!eWOErHSaiddIbsUNsJ:nixos.org

NixOS CUDA

325 Members
CUDA packages maintenance and support in nixpkgs | https://github.com/orgs/NixOS/projects/27/ | https://nixos.org/manual/nixpkgs/unstable/#cuda64 Servers

Load older messages


SenderMessageTime
27 Apr 2026
@yorik.sar:matrix.orgyorik.sarIt’s not our commit message. We’re just syncing branch from NixOS/nixpkgs to nixos-cuda/nixpkgs using API.05:54:42
@shadowrz:nixos.devYorusaka MiyabiOh, forgot this05:56:00
@yorik.sar:matrix.orgyorik.sarhttps://docs.github.com/en/rest/branches/branches?apiVersion=2026-03-10#sync-a-fork-branch-with-the-upstream-repository - this one to be precise05:57:01
@yorik.sar:matrix.orgyorik.sarI have no idea why GitHub decides to add an event to every issue because of this.05:58:04
@yorik.sar:matrix.orgyorik.sarMaybe the previous event from the master was stuck in split brain or smth.05:59:15
@hexa:lossy.networkhexaif it appears on a handful of PRs per day probably not09:15:27
@yorik.sar:matrix.orgyorik.sarDoes it trigger something in our infra?09:16:24
@hexa:lossy.networkhexa I saw it on two consecutive PRs 09:16:41
@hexa:lossy.networkhexaand didn't look further09:16:47
@ninja:worldethicaldataforum.orgNinja joined the room.14:31:41
@sporeray:matrix.orgRobbie BuxtonI’ve been having some fun recently with libcupti infinitely recursing during builds trying to find libcuda (ignoring the stub). This seems to be a new issue only in CUDA 13 plus. It doesn’t try and open libcuda in 12. Curious if anyone has run into this before?19:01:29
@sigmanificient:matrix.orgSigmanificient joined the room.21:20:15
@hexa:lossy.networkhexa
Download
21:20:33
@hexa:lossy.networkhexapretty sure the name is confusing people21:20:50
@hexa:lossy.networkhexahydra-channel-foo makes it sound official21:20:56
@sigmanificient:matrix.orgSigmanificientyeah 21:21:03
@amaan:amaanq.comamaan joined the room.21:48:43
@glepage:matrix.orgGaétan Lepage True, we can rename this. cc yorik.sar 22:20:11
@glepage:matrix.orgGaétan LepageI have not run into this. If you find out why, I can surely make a PR.22:20:30
@alexch01:matrix.orgAlex C joined the room.23:00:47
@sporeray:matrix.orgRobbie BuxtonAh it’s no stress and easy to get around, just wanted to make sure I wasn’t wasting work if someone had already solved it23:10:13
30 Apr 2026
@glepage:matrix.orgGaétan Lepage FYI: cudaPackages_13 now points cudaPackages_13_2 (which is at 13.2.0). 15:14:44
@glepage:matrix.orgGaétan Lepage * FYI: cudaPackages_13 now points to cudaPackages_13_2 (which is at 13.2.0). 15:15:10
@lostmsu:matrix.orglostmsu joined the room.17:30:10
@hexa:lossy.networkhexa Gaétan Lepage is cache.nixos-cuda.org being hammered or why is it so slow to respond? 23:47:40
@hexa:lossy.networkhexaand any reason there's no ipv6 support?23:54:15
1 May 2026
@glepage:matrix.orgGaétan Lepage I am not aware. It seems okay right now, but maybe it was under heavy load. SomeoneSerge (matrix works sometimes) recently added the Anubis firewall to the web portal.
Also, it's the same machine hosting the actual hydra instance (including evaluator and queue runner) and evals are sometimes giving it a hard time.
21:07:45
@glepage:matrix.orgGaétan Lepage I didn't initially set up ipv6, but now it is available (2a01:4f9:3071:1108::1). Do you mean that the hydra instance is not correctly configured w.r.t. ipv6? 21:11:13
@hexa:lossy.networkhexadns.nix has ipv6 but that has not been propagated into dns22:08:46
@hexa:lossy.networkhexa
❯ nix eval .#dns --json | jq
{
  "domains": {
    "nixos-cuda": {
      "dnsProvider": "hetzner_v2",
      "domain": "nixos-cuda.org",
      "records": [
        {
          "address": "185.199.108.153",
          "label": "@",
          "type": "a"
        },
        {
          "address": "185.199.109.153",
          "label": "@",
          "type": "a"
        },
        {
          "address": "185.199.110.153",
          "label": "@",
          "type": "a"
        },
        {
          "address": "185.199.111.153",
          "label": "@",
          "type": "a"
        },
        {
          "address": "144.76.101.55",
          "label": "ada",
          "type": "a"
        },
        {
          "address": "95.216.20.88",
          "label": "atlas",
          "type": "a"
        },
        {
          "address": "37.27.129.22",
          "label": "hydra",
          "type": "a"
        },
        {
          "address": "2a01:4f9:3071:1108::1",
          "label": "hydra",
          "type": "aaaa"
        },
        {
          "address": "45.154.216.118",
          "label": "oxide-1",
          "type": "a"
        },
        {
          "address": "95.216.72.164",
          "label": "pascal",
          "type": "a"
        },
        {
          "label": "www",
          "target": "@",
          "type": "cname"
        },
        {
          "label": "cache",
          "target": "hydra",
          "type": "cname"
        },
        {
          "label": "grafana",
          "target": "hydra",
          "type": "cname"
        }
      ],
      "registrar": "none"
    }
  },
  "outputs": {
    "config": "// nixos-cuda\nD(\"nixos-cuda.org\",\nNewRegistrar(\"none\"),\nDnsProvider(NewDnsProvider(\"hetzner_v2\")),\nPURGE,\nA(\"@\", \"185.199.108.153\"),\nA(\"@\", \"185.199.109.153\"),\nA(\"@\", \"185.199.110.153\"),\nA(\"@\", \"185.199.111.153\"),\nA(\"ada\", \"144.76.101.55\"),\nA(\"atlas\", \"95.216.20.88\"),\nA(\"hydra\", \"37.27.129.22\"),\nAAAA(\"hydra\", \"2a01:4f9:3071:1108::1\"),\nA(\"oxide-1\", \"45.154.216.118\"),\nA(\"pascal\", \"95.216.72.164\"),\nCNAME(\"www\", \"@\"),\nCNAME(\"cache\", \"hydra\"),\nCNAME(\"grafana\", \"hydra\"))",
    "creds": "! sops decrypt --extract \"['dns-creds']\" /nix/store/r1vznya21dycgk664f6x0pqn3qmwlws0-source/.secrets.yaml"
  },
  "settings": {
    "sops": {
      "extractString": "['dns-creds']",
      "file": "/nix/store/r1vznya21dycgk664f6x0pqn3qmwlws0-source/.secrets.yaml"
    }
  }
}
22:29:02

Show newer messages


Back to Room ListRoom Version: 9