15:42 ppw: hey, does nouveau have a "power-save" feature?
15:49 karolherbst: ppw: depends on what you mean, but generally we to turn off GPUs on laptops
15:50 ppw: karolherbst: uh, well let's say, for instance, forcing a low core and memory clock.
15:50 karolherbst: we do support reclocking for some GPUs, and sometimes explicitly setting the perf state might lead to lower voltages, but GPUs tend to boot in low power mode in avarage
15:51 karolherbst: most of power savings actually come from higher performance though and getting the GPU into idle states
16:00 swiftgeek: karolherbst: i melted some nv50 on a lappy, during windows 8 preview or sth installation xD
16:00 swiftgeek: for some reason mobile geforce 9600M GT or sth, always booted either overclocked or even more overclocked
16:01 karolherbst: yeah.. that stuff was more common for those older GPUs
16:02 karolherbst: starting with Fermi/Kepler the clocks went down a lot by default
16:02 swiftgeek: i wonder when I will get around to making MXM2 with newer gpu just for the sake of it xD
16:03 swiftgeek: would be nice to have some reference mxm3 files and just modify edge connector area
16:06 swiftgeek: oh well I would need to first figure out stuff over JTAG and memory controller to make something reliable
18:32 johns: hi all, trying to get a new (to me, but used) card working, and I'm getting this: https://paste.debian.net/1247856/ error about the PCI ROM data signature
18:32 johns: lspci output shows: https://paste.debian.net/1247853/
18:36 johns: xorg log shows nouveau being loaded
18:37 johns: but failing to open the drm device
18:45 johns: I went through the Troubleshooting -- no modesetting stuff in the kernel commandline, etc
19:06 karolherbst: mhhh
19:07 karolherbst: maybe your vbios simply has a wrong signature, but works regarldess... dunno
19:07 karolherbst: johns: this is on a desktop, right?
19:08 johns: karolherbst: yeah. currently chasing the possibility that there may be a physical bios switch on the card..
19:09 johns: no, I was confused
19:12 karolherbst: sometimes OEMs do strange things, maybe it's nouveau which is wrong... hard to tell. Does it work with the nvidia driver?
19:20 johns: karolherbst: haven't tested that yet, was hoping not to, but looks like I might have to
19:20 johns: it is a used card, so perhaps the previous user did something..
19:20 karolherbst: would be good to know if nvidia rejects that vbios as well or not...
19:21 karolherbst: maybe it's a "relabeld" GPU or something
19:21 karolherbst: dunno
19:21 karolherbst: or maybe the vbios is modded
19:21 karolherbst: who knows
19:25 johns: I wonder if I could reflash the vbios?
19:28 karolherbst: not really
19:28 karolherbst: mhhh
19:28 karolherbst: or actually...
19:28 karolherbst: if the tooling updates the checksum it should work
20:25 johns: interestingly I get that same error when I try to dump the rom
20:28 karolherbst: ahh yeah... the tools we have in envytools check that as well
20:29 karolherbst: but I think they still dump it? maybe there was a flag for it.. dunno
20:31 johns: this is just trying to dump it with cat
20:46 karolherbst: mhhh... strange
20:47 karolherbst: johns: might be able to use nvagetbios from https://github.com/envytools/envytools
20:48 johns: karolherbst: thanks, checking it out..
21:25 johns: karolherbst: well, nvagetbios started doing something and then caused a machine reboot..
21:25 karolherbst: :(
21:25 karolherbst: johns: might have to pass in "-s prom"
21:26 johns: it got as far as "Attempt to extract the vbios from card 0 (nve4) using PRAMIN." which was very exciting
21:29 johns: karolherbst: okay, great, that gave me https://paste.debian.net/1247892/ -- and a 1MB file of some variety..
21:30 johns: which apparently contains no strings
21:37 johns: 1MB seems very large, the official default bios seems to be 179k?
22:34 johns: with debug on, https://paste.debian.net/1247897/
22:45 johns: whatever I extracted with nvagetbios probably isn't anything, hexdump shows "0080000 1100 badf 1100 badf 1100 badf 1100 badf"