| 21 Mar 2025 |
 K900 | No idea, I think AMD people need to figure that out | 16:32:34 |
 nyanbinary 🏳️⚧️ | doesn't look like thee issue is getting much attention tho :p | 16:37:59 |
| nyanbinary 🏳️⚧️ | sadly | 16:38:01 |
| nyanbinary 🏳️⚧️ | oh also maybe Atemu can provide more info | 16:38:37 |
| nyanbinary 🏳️⚧️ | since you made the issue? | 16:38:41 |
 Atemu | I don't know more than what's written in the issue | 16:41:10 |
| Atemu | Oh wait now that you mention it, @cyclopentane:aidoskyneen.eu did not appear to repro on her FW16 running sway two days ago. I haven't tried using sway yet but it appears to be compositor-specific in some way | 16:43:58 |
| Atemu | Go try that if you feel like debugging | 16:44:16 |
| 22 Mar 2025 |
 Sean | can someone help me debug monster hunter wilds? | 00:14:42 |
| Sean | essentially in oilwell basin it loads a bunch of particles and unless I'm on "very low" the gpu disappears and re-appears over and over again | 00:15:26 |
| Sean | which results in like 2 FPS and 800ms of delay between inputs | 00:16:03 |
| Sean | just read the stuff above. should I use mesa 24? | 00:19:34 |
| Sean | it doesn't start at all in mesa 24 | 00:43:24 |
 olivia | the gpu disappears and re-appears over and over again
what do you mean by this? | 00:56:18 |
| Sean | I mean in the proton logs I can see the vulkan device lost, over and over | 00:56:54 |
| olivia | https://gitlab.freedesktop.org/mesa/mesa/-/issues/12809 (which I think is the issue that the previous conversation was about, from skimming) seems unlikely to be related | 00:57:22 |
| olivia | ah, that makes sense | 00:57:29 |
| olivia | is there anything in the kernel logs? | 00:57:34 |
| olivia | I'm assuming this is amd? | 00:57:39 |
| Sean | oh I have this issue but yeah unrelated | 00:57:41 |
| Sean | let me check | 00:57:50 |
| olivia | * I'm assuming this is amd/radv? | 00:57:54 |
| Sean | no nothing just this before it even starts happening
Mar 21 21:01:08 nixos-desktop kernel: umip: MonsterHunterWi[79400] ip:1569dbf11 sp:8280eee8: SGDT instruction cannot be used by applications.
Mar 21 21:01:08 nixos-desktop kernel: umip: MonsterHunterWi[79400] ip:1569dbf11 sp:8280eee8: For now, expensive software emulation returns the result.
Mar 21 21:01:08 nixos-desktop kernel: umip: MonsterHunterWi[79400] ip:1569dbf11 sp:8280eee8: SGDT instruction cannot be used by applications.
Mar 21 21:01:08 nixos-desktop kernel: umip: MonsterHunterWi[79400] ip:1569dbf11 sp:8280eee8: For now, expensive software emulation returns the result.
Mar 21 21:01:10 nixos-desktop kernel: umip: MonsterHunterWi[79400] ip:7fc2c65b sp:8280fc30: SGDT instruction cannot be used by applications.
| 01:01:40 |
| olivia | that's probably unrelated | 01:02:17 |
| Sean | I got this after
Mar 21 21:01:31 nixos-desktop kernel: mce: [Hardware Error]: Machine check events logged
Mar 21 21:01:31 nixos-desktop kernel: [Hardware Error]: Corrected error, no action required.
Mar 21 21:01:31 nixos-desktop kernel: [Hardware Error]: CPU:0 (19:21:2) MC27_STATUS[Over|CE|MiscV|-|-|-|SyndV|-|-|-]: 0xd82000000002080b
Mar 21 21:01:31 nixos-desktop kernel: [Hardware Error]: IPID: 0x0001002e00000500, Syndrome: 0x000000005a020001
Mar 21 21:01:31 nixos-desktop kernel:
Mar 21 21:01:31 nixos-desktop kernel: [Hardware Error]: Power, Interrupts, etc. Ext. Error Code: 2
Mar 21 21:01:31 nixos-desktop kernel: [Hardware Error]: cache level: L3/GEN, mem/io: IO, mem-tx: GEN, part-proc: SRC (no timeout)
| 01:02:37 |
| Sean | let me try running my RAM at 2666 | 01:03:00 |
| olivia | I'm kinda surprised that proton even attempts to recover from VK_ERROR_DEVICE_LOST lol. That sounds like an enormous pain | 01:03:35 |
| olivia | oh, possible they don't have any special handling to quit the program on device loss, and so they're just getting the same error over and over again with every api call. That seems more likely | 01:04:02 |
| Sean | In reply to @olivia:computer.surgery
oh, possible they don't have any special handling to quit the program on device loss, and so they're just getting the same error over and over again with every api call. That seems more likely well it must recover somehow. since the game continues running | 01:08:16 |
| olivia | ah, yeah | 01:08:46 |
