08:57 anarsoul: so I'm working on improving combiner unit usage in ppir, but it's either I'm missing something in encoding (will check tomorrow) or there is some weird interaction with combiner and branch
08:58 anarsoul: e.g.
08:58 anarsoul: load.v $2.x 1.x, load.u 1, mul.s2 $0.y ^uniform.x $2.xxxx, branch.lt $0.x $0.y 18
08:58 anarsoul: ^^ this doesn't work
08:58 anarsoul: but
09:06 anarsoul: load.v $2.x 1.x, load.u 1, mul.s0 $0.y ^uniform.x $2.x, branch.lt $0.x $0.y 18
09:06 anarsoul: works fine
09:08 anarsoul: I suspect it's something about branch
18:37 anarsoul: down to 6 failures (4 actual failures, but one causes job timeout and kills 2 other tests)
21:20 anarsoul: darn, I hate pp select :)
23:06 anarsoul: it seems to be working and improvement for some shaders is significant
23:06 anarsoul: enunes: when you get back please take a look at https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33540