| 18 Apr 2024 |
 K900 | We should really start screaming at people if they run ZFS with a non-LTS kernel tbh | 09:12:43 |
 Atemu | Huh, it usually happens in autumn | 09:12:45 |
| Atemu | And remove that latestCompatibleLinuxKernel or whatever | 09:13:02 |
| Atemu | Or make it point at LTS | 09:13:14 |
| Atemu | Which is the default anyways but ig this passthru was pasted into the official ZFS docs at some point... | 09:13:44 |
| Atemu | * Which is the default anyways but this passthru was pasted into the official ZFS docs at some point... | 09:15:06 |
| Atemu | And now for something completely different: I've built myself a little module for LACT which can be used like this. Is there interest in upstreaming this module? | 09:20:23 |
| K900 | What is LACT? | 09:22:10 |
| K900 | (baby don't hurt me~) | 09:22:19 |
| Atemu | It's a utility for configuring AMDGPUs | 09:25:48 |
| Atemu | Power mode, fan speed etc. | 09:26:03 |
| Atemu | https://github.com/ilya-zlobintsev/LACT | 09:26:19 |
| Atemu | It works quite a bit better than CoreCtrl IMHO by virtue of having a proper headless daemon | 09:27:54 |
| Atemu | Oh and I think it can OC too but who OC's these days? | 09:34:02 |
| Atemu | K900: If you do stress -c 32, is your 7950x3D also up against 89°C and only at ~4.2GHz? | 09:45:58 |
| Atemu | I feel like I'm hitting a thermal limit here | 09:46:11 |
| K900 | In reply to @atemu12:matrix.org
It works quite a bit better than CoreCtrl IMHO by virtue of having a proper headless daemon Ooh interesting | 10:07:35 |
| K900 | In reply to @atemu12:matrix.org
K900: If you do stress -c 32, is your 7950x3D also up against 89°C and only at ~4.2GHz? Hold on, I don't have stress-ng, but I might have cinebench | 10:07:49 |
| K900 | So yeah CB R23 multi-core is getting me to the 89C limit | 10:18:52 |
| K900 | But I am getting better clocks | 10:18:56 |
| K900 | 4.7GHz or so on CCD0 and 4.8-4.9 on CCD1 | 10:19:04 |
| K900 | I can probably get it a bit higher with a more aggressive curve optimizer | 10:21:08 |
| K900 | I'm currenly running -10 all core | 10:21:12 |
| K900 | And I know this chip can go to like -20-25 | 10:21:46 |
| K900 | But I've seen a few weird glitches that I can't attribute to undervolting but also can't NOT attribute to undervolting | 10:22:13 |
| K900 | So I've been running it a bit more conservative | 10:22:22 |
 SigmaSquadron | Redacted or Malformed Event | 10:24:54 |
| Atemu | In reply to @k900:0upti.me
But I am getting better clocks Damn, indeed if I up the fan speed of CPU and case, it clocks quite a bit higher | 11:30:13 |
| Atemu | Already at 4.3-4.4 but my case is still rather hot from the stress test I just did | 11:30:35 |
| Atemu | In reply to @k900:0upti.me
Hold on, I don't have stress-ng, but I might have cinebench I'm using regular stress I think | 11:34:29 |
