| 15 Oct 2023 |
 @atemu12:matrix.org | Is your CPU still clocking as high as it should? | 12:46:19 |
 Charlotte 🦝 (it/rac/racs/racself/🦝/plush) | In reply to @atemu12:matrix.org
Charlotte 🦝 (it/rac/racs/racself/🦝/plush): Well, do the clocks go down? Clocks can be queried via sysfs and there are a bunch of tools to display them in fancy GUIs too. i forgot (it's been a while since i investigated this) but i remember the clocks being lower, but not that much | 12:47:35 |
| @atemu12:matrix.org | If it's just a few %, it doesn't mattere | 12:47:59 |
| @atemu12:matrix.org | * If it's just a few %, it doesn't matter | 12:48:01 |
| @atemu12:matrix.org | Note that even if the clocks were super low, that still might not matter | 12:48:11 |
| Charlotte 🦝 (it/rac/racs/racself/🦝/plush) | In reply to @atemu12:matrix.org
Is your CPU still clocking as high as it should? yeah cpu performance was unchanged and cpu was not fully utilized on any core | 12:48:26 |
| @atemu12:matrix.org | If GPU util is very low, it'll clock itself down to save power | 12:48:36 |
| Charlotte 🦝 (it/rac/racs/racself/🦝/plush) | In reply to @atemu12:matrix.org
Note that even if the clocks were super low, that still might not matter yeah due to power management, it only clocks up if there's work to do | 12:48:45 |
| @atemu12:matrix.org | Exactly | 12:48:52 |
| Charlotte 🦝 (it/rac/racs/racself/🦝/plush) | but yeah the friend that i asked previously pointed at power management settings but they either seem to be read only or not affect it much | 12:49:29 |
| @atemu12:matrix.org | In reply to @lotte:chir.rs
yeah cpu performance was unchanged and cpu was not fully utilized on any core CPU util is hard to measure correctly. I'd trust a measurement of it using one core but I wouldn't trust a measurement showing it's not. | 12:50:05 |
| Charlotte 🦝 (it/rac/racs/racself/🦝/plush) | anyways i was just asking if anyone else encountered/solved this, we don't need to debug it, I'm okay rebooting before gaming | 12:50:30 |
| @atemu12:matrix.org | I'm experiencing a "fun" bug where setting fan speed via corectrl/sysfs makes the GPU do weird magic on suspend and on resume I get 575MHz CPU clocks | 12:51:46 |
| @atemu12:matrix.org | That's why I was asking whether your CPU clocks are still fine | 12:52:19 |
|  @hydridity:matrix.org joined the room. | 13:26:50 |
| @hydridity:matrix.org changed their profile picture. | 13:27:18 |
 @Lownin:matrix.org | This is weird.
I'm running NixOS 23.05, running KDE with xorg, and I have an Nvidia 3070 TI.
As of a few days ago, if I have only one monitor plugged in, and enable "Force Full Composition Pipeline" when I try to start a 3D game, the game appears to lock up as soon as it starts. Tested a few so far. If I alt+tab away from the game, after ~10 seconds of no feedback, the game will start updating when it no longer has focus. As soon as I click back into it, it locks up again and the cycle repeats.
This all goes away when I have a 2nd monitor plugged in and force full composition pipeline on either the main screen or both screens. Additionally, if I try to enable force full composition pipeline in my configuration.nix, KDE seems to behave perhaps the same way--when kwin starts, I get just a black screen.
Anyone have any ideas? | 16:07:09 |
| @atemu12:matrix.org | Buy an AMDGPU? | 16:17:35 |
| @Lownin:matrix.org | Not helpful. | 16:17:49 |
| @atemu12:matrix.org | I know but we really can't do anythish about Nvidia bugs rn | 16:18:14 |
| @atemu12:matrix.org | * I know but we really can't do anything about Nvidia bugs rn | 16:18:22 |
| @Lownin:matrix.org | * This is weird.
I'm running NixOS 23.05, running KDE with xorg, and I have an Nvidia 3070 TI.
As of a few days ago, if I have only one monitor plugged in, and enable "Force Full Composition Pipeline" when I try to start a 3D game, the game appears to lock up as soon as it starts. Tested a few so far. If I alt+tab away from the game, after ~10 seconds of no feedback, the game will start updating when it no longer has focus. As soon as I click back into it, it locks up again and the cycle repeats.
This all goes away when I have a 2nd monitor plugged in and force full composition pipeline on either the main screen or both screens. ~Additionally, if I try to enable force full composition pipeline in my configuration.nix, KDE seems to behave perhaps the same way--when kwin starts, I get just a black screen.~ This part seems to not be happening anymore. The configuration is set correctly, but the same problem occurs as above like then I set it manually.
Anyone have any ideas? | 16:18:58 |
| @Lownin:matrix.org | * This is weird.
I'm running NixOS 23.05, running KDE with xorg, and I have an Nvidia 3070 TI.
As of a few days ago, if I have only one monitor plugged in, and enable "Force Full Composition Pipeline" when I try to start a 3D game, the game appears to lock up as soon as it starts. Tested a few so far. If I alt+tab away from the game, after ~10 seconds of no feedback, the game will start updating when it no longer has focus. As soon as I click back into it, it locks up again and the cycle repeats.
This all goes away when I have a 2nd monitor plugged in and force full composition pipeline on either the main screen or both screens. ~~Additionally, if I try to enable force full composition pipeline in my configuration.nix, KDE seems to behave perhaps the same way--when kwin starts, I get just a black screen.~~ This part seems to not be happening anymore. The configuration is set correctly, but the same problem occurs as above like then I set it manually.
Anyone have any ideas? | 16:19:15 |
| @Lownin:matrix.org | * This is weird.
I'm running NixOS 23.05, running KDE with xorg, and I have an Nvidia 3070 TI.
As of a few days ago, if I have only one monitor plugged in, and enable "Force Full Composition Pipeline" when I try to start a 3D game, the game appears to lock up as soon as it starts. Tested a few so far. If I alt+tab away from the game, after ~10 seconds of no feedback, the game will start updating when it no longer has focus. As soon as I click back into it, it locks up again and the cycle repeats.
This all goes away when I have a 2nd monitor plugged in and force full composition pipeline on either the main screen or both screens.
Anyone have any ideas? | 16:20:10 |
| @Lownin:matrix.org | Is this a known bug? I looked through nvidia and nix issue trackers and didn't find it.
I also don't think the nvidia driver updated since it was working? Maybe something else, like KDE or the kernel, but I don't have a Nix generation old enough to start checking. | 16:21:31 |
 K900 | The usual answer for Nvidia bugs is "complain to Nvidia" | 16:21:32 |
 @woobilicious:matrix.org | Complaining to AMD, despite it being open source isn't much better tbh. | 16:47:35 |
| K900 | I've had nothing but great experiences with AMD kernel people | 16:49:17 |
| K900 | And Mesa people in general | 16:49:24 |
| @woobilicious:matrix.org | this was the drm bug-report area.
I've been ignored for 6 months, had my issue closed due to a related bug that wasn't actually fixed in said patch, and bug report for that issue had obvious signs it wouldn't fix it because the bisect dates made no sense with the time of first reports.
And because DisplayPort spec is closed, I don't really know how to create a work around. | 16:55:59 |
