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/
uis has joined #lima
uis_ has quit [Ping timeout: 480 seconds]
<uis>
Thank you very much for your work.
<uis>
(Who knew Vasyans from opennet were this good?)))
chewitt has quit [Quit: Zzz..]
<enunes>
anarsoul: indeed thanks for all this work :)
<enunes>
anarsoul: I started looking into the optimizer one but there seem to be some debug leftovers, is that still in some wip state?
<enunes>
also I think it would be best to fold the commit which fixes the deqp tests with the CI updates rather than a separate commit
tlwoerner has joined #lima
<anarsoul[m]>
I kind of hoped to avoid running deqp for each of the commits since I have no idea what fixed these tests :)
<anarsoul>
enunes: these aren't really debug leftovers
<anarsoul>
I'll move assigning discard block index into a separate commit though
<anarsoul>
(and it doesn't have to be INT_MAX, it can be the last block + 1 index)
<anarsoul>
uis: I haven't been on opennet for a over a decade. Let's say I have conflicting political views with opennet's administration
<uis>
You replied to my post in 2020 about panfrost.
<anarsoul>
oh really? ok, then half a decade
<anarsoul>
(I just checked and it looks like opennet is blocking Canadian IPs. heh)
<anarsoul>
ppir_add_write_after_read_deps() bites me again
<anarsoul>
it doesn't account partial reg reads/writes and as a result may create a circular dependency
<anarsoul>
*doesn't account for
yuq825_ has joined #lima
yuq825 has quit [Ping timeout: 480 seconds]
Net147 has quit [Quit: Quit]
Net147 has joined #lima
<anarsoul>
oof, it's actualy a different issue :\
<anarsoul>
I'd say our comp->var_nodes[] usage is completely broken