| 16 Nov 2024 |
 Alexandros Liarokapis | * Got any resources I can look into? | 08:06:15 |
| Alexandros Liarokapis | Actially I think the wiki page has enough info to get me started | 08:06:31 |
| Alexandros Liarokapis | * Actually I think the wiki page has enough info to get me started | 08:06:41 |
| Alexandros Liarokapis | .. or not, it is mainly nixos based. | 08:15:50 |
| Alexandros Liarokapis | i guess I may as well try it | 08:16:07 |
 hexa |
error: tensorflow-gpu-2.13.0 not supported for interpreter python3.12
| 20:45:57 |
| hexa | the sound of nixos 24.05 hits hard | 20:46:03 |
| hexa | * error: tensorflow-gpu-2.13.0 not supported for interpreter python3.12
| 20:46:08 |
| hexa | * error: tensorflow-gpu-2.13.0 not supported for interpreter python3.12
| 20:46:12 |
| 17 Nov 2024 |
 Gaétan Lepage | Yes... Let's hope zeuner finds the time to end the TF bump... | 10:38:39 |
| 18 Nov 2024 |
| hexa | wyoming-faster-whisper[4505]: File "/nix/store/dfp38l0dy3n97wvrgz5i62mwvsmshd3n-python3.12-faster-whisper-unstable-2024-07-26/lib/python3.12/site-packages/faster_whisper/transcribe.py", line 145, in __init__
wyoming-faster-whisper[4505]: self.model = ctranslate2.models.Whisper(
wyoming-faster-whisper[4505]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^
wyoming-faster-whisper[4505]: RuntimeError: CUDA failed with error unknown error
systemd[1]: wyoming-faster-whisper-medium-en.service: Main process exited, code=exited, status=1/FAILURE
| 02:09:21 |
| hexa | also loving unknown error errors | 02:09:26 |
| hexa | wyoming-faster-whisper[4745]: File "/nix/store/dfp38l0dy3n97wvrgz5i62mwvsmshd3n-python3.12-faster-whisper-unstable-2024-07-26/lib/python3.12/site-packages/faster_whisper/transcribe.py", line 145, in __init__
wyoming-faster-whisper[4745]: self.model = ctranslate2.models.Whisper(
wyoming-faster-whisper[4745]: ^^^^^^^^^^^^^^^^^^^^^^^^^^^
wyoming-faster-whisper[4745]: RuntimeError: CUDA failed with error no CUDA-capable device is detected
| 02:10:44 |
| hexa | baby steps | 02:10:46 |
| hexa | I can confirm the card is still seated correctly 😄 | 02:10:58 |
| hexa | hardening at work | 02:18:46 |
 connor (burnt/out) (UTC-8) | Ugh I don’t like computers | 05:10:46 |
| connor (burnt/out) (UTC-8) | Anyway in the interest of splitting my attention ever more thinly I decided to start trying to work on some approach toward evaluation of derivations and building them
The idea being to have
- a service which is given a flake ref and an attribute path and efficiently produces a list of attribute paths to derivations exiting under the given attribute path and stores the eval time somewhere
- a service which is given a flake ref and an attribute path to a derivation and produces the JSON representation of the closure of derivations required to realize the derivation, again storing eval time somewhere
- a service which functions as a job scheduler, using historical data about costs (space, time, memory, CPU usage, etc.) and information about locality (existing store paths on different builders) to realize a derivation, which is updated upon realization of a derivation
| 05:18:41 |
| connor (burnt/out) (UTC-8) | Because why have one project when you can have many? | 05:18:55 |
| connor (burnt/out) (UTC-8) | https://github.com/ConnorBaker/nix-eval-graph
And I’ve decided to write it in Rust, which I am self teaching.
And I’ll probably use a graph database, because why not.
And I’ll use NixOS tests for integration testing, because also why not. | 05:20:02 |
| connor (burnt/out) (UTC-8) | All this is to say I am deeply irritated when I see my builders copying around gigantic CUDA libraries constantly. | 05:20:31 |
| connor (burnt/out) (UTC-8) | Unrelated to closure woes, I tried to package https://github.com/NVIDIA/MatX and https://github.com/NVIDIA/nvbench and nearly pulled my hair out. If anyone has suggestions for doing so without creating a patched and vendored copy of https://github.com/rapidsai/rapids-cmake or writing my own CMake for everything, I’d love to hear! | 05:23:26 |
| connor (burnt/out) (UTC-8) | Also, anyone know how the ROCm maintainers are doing? | 05:26:35 |
 SomeoneSerge (back on matrix) | In reply to @connorbaker:matrix.org
Anyway in the interest of splitting my attention ever more thinly I decided to start trying to work on some approach toward evaluation of derivations and building them
The idea being to have
- a service which is given a flake ref and an attribute path and efficiently produces a list of attribute paths to derivations exiting under the given attribute path and stores the eval time somewhere
- a service which is given a flake ref and an attribute path to a derivation and produces the JSON representation of the closure of derivations required to realize the derivation, again storing eval time somewhere
- a service which functions as a job scheduler, using historical data about costs (space, time, memory, CPU usage, etc.) and information about locality (existing store paths on different builders) to realize a derivation, which is updated upon realization of a derivation
Awesome! I've been bracing myself to look into that too. What's your current idea regarding costs and locality? | 07:09:42 |
| SomeoneSerge (back on matrix) | In reply to @connorbaker:matrix.org
Unrelated to closure woes, I tried to package https://github.com/NVIDIA/MatX and https://github.com/NVIDIA/nvbench and nearly pulled my hair out. If anyone has suggestions for doing so without creating a patched and vendored copy of https://github.com/rapidsai/rapids-cmake or writing my own CMake for everything, I’d love to hear! we'd need to do that if were to package rapids itself too, wouldn't we? | 07:11:11 |
