| 17 Jan 2023 |
 K900 | Timing? | 21:02:47 |
| K900 | Try to stick a sleep 1 in there? | 21:02:56 |
 Arson Copperflame | Right not I'm rebuilding with it removed to make sure I'm not hallucinating or something | 21:03:48 |
| Arson Copperflame | But it's a good idea, i'll try that next | 21:04:23 |
| Arson Copperflame | Since I made that commit, it somehow works even with the for ... echo $x removed... So maybe it was something external after all that got fixed by forcing nix to build a new tarball (either by changing a file or having a different git rev) | 21:20:53 |
| 18 Jan 2023 |
 Sandro 🐧 | Should we do a wsl maintainer team and put programs like wslu under it? | 13:12:57 |
| Arson Copperflame | Do you mean in nixpkgs? | 14:15:19 |
| Sandro 🐧 | yes | 14:56:08 |
| Arson Copperflame | I guess you'd have to ask the maintainer of the wslu package, but I think it'd make sense | 16:06:01 |
| 19 Jan 2023 |
| K900 | Oh no | 07:54:53 |
| K900 | The latest preview update breaks us | 07:55:03 |
| K900 | I think they fixed /dev/shm | 07:55:08 |
| K900 | https://github.com/nix-community/NixOS-WSL/pull/199 | 08:11:55 |
| Sandro 🐧 | In reply to @k900:0upti.me
I think they fixed /dev/shm fixed or "fixed"? | 10:08:51 |
| K900 | Fixed | 10:09:06 |
| K900 | Like actually | 10:09:08 |
| K900 | But that broke our fix | 10:09:14 |
| 21 Jan 2023 |
|  Buxel joined the room. | 22:13:42 |
| 22 Jan 2023 |
|  Mic92 left the room. | 20:51:42 |
 lambadada | Has anyone gotten CUDA working in WSL? How do I check if WSL can actually see the card? | 23:26:34 |
| K900 | It's not actually ever going to "see the card" | 23:36:30 |
| K900 | At best it'll see a virtual device that will forward compute workloads to the real ome | 23:36:42 |
| K900 | * At best it'll see a virtual device that will forward compute workloads to the real one | 23:36:48 |
| lambadada | Has anyone gotten CUDA working in WSL? How do I check if WSL can actually see the virtual device that will forward compute workloads to the real one? | 23:37:55 |
| K900 | What I'm trying to say is there is no device | 23:38:16 |
| K900 | You need to run an actual CUDA application | 23:38:24 |
| K900 | And you need to point it at the WSL-provided (really vendor-provided) drivers that are in /usr/lib/WSL somewhere | 23:38:59 |
| K900 | * And you need to point it at the WSL-provided (really vendor-provided) drivers that are in /usr/lib/wsl somewhere | 23:39:08 |
| lambadada | Okay, I'll look into that. They do mention a library from the windows side that can be invoked. | 23:40:28 |
| lambadada | Will check back once I have a bit more info. | 23:40:34 |
