ChanServ changed the topic of #lima to: Development channel for open source lima driver for ARM Mali4** GPUs - Kernel driver has landed in mainline, userspace driver is part of mesa - Logs at https://oftc.irclog.whitequark.org/lima/
<anarsoul|2>
enunes: what do you think about replacing ppir_comp->var_nodes with proper hash tables for SSAs and regs?
<anarsoul|2>
with a fixed number of var_nodes (which is passed from nir) it's not possible to transform ppir
yuq825 has joined #lima
<enunes>
anarsoul|2: I guess it makes sense, this whole var_nodes stuff and register index handling has always been a mess
<enunes>
for now I think it is important to first do what we need for this nir_legacy handling for ppir and reg load/store for gpir to unblock the entire nir register rework
<enunes>
then I guess after that it is better to do our own backend reworks
yuq825 has quit []
<anarsoul|2>
sounds good
camus has quit [Remote host closed the connection]
camus has joined #lima
<anarsoul|2>
and we also need to improve printing of ppir representation. It's not really readable
<enunes>
slowly getting memories of ppir back again
<anarsoul|2>
:)
<anarsoul|2>
yeah, too many corner cases
<anarsoul|2>
I don't really remember why we don't keep successors from a different block in the list