonably standard boot story | 00:38:13 |
hexa | like, we started providing aarch64-linux support roughly when there was uefi-boot support and server grade hardware | 00:39:00 |
hexa | the early machines were probably cavium thunderx | 00:40:43 |
hexa | then qualcomm centriq | 00:40:46 |
hexa | and now ampere altra | 00:40:50 |
hexa | and then there is work-on-arm https://github.com/WorksOnArm/equinix-metal-arm64-cluster/issues/30 | 00:45:08 |
charuto | is something like https://milkv.io/jupiter or similar SBCs not good candidates in terms of hardware? | 01:16:34 |
artemist | That specific board has no mainline linux and only 8 cores of unknown but probably poor performance. The Milk-V pioneer is a bit closer (64 cores, more RAM supported), but mainline linux on it isn't usable yet. It's mATX which you can get rack cases for, but you're still DIYing with dev hardware | 02:43:55 |
artemist | I suspect loongarch will be ready for servers before risc-v, but there are probably very few nixos users on it | 02:47:21 |
| myrkskog joined the room. | 03:16:34 |
dramforever | that makes sense, more of the riscv effort is focused on the embedded side just because there the customizablility matters more | 05:57:05 |
| linj joined the room. | 13:32:44 |
John Ericson | dramforever: I am very found for cross compilation because the work I've put in, of course :) | 15:46:08 |
John Ericson | But I wanted to point out a nice thing that can be done is cross compiling the packages, and then running tests natively | 15:46:28 |
John Ericson | to need fewer Risc-V builders (whether physical or an emulator) | 15:47:08 |
John Ericson | Also, content-addressing derivations, which really need to get rolled out this year, should help make mixing native and cross builds less tedious | 15:47:43 |
aleksana (force me to bed after 18:00 UTC) | In reply to @Ericson2314:matrix.org Also, content-addressing derivations, which really need to get rolled out this year, should help make mixing native and cross builds less tedious Content addressed is only blocked by one or two hydra problems now | 16:03:27 |
John Ericson | yes | 16:04:01 |
John Ericson | basically just me having too many things I worry about at once :) | 16:04:13 |
John Ericson | the hydra problems aren't even that bad | 16:04:19 |
| pasukon removed their display name pasukon. | 16:06:41 |