11:07 <rellla> cwabbott: are you aware of https://freenode.irclog.whitequark.org/lima/2020-06-19 ?

11:18 <rellla> i extended and prettified gpir debugs and try to understand regalloc and scheduler. besides that the issue should be the mentioned limited number of stores/loads per instruction, did you already do some investigations on it or can you give me hints wich code paths i should follow to track the issue down?

11:23 <rellla> http://imkreisrum.de/deqp/512/log is the gpir log (with my extensions) - would be cool, if you can have a look ... perhaps you can at least localize the root of all evil :)

11:26 <rellla> anarsoul, enunes: you are welcome as well :)

11:28 <rellla> btw, enunes, according to https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7825#note_710067, do you plan to look into integrating shader cache?

11:44 <enunes> rellla: yes, but I havent started it yet, I had https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3048 in my list first

11:45 <enunes> were you working on the shader cache?

11:50 <rellla> enunes: i began to read vc4 and iris code but only added some basic cache code. i stopped at the point, what to use for the key and how/where to implement it

11:50 <rellla> there is more brain work necessary first :)

11:51 <rellla> feel free to pick the task

11:59 <enunes> rellla: ok, I probably will then, I'll let you know when I go back to it

19:54 <anarsoul> rellla: I won't have much time to look into it till end of January 2021

19:55 <anarsoul> I'm pretty busy with my daytime job and personal stuff till then