15:32 dviola: zmike: so, I tried getting the apitrace for this again: https://gitlab.freedesktop.org/mesa/mesa/-/work_items/15223 -- but I haven't had much luck, the resulting trace is always a 4+ GB trace with frames that can't be replayed, let alone reproduce the issue
15:32 dviola: that's with windows apitrace, with linux apitrace I get a trace with 0 frames
15:35 zmike: on linux you could try directly injecting with LD_PRELOAD=/path/to/apitrace/wrappers/glxtrace.so in the steam launch cmd
15:35 zmike: otherwise sometimes it be like that
15:36 dviola: I'll try that, thanks
15:36 dviola: it seems like the issue happens on windows too, for some users
15:37 dviola: though it's unclear why my integrated gpu can't reproduce the issue, only the 6700XT can
15:40 dviola: I noticed a previous dev added a workaround for this: https://gitlab.freedesktop.org/mesa/mesa/-/commit/1af72fa4d665b9847dff9b22d7a7dea01c0960c7
15:41 dviola: without that it just crashes
15:48 dviola: another dev*
16:20 dviola: it's stuck at the loading screen now, but the LD_PRELOAD thing works
16:21 dviola: the resulting trace file still doesn't make much sense when I glretrace -w doom.trace
16:22 dviola: just a bunch of white textures and not much else
17:31 dviola: I see a bunch of conditions that mention CHIP_RAVEN in src/gallium/drivers/radeonsi and src/amd/vulkan/ but nothing for CHIP_NAVI22
17:32 dviola: seems like the conditions are for disabling DCC in raven in some cases, etc
18:30 dviola: I think I should be able to point LD_LIBRARY_PATH to my custom mesa lib from Steam's cmd right?
18:32 karolherbst: mhh.. I'm running into a weird issue where nir_lower_vars_to_ssa is replacing variables with undefs, even though they have a non undef valued stored...