!xmLtiCaAJxfhURjrXl:matrix.org

NixOS RISC-V

231 Members
NixOS on RISC-V https://wiki.nixos.org/wiki/RISC-V https://pad.lassul.us/NixOS-riscv64-linux https://github.com/orgs/NixOS/teams/risc-v70 Servers

Load older messages


SenderMessageTime
10 Mar 2024
@5m5z3q888q5prxkg:chat.lightnovel-dungeon.deKREYREN ⚡️🦀 (TH1520 in Licheepi 4A)
ERROR: tests/cp/proc-short-read
===============================

cp: skipping file '/proc/cpuinfo', as it was replaced while being copied
sed: can't read 1: No such file or directory
proc-short-read.sh: set-up failure: 
ERROR tests/cp/proc-short-read.sh (exit status: 99)

Seems that coreutils are failing to build through qemu x86_64-linux -> riscv64-linux

17:03:58
@raitobezarius:matrix.orgraitobezariusBut a lot of code was in suboptimal state to have a realistic cluster AFAIK15:49:13
@raitobezarius:matrix.orgraitobezariusI am still interested to build one15:49:20
@raitobezarius:matrix.orgraitobezariusBut maybe we should define a core set of packages15:49:27
@thefossguy:matrix.orgPratham Patel (you can mention me)The VisionFive2, though having the same cores as the HiFive Unmatched, has come uarch improvements (kinda like a "v2" of the core), making it a bit faster for compilation. But gets severely hindered by half the RAM (8) compared to what you get on the Unmatched (16).15:50:25
@raitobezarius:matrix.orgraitobezariusI'd not say that RAM is the biggest deal tbf15:51:04
@raitobezarius:matrix.orgraitobezariusYou can eval nixpkgs on a x8615:51:09
@thefossguy:matrix.orgPratham Patel (you can mention me) Not the biggest, but is a pain point tbh 15:51:19
@raitobezarius:matrix.orgraitobezariusFor what exactly?15:51:28
@thefossguy:matrix.orgPratham Patel (you can mention me) and no PCIe support yet (patches sent, getting merged and might be in 6.10), so can't use swap that "well" 15:52:32
@raitobezarius:matrix.orgraitobezariusYeah I have seen the PCIe circus15:54:22
@raitobezarius:matrix.orgraitobezariusThat's my biggest annoyance15:54:29
@raitobezarius:matrix.orgraitobezariusBecause NVMe for builders would be very nice 15:54:43
@thefossguy:matrix.orgPratham Patel (you can mention me)
In reply to @raitobezarius:matrix.org
For what exactly?
well, fedora packaging might be crazy but something this insane might've been a need
https://matrix.to/#/!GhnNOKFLyvWBmnjqRS:fedoraproject.org/$C8oSRQ2RM6rpaAjTQpshPYtYs1bXbMRlXWIUdcT648A?via=fedoraproject.org&via=fedora.im&via=matrix.org
15:54:45
@raitobezarius:matrix.orgraitobezariusI'd be curious which package is this one15:55:36
@raitobezarius:matrix.orgraitobezariusI wonder if this is not just a spec for speed reason rather than actual requirement, e.g. chromium linkage15:55:56
@thefossguy:matrix.orgPratham Patel (you can mention me)
In reply to @raitobezarius:matrix.org
Because NVMe for builders would be very nice

highly agree, but no idea how nice

AFAIK, the U74 is equivalent to A55 (correct me if I'm wrong) and haven't used an NVMe on it yet. On my RK3588, I saw a literal 100% improvement in building a defconfig kernel from the NVMe compared to when the source was on the SD Card

15:57:05
@thefossguy:matrix.orgPratham Patel (you can mention me)
In reply to @raitobezarius:matrix.org
I'd be curious which package is this one
I'm not even kidding, I'm too scared to look
15:57:21
@raitobezarius:matrix.orgraitobezarius
In reply to @thefossguy:matrix.org

highly agree, but no idea how nice

AFAIK, the U74 is equivalent to A55 (correct me if I'm wrong) and haven't used an NVMe on it yet. On my RK3588, I saw a literal 100% improvement in building a defconfig kernel from the NVMe compared to when the source was on the SD Card

Not really surprised
15:57:42
@thefossguy:matrix.orgPratham Patel (you can mention me)the halving was surprising for me though, I knew it would be better, but didn't know that it would be by this much!15:58:11
@skeuchel:matrix.orgSteven Keuchelis that with system or user mode emulation?17:08:58
@alex:tunstall.xyzAlex
In reply to @raitobezarius:matrix.org
I'd not say that RAM is the biggest deal tbf
It is necessary for certain packages.
(e.g. the Zig compiler, unless that's now been updated to a less memory-intensive version?)
17:14:20
@raitobezarius:matrix.orgraitobezarius
In reply to @alex:tunstall.xyz
It is necessary for certain packages.
(e.g. the Zig compiler, unless that's now been updated to a less memory-intensive version?)
how much RAM the Zig compiler take in your experience?
17:17:03
* @5m5z3q888q5prxkg:chat.lightnovel-dungeon.deKREYREN ⚡️🦀 (TH1520 in Licheepi 4A) submitted https://github.com/NixOS/nixpkgs/issues/294775 | Build failure: coreutils (riscv64-linux) 17:16:32
@raitobezarius:matrix.orgraitobezariusjust to have a rough idea17:17:06
@5m5z3q888q5prxkg:chat.lightnovel-dungeon.deKREYREN ⚡️🦀 (TH1520 in Licheepi 4A)https://github.com/misuzu/nixos-vf2/blob/master/flake.nix#L3 Is this binary cache usable?17:19:33
@raitobezarius:matrix.orgraitobezariusit'd be interesting to collect that data tbh17:17:10
@sorear:matrix.orgsoreari feel like "know who runs it" is the bare minimum security-wise for a binary cache17:49:02
@alex:tunstall.xyzAlex
In reply to @raitobezarius:matrix.org
how much RAM the Zig compiler take in your experience?
Last time I tried to build it, it got OoM killed (8 GiB VF2), so I have no idea.
18:59:11
@alex:tunstall.xyzAlexIIRC I also tried with a lot of swap (can't remember how much), but it thrashed and still eventually failed.19:00:55

Show newer messages


Back to Room ListRoom Version: 10