| 17 Jan 2025 |
 matthewcroughan | what is "output rate", I don't see it documented anywhere | 18:47:11 |
 K900 | The rate the actual output runs at | 18:50:24 |
| K900 | Which is pretty much always 48k unless you have it configured differently | 18:51:19 |
| 18 Jan 2025 |
|  Rachel Astkhyan 🇦🇲 ❤️🦋 changed their display name from /dev/westpain ⛟ to Rachel (Stellessia). | 16:29:10 |
| Rachel Astkhyan 🇦🇲 ❤️🦋 changed their display name from Rachel (Stellessia) to Rachel Astkhyan 🇦🇲 ❤️🦋. | 16:29:57 |
| 19 Jan 2025 |
| matthewcroughan | K900: Do you have a config for overclocking rk3588? | 03:37:22 |
| matthewcroughan | I figure if anyone has tried, it's you :D | 03:37:43 |
| K900 | No | 07:29:02 |
| K900 | You'll probably kill it | 07:29:09 |
|  Waldemar Tomme (they/them) changed their display name from Waldemar Tomme to Waldemar Tomme (they/them). | 08:17:06 |
|  @rgrunbla:matrix.org left the room. | 09:36:52 |
| matthewcroughan | I found that a task I'm trying to run, works better on cores 4-7 than on 0-3 | 13:13:15 |
| matthewcroughan | it executes in 2 seconds on 4-7 and in like 20 seconds on 0-3, so the linux scheduler isn't doing the right thing | 13:13:42 |
| matthewcroughan | If a single threaded app happens to spawn on a slow core, it has impacts | 13:13:53 |
| matthewcroughan | This was my intuition, but you told me a few days ago that my intuition was wrong, how come you thought I was wrong? | 13:14:29 |
| K900 | The scheduler will move tasks between cores | 13:25:16 |
| K900 | If you don't force them with taskset | 13:25:22 |
|  sinan changed their profile picture. | 14:02:51 |
 tpw_rules | K900: how out of date can libgbm be compared to mesa? i noticed last time you updated mesa you also put the libgbm update on staging | 19:37:54 |
| K900 | After 24.3, any | 19:50:37 |
| K900 | Before 24.3, they need to be exactly synced | 19:50:44 |
| tpw_rules | ok, that's what i gathered. just keeping our libgbm up to date for consistency? | 20:26:43 |
| tpw_rules | * ok, that's what i gathered. you are just keeping our libgbm up to date for consistency? | 20:27:03 |
| K900 | The idea is that we bump to latest libgbm every staging cycle | 20:27:09 |
| K900 | But we can bump the actual drivers on master | 20:27:22 |
| K900 | Without having to wait a cycle | 20:27:27 |
| K900 | Because bumping it on staging is effectively free | 20:27:49 |
| K900 | The actual GBM loader bits haven't meaningfully changed in years | 20:29:32 |
| K900 | But it's easier that way | 20:29:38 |
| K900 | And future work is probably getting the loader moved out of Mesa entirely | 20:30:06 |
