07:36kreyren: What prevents us from using nouveau on newer GPUs?
07:36kreyren: Just the power management?
07:38airlied: signed firmware is usually the last missing piece for enabling newer gpus
07:39airlied: I think still waiting on ampere signed fw
12:45karolherbst: imirkin: https://lichess.org/training/6c42j :3
14:31imirkin: karolherbst: yeah, it's a nice one. had to calculate that one out first to make sure the pawn could run in time.
14:41karolherbst: :)
14:41karolherbst: the first move was quite obvious to mee, but I had to think a little how to continue :D
14:42imirkin: right
15:17imirkin: karolherbst: https://lichess.org/1U6hsydcnUXG -- the kind of high quality game you could find yourself playing at my (shitty) level :)
15:42karolherbst: ... :D
16:53Lyude: imirkin: I never got to doing the work for maxwell, there was a regression at one point that changed where we were loading the clockgate packsg that also needs to be looked at
16:53Lyude: there should still be a wip maxwell branch on my github
16:55imirkin: Lyude: but sounds like kepler was in
16:55imirkin: s/was/is
16:55Lyude: imirkin: yeah it is - but we do need to verify that it works as the last time I checked the power consumption seemed to be higher then when I first tested the kepler work on some cards
16:55imirkin: lol
16:55imirkin: Lyude: also did you cover GK208's in your testing?
16:56Lyude: it -did- work as expected at one point, I definitely know that
16:56imirkin: those are the low-power chips of the gen
16:56Lyude: imirkin: I'm not sure - I may have but I don't remember which chips I tested specifically
16:56imirkin: k
16:56Lyude: if you have any questions let me know though
16:56Lyude: *other questions