01:33 luc: hi, i wonder how serious the perf on intel igpu (e.g. Arc 140T integrated in Ultra 9 285H processor) would be affected on a system in a single-channel memory configuration? i got glxgears 4000+ fps on the latest mesa iris driver. i thought it would perform better. https://paste.rs/FY2qC
12:59 karolherbst: Soo.. when radv exports memory that has the RADEON_FLAG_GL2_BYPASS flag, does it also need to get imported with that flag? Wondering what's the high level requirement there
16:06 ted_: https://drive.google.com/file/d/1XBBCz9mpykdPCkSk4dslIyg0J6pkjYHq/view?usp=sharing
16:07 ted_: Issue I encountered with Mesa - this write-up is from an AI assisted debugging run.
16:32 ted_: Let me know if this helpful.
16:44 pinchartl: ted_: it likely won't
16:45 ted_: For me it addressed not being able to open Chome in Kubuntu running on Wayland.
16:45 ted_: Kubuntu stable - did not try bleeding edge.
16:45 ted_: GPU acceleration was not working until this fix.
16:46 ted_: I have a patch zip I will share as well.
16:47 pinchartl: I don't think a link to a google drive is what overworked maintainers need
17:07 ted_: I was unable to put a PR in, I have requested permission to do so following the documented process. PR will be put through when that access is granted.
17:14 pixelcluster: note mesa's policy on "AI"-generated issues/PRs: https://docs.mesa3d.org/submittingpatches.html
17:20 zmike: your best bet is having your AI generate a simplified repro case and filing a ticket
17:20 zmike: i.e., a test
17:37 soreau: does anyone know where software drivers like llvmpipe get the value for GL_MAX_TEXTURE_SIZE ?
17:38 soreau: is it hard coded or does it actually have some algorithm behind it?
17:40 llyyr: grep LP_MAX_TEXTURE_SIZE
17:40 llyyr: comment explains why it is what it is
17:41 soreau: llyyr: that does explain it, thanks