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/
camus1 has joined #lima
camus has quit [Ping timeout: 480 seconds]
camus has joined #lima
camus1 has quit [Ping timeout: 480 seconds]
Viciouss has quit [Remote host closed the connection]
Viciouss has joined #lima
<rellla>
are there still bugs? :)
camus1 has joined #lima
camus has quit [Ping timeout: 480 seconds]
camus has joined #lima
camus1 has quit [Remote host closed the connection]
camus1 has joined #lima
camus has quit [Ping timeout: 480 seconds]
camus has joined #lima
camus1 has quit [Ping timeout: 480 seconds]
<enunes>
rellla: I think something that can still be improved and can benefit applications is ppir. even some glmark2 benchmarks seem to be bottlenecked mostly by fragment shaders at this point
<rellla>
enunes: do you have sth special in mind?
<enunes>
rellla: I think merging instructions in general, for example something I was looking at some weeks ago was branches, we are still defaulting to creating a separate instruction just for the branch
<enunes>
another thing I noticed was fragcoord, we are inserting a bunch of code for transformations that the binary compiler simply doesn't, I wonder if we need that
<enunes>
seemed to be related to some pipe caps that set which corner is the origin, or something like that
<enunes>
I'm sure there are other places that can be improved, maybe by looking at the shader output for some popular applications
camus has quit [Remote host closed the connection]