| 25 Nov 2024 |
 connor (burnt/out) (UTC-8) | You can see what I eventually hope to move in-tree here: https://github.com/ConnorBaker/cuda-packages
Here’s the update script I’ve made for the different redists: https://github.com/ConnorBaker/cuda-packages/tree/main/scripts/cuda-redist | 07:01:12 |
| connor (burnt/out) (UTC-8) | Ugh we should write an update for the post Tom made on discourse (https://discourse.nixos.org/t/community-team-updates/56458)
@someoneserge anything we should mention in particular?
I think I started a draft for an update earlier this year so I’ll see if I can find it :/ | 07:03:39 |
 SomeoneSerge (back on matrix) | In reply to @connorbaker:matrix.org
Ugh we should write an update for the post Tom made on discourse (https://discourse.nixos.org/t/community-team-updates/56458)
@someoneserge anything we should mention in particular?
I think I started a draft for an update earlier this year so I’ll see if I can find it :/ Let's make a shared pad for the draft? | 14:18:46 |
| SomeoneSerge (back on matrix) | Also maybe we've already reached the point where a room-wide voice call could be a better way to list the "challenges" | 14:41:51 |
 hexa (UTC+1) | is anyone here aware why tensorflow 2.13.0 on 24.11 now requires AVX CPU instructions and the same version on 24.05 did not? https://github.com/NixOS/nixpkgs/issues/358973 | 17:49:22 |
| hexa (UTC+1) | yes, the answer is dependencies, likely protobuf from the call trace | 17:49:39 |
| hexa (UTC+1) | * the likely answer is dependencies, should be protobuf from the call trace | 17:49:58 |
| SomeoneSerge (back on matrix) | I was about to post "you sure it's not the pypi garbage" and then thought "I'm surely just being biased" | 18:07:06 |
| SomeoneSerge (back on matrix) | HOWEVER
❯ nix path-info --derivation github:NixOS/nixpkgs/nixos-24.11#wyoming-openwakeword --recursive | grep tensorflow
[45.7 MiB DL]/nix/store/7jriah484029w7m5rnf0mjgqq07bw8s9-tensorflow_estimator-2.15.0-py2.py3-none-any.whl.drv
/nix/store/6h3v0mb1cx0grqmhizjg9a7jfn9vqfg8-python3.12-tensorflow-estimator-2.15.0.drv
/nix/store/9f7kqb69yj916mz050mjgj2v647n5qjb-tensorflow_cpu-2.18.0-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.drv
/nix/store/38hvm2cpx13vcnszrx3r5ra1ygh95vpq-python3.12-tensorflow-2.18.0.drv
| 18:07:13 |
| SomeoneSerge (back on matrix) | ^^^ tensorflow of a smoker | 18:07:53 |
| SomeoneSerge (back on matrix) | * ^^^ tensorflow of a heavy smoker | 18:08:13 |
| SomeoneSerge (back on matrix) | Tensorflow of a person only doing mild Bazel drugs:
❯ nix path-info --derivation github:NixOS/nixpkgs/nixos-24.05#wyoming-openwakeword --recursive | grep tensorflow
/nix/store/d66q5sh2fp6qdp69wwx3a2xmziar086m-tensorflow_estimator-2.11.0-py2.py3-none-any.whl.drv
/nix/store/62kz9wzp6diivxh8ab7l7gc5lgzcvxdg-python3.11-tensorflow-estimator-2.11.0.drv
/nix/store/5w169ckmwk0jfgsfrc0dv59di5smj21r-tensorflow-deps-merged.drv
/nix/store/ixj73jp45cdnx3p694gs3s9avd07fqfk-tensorflow-2.13.0-deps.tar.gz.drv
/nix/store/6sdqpx9d5hr89yz1gg8mncwhz6lm7qxp-tensorflow-2.13.0.drv
/nix/store/isjphw15ax682z95qrmsh92j108qy3p6-python3.11-tensorflow-2.13.0.drv
| 18:08:35 |
| SomeoneSerge (back on matrix) | hexa (UTC+1): | 18:08:44 |
| hexa (UTC+1) | lol | 18:09:25 |
| hexa (UTC+1) | we have tensorflow 2.18 on 24.11? 😲 | 18:09:49 |
| hexa (UTC+1) | did y'all backport the tf-bin pin? | 18:10:01 |
| hexa (UTC+1) | oh no, I moved it to tf-bin because of py312 | 18:10:37 |
| hexa (UTC+1) | is anyone else getting older by the minute? | 18:10:46 |
| hexa (UTC+1) | * is anyone else here getting older by the minute? | 18:10:51 |
| SomeoneSerge (back on matrix) | In reply to @hexa:lossy.network
is anyone else here getting older by the minute? Yes. | 18:11:03 |
| 26 Nov 2024 |
 sielicki | would be willing to attend, let me know a date | 01:44:59 |
| sielicki | really awesome stuff | 01:46:03 |
| SomeoneSerge (back on matrix) | In reply to @ss:someonex.net
Also maybe we've already reached the point where a room-wide voice call could be a better way to list the "challenges" Let's try a jitsi? Maybe just a half an hour call, one round of "I use (contribute to) or considered using (contributing to) (cuda in) nixpkgs, here's what challenges I'm facing", notes in a hedgedoc pad.
https://crab.fit/nixpkgs-cuda-status-289220
| 14:23:11 |
| connor (burnt/out) (UTC-8) | I’m crunching to try to finish something for work today and tomorrow so I might be scarce | 14:24:06 |
| SomeoneSerge (back on matrix) | In reply to @connorbaker:matrix.org
I’m crunching to try to finish something for work today and tomorrow so I might be scarce Yeah ofc, just see if anything on crabfit might work for you | 14:25:18 |
| SomeoneSerge (back on matrix) | In reply to @ss:someonex.net
Let's try a jitsi? Maybe just a half an hour call, one round of "I use (contribute to) or considered using (contributing to) (cuda in) nixpkgs, here's what challenges I'm facing", notes in a hedgedoc pad.
https://crab.fit/nixpkgs-cuda-status-289220
CC sielicki | 14:40:36 |
| 27 Nov 2024 |
| connor (burnt/out) (UTC-8) | oh hell yeah I got warp packaged https://github.com/ConnorBaker/cuda-packages/blob/main/cuda-packages/common/warp.nix | 08:30:44 |
| connor (burnt/out) (UTC-8) | If I can package OpenCV 4.10, I could conceivably stop using micromamba to manage dependencies for the rewrite of http://aggregate.org/DIT/PARSEK/ I'm working on | 08:31:30 |
| SomeoneSerge (back on matrix) | In reply to @ss:someonex.net
Let's try a jitsi? Maybe just a half an hour call, one round of "I use (contribute to) or considered using (contributing to) (cuda in) nixpkgs, here's what challenges I'm facing", notes in a hedgedoc pad.
https://crab.fit/nixpkgs-cuda-status-289220
So seems like 3PM UTC+00 today or tomorrow are ok? | 09:15:07 |
|  wrvsrx joined the room. | 12:48:50 |
