12:34glehmann: agd5f: I got a navi 31 this week, and it's completely unusable: https://gitlab.freedesktop.org/drm/amd/-/issues/2518
12:59mech_pilot: was it also like this with rdna2 gen? on the beginning?
13:21agd5f: glehmann, never seen that before. are you using X? if so, make sure you have this patch: https://cgit.freedesktop.org/xorg/driver/xf86-video-amdgpu/commit/?id=9c959fac3af28d191105f63236096ad456dca614
13:41glehmann: agd5f: yes, I'm using X and I use 23.0 which has that commit
13:57tonyk: agd5f: I see that registers mmCP_IB1_BASE_[HI, LO] (at GFX10) can be used to set the indirect buffer to be executed, but do you know which register should I read to show the current instruction pointer of the IB being executed?
14:10glehmann: agd5f: it looks like my issue is only happening with uncomposited X. kde wayland and kde X with compositor work fine. I was previously using lxde and kde X with compositor disabled also has the same issue
15:32agd5f: tonyk, I don't recall off hand. There is a CP register you can read to dump the current command fifo, I don't recall if there is a IB rptr like register off hand. mareko?
17:04Venemo: agd5f: wouldn't the rptr have this info?
17:33agd5f: Venemo, the rptr and wptr only affect the ring AFAIK. I'm not sure if there is something similar for IBs
17:33Venemo: I see