08:22tzimmermann: skeggsb, could you please review the nouveau patches in https://patchwork.freedesktop.org/series/71871/ ?
08:24skeggsb: tzimmermann: they look fine, i'll do an explicit R-b on-list though
08:24skeggsb: sorry, i did look at them already but didn't reply...
08:25tzimmermann: skeggsb, not much is left to do for this patchset. so thanks for your review
08:26skeggsb: thanks for that cleanup, it always seemed a bit dumb to have to do that lookup
10:08ric96: karolherbst: does nouveau.NvFanless=true still hold true ?
10:08ric96: Gonna rewire fans on my gtx960 for it, 100% rpm always.
10:16karolherbst: ric96: the patches were never upstream
10:18ric96: ... sooo no relocking at all for now?
10:19HdkR: :%s/ for now//
10:20ric96:proceeds to take out a shiny new gt710
10:20HdkR: Oh hey, I have one of those on a shelf
10:20ric96: Every one does, if you are in this channel, kinda mandatory
10:21HdkR: Literally the worst GPU I have :P
10:22ric96: I've got a hd4560 i think... Talking about the worst.
10:22ric96: karolherbst: do you maintain a tree somewhere whith those patches?
11:05RSpliet: ric96: I most certainly do not have a gt71... *peers at the shelf* wait, where did that come from?
13:05karolherbst: ric96: not really
17:18ric96: ok so this might seem far fetched. I was wondering if its possible to boost the lowest/base clock to a higher value by doing some gpu bios mods?
17:18ric96: This is for maxwell btw
17:19imirkin_: bios is signed iirc
17:19karolherbst: imirkin_: I think only runtime uploaded vbios' have to be signed, but flashed ones don't...
17:19karolherbst: or something like this
17:20ric96: Yeah flashed ones don't have to be signed. Huge custom bios community because of that
17:21ric96: Peeps figuring out turned off cuda corea and re enabling them for performance and stuff like that
17:21karolherbst: ric96: the biggest issue with the patches is that we need the PMU for secboot and there is tons of reworks going on in this area
17:21karolherbst: it's all quite annoying to get everything working correctly
17:22karolherbst: also.. I still need to fix our power sensor code :(
17:39ric96: So did some search. Turns out you can modify clock states using modded bios, albeit using windows. https://www.techpowerup.com/download/maxwell-ii-bios-tweaker/
17:49karolherbst: ric96: that's something else
17:49karolherbst: you can only moify runtime perf states, not the initial booting state
17:51ric96: So the p0-p8 states are post boot. And there is a separate boot state?
17:53karolherbst: there are fixed register values in the initialization code within the vbios
17:53karolherbst: but that's too advances for the bios modding community
18:00ric96:proceeds to google hex editing vbios
18:01ric96:doesn't find much
22:33Lyude: skeggsb: not something I'm planning on working on right away but I'm curious if you'd know if it's possible for us to reset the display engine on nvidia hw while not resetting anything else?
22:33imirkin_: 0x200 probably?
22:34Lyude: imirkin_: think that works on volta+ as well?
22:35imirkin_: as much as it does on other gens :)
22:35imirkin_: "reset" is a strong word though
22:35Lyude: (being able to do resets like this would be nice for stuff like igt)
22:35imirkin_: just flipping the bit off then on may not be a total reset
22:35Lyude: imirkin_: basically just anything that gets display channels unhung
22:35imirkin_: yeah, like i said, dunno
22:35Lyude: if other things break we tried our best
22:37Lyude: also - if anyone's wondering on the crc stuff I am still working on it, got sidetracked by a lot of work stuff and now it's ready to submit... except I found an issue rerunning some of the igt tests I wrote for this on turing :s
22:39Lyude: it is probably something silly though
22:39Lyude: at least I know the tests work :)
22:48skeggsb: Lyude: not... safely
22:48skeggsb: pulling it down with NV_PMC_ENABLE tends to result in it never coming back again