| 3 Jan 2024 |
 @atemu12:matrix.org | crushing-smite: It's recording the physical screen and passing that recording through shared memory AFAIUI | 10:50:43 |
 K900 | It doesn't exist until there's a physical plug because the GPU doesn't see any displays connected to it | 10:50:46 |
| K900 | So it's just not enabling graphical output at all | 10:50:55 |
 crushing-smite | :( | 10:51:05 |
| crushing-smite | what a waste of non-wasted electricity | 10:51:14 |
 philipp | It's not that bad if you just use on of these fake display plugs. | 10:51:37 |
| K900 | It's a negligible amount of electricity if you just use a dummy plug or your existing monitor | 10:52:02 |
| K900 | The GPU usually can't tell if it's the active input on your monitor | 10:52:10 |
| K900 | So it'll still render even if you have the monitor set to not display it | 10:52:20 |
| crushing-smite | yeah, but I mean, what a waste of electricity as in GPU cycles | 10:52:51 |
| K900 | GPU cycles aren't really wasted | 10:53:05 |
| crushing-smite | could have still rendered stuff even if it knows there's no output 🤔 | 10:53:09 |
| K900 | Well except the scanout hardware | 10:53:15 |
| K900 | Which is basically idle anyway | 10:53:20 |
| K900 | Because the display is what's driving the scanout | 10:53:24 |
| @atemu12:matrix.org | It could, theoretically. Windows drivers just simply don't do that. | 10:53:38 |
| K900 | As in, the display will ask the GPU to feed it frames | 10:53:40 |
| crushing-smite | aha, so the trigger for GPU's action is actually the display? | 10:53:43 |
| K900 | The GPU will still render frames | 10:53:52 |
| crushing-smite | interesting | 10:53:53 |
| K900 | Even if not asked | 10:53:55 |
| crushing-smite | what a rabbit hole this whole thing is | 10:54:18 |
| K900 | But moving the data to the display is initiated by the display | 10:54:18 |
| crushing-smite | all just to play some games | 10:54:24 |
| K900 | (usually) | 10:54:27 |
| K900 | (with VRR it's kind of inverted but also not really) | 10:54:37 |
| crushing-smite | it also now appears, that regardless of using looking glass, I should first set GPU passthrough via the conventional means | 10:55:59 |
| crushing-smite | And also tweak the KVM for optimized performance | 10:56:11 |
| crushing-smite | Any tips/mans/blogs on that part? | 10:56:24 |
| K900 | Yes | 10:56:24 |
