13:53 <kusma> zmike: Seems the line-stipple lowering was less nasty than I *feared*... Let's see if you agree ;)

13:53 <zmike> uh oh

14:10 <zmike> yeah this isn't too bad at all

17:37 <anholt> wait, zink doesn't have a get_disk_shader_cache()?

17:37 <anholt> I was wondering why it felt like I was getting no shader caching in asphalt9

17:38 <zmike> it has caching but not for that

18:27 <kusma> look, we need a cache for the cache for the cache!

21:22 <anholt> zmike: tc-readahead breaks the among_us angle trace. NULL function pointer in tc_batch_execute()'s "iter += execute_func[call->call_id](pipe, call, last);"

21:39 <zmike> interesting

21:39 <zmike> I'll check it out tomorrow

21:39 <zmike> just fixed a couple minor bugs in it this afternoon before I clocked out but didn't have the energy to push

21:40 <zmike> I'm wondering if I need to add fd's buffer shadowing in order to effectively land that since the current cmdbuf reordering isn't quite good enough to fully avoid transfer ops splitting renderpasses

21:41 <zmike> passes cts, but not exactly a high bar