02:14imirkin: anyone have a fermi plugged in that can do a slight amount of testing for me? (would require ability to build a patched mesa tree)
02:23gnarface: which one is fermi again?
02:26gnarface: oh, i do have one of those but nothing to plug it into that has the right power hookups
02:26gnarface: except what i'm using right now
02:26gnarface: those side connectors are annoying
02:27imirkin: that's ok - no worries
02:29gnarface: well, i assume someone who knows more about the driver would be more helpful in the effort anyway but if you are still having problems finding someone in a couple days let me know
02:31gnarface: (i have another machine here i had planned on putting it into, but reluctance to saw holes in the case so far has prevented me from doing so)
02:34imirkin: fwiw i have loads of GPUs, a fermi included
02:34imirkin: i was mostly hoping to avoid having to plug one in myself
02:34imirkin: (i've even managed to somehow acquire 2... both GF108's ... oh well)
02:36gnarface: i really liked this card actually
02:36gnarface: but having only 1GB of ram became a liability for certain things...
02:36imirkin: but then time happened? :)
02:36gnarface: well i mean i still like it
02:37gnarface: i wish i could keep it and just upgrade the ram
02:39imirkin: too bad it's soldered on
02:40gnarface: video cards used to be upgradeable
02:40gnarface: why did they stop doing that?
02:40gnarface: empty slots cost that much, really?
02:41gnarface: i've owned, i think, 3 video cards in the past that had spare upgrade slots for extra ram
02:42gnarface: two of them technically they were sockets, not slots...
02:42gnarface: but you get my point
02:45imirkin: i don't remember a single upgradable video card
02:45gnarface: well we're talking pre-PCI-E
02:45imirkin: well like the cirrus logic type cards
02:46imirkin: had chips you could remove
02:46gnarface: yea, PCI and early AGP
02:46imirkin: but they were definitely not meant for user upgrades
02:46gnarface: there were upgradeable ones
02:46gnarface: in the < 8MB era
02:46imirkin: s3 virge was definitely not
02:46gnarface: you're sure?
02:46imirkin: i'm just trying to walk down the boards i've had
02:46imirkin: ati 7000 wasn't
02:46imirkin: that was my first agp boards
02:47gnarface: e they were S3 and Matrox
02:47imirkin: i even ran into a 3dlabs board - also not user upgradeable
02:47gnarface: S3 and Matrox at least those two brands for sure
02:48imirkin: find it on google image search :)
02:49gnarface: search for matrox g300
02:49gnarface: some of the pictures that come up have it, some don't
02:49gnarface: here's one but it's a g200
02:50gnarface: they were good cards, really
02:50gnarface: good linux support
02:51imirkin: hm, i guess those matrox boards really did have a separate slot
02:51imirkin: i never had one
02:51imirkin: so-dimm presumably
02:51gnarface: the S3 virge that i'm thinking of might have been actually a Virge DX ?
02:52gnarface: was that a separate thing or were they all "DX" ?
02:53gnarface: and way back in the day i had one i think had the same type of sockets as the S3 that was called uh... Paradise Pipeline PCI?
02:54imirkin: dunno. either way hopefully it's been sufficiently demonstrated that most video boards are *not* user upgradeable
02:54imirkin: even if a handful were
02:54gnarface: i might be remembering wrong, that one may not have had sockets. i might be confusing it for the ati mach64 i got from some old mac
02:55gnarface: anyway, for a while i had the impression it was a common trend
02:59imirkin: 3dfx boards didn't have it either
02:59imirkin: dunno man
02:59imirkin: i think more the exception than the rule
02:59gnarface: it wouldn't be the only strange bubble i've existed in
02:59gnarface: anyway, i'm for it
03:25pl44c: For note #4 on here https://nouveau.freedesktop.org/wiki/VideoAcceleration/ would anyone happen to have an explanation on why the way trailers are encoded would be different than other videos?
03:25imirkin: it was just an observation
03:25imirkin: nothing specific to trailers
03:26pl44c: so should that note be removed with more information with the type of h264 encoding it was lol
03:46imirkin: pl44c: i couldn't identify the cause
12:09karolherbst: orbea: I just pushed the nouveau fix for nine
15:13orbea: karolherbst: nice, though honestly I am using amdgpu now... :)
15:46imirkin: karolherbst: i don't suppose you have a fermi plugged in?
15:51orbea: karolherbst: i suspect it fixed this one too https://github.com/iXit/Mesa-3D/issues/324
15:51karolherbst: imirkin: nope, I don't
15:51karolherbst: imirkin: I don't have direct access to a fermi afaik
15:51imirkin: ok no worries
15:52karolherbst: orbea: will check the trace
16:20orbea: i guess the op said its a different issue, but your fix may of helped that game in other ways
16:20orbea: might be a nouveau issue, iiv3 said he could not reproduce it with r600
16:29karolherbst: orbea: replaying the trace causes crashes here
16:33orbea: i'm going to have to really update my wine later and see how it is with radeonsi
16:34orbea: btw, do I need to actually remove the amd card and plug the nvidia back in to switch? Might be useful for testing...
16:34karolherbst: don't think so
16:36imirkin: assuming you can physically get everything in there, no problem with switching back and forth between GPUs
16:36imirkin: esp something like prime makes it all easier
16:36orbea: there is plenty of room in the chasis at least
16:36orbea: but I'll have to look into prime now :)
16:37imirkin: well, the standard use-case is you have an iGPU + dGPU
16:38orbea: at least being able to switch with a reboot would be better, its hard to get the case out to swtich hardware...
16:39imirkin: however there's nothing technically different between an igpu and dgpu... it's all the same as far as the software is concerned
16:39karolherbst: orbea: it crashes inside gallium nine, but wine is totally painful to debug, so I won't do that
16:39orbea: yea, understandable
16:40orbea: i bisected a wine bug a while ago and then got told to bisect which dll is causing the issue, still procrastinating...
16:43karolherbst: orbea: mhh, seems like that bug was about some further issues
16:43karolherbst: and without the sampler fix it was worse
16:43orbea: i misread the first time
16:44karolherbst: mhh, if I could only replay that trace :D
16:44karolherbst: (or just download the game I guess
16:45orbea: if its crashing in nine, then maybe you could get the guys in #d3d9 to debug that :P
16:45orbea: and are you using the right apitrace?
16:46karolherbst: ... so you can't replay a mingw trace with the msvc one?
16:46orbea: I'm not sure on the details, but I've been told to use the msvc version
16:48orbea: but I think that would be a different error actually
16:48karolherbst: mhh, crashes with the msvc one
16:48karolherbst: and winedbg is as useless as ever
16:53karolherbst: I just download the game
16:56orbea: it works well in wine without nine and nouveau at least
16:57orbea: there was a xorg issue blocking nine games from starting back then so I dont think I ever ran it with nine
16:58karolherbst: I see
16:58karolherbst: last time I looked into that, the game ran better through wine
16:58karolherbst: but that was shortly after the linux release
16:58karolherbst: might be different now
17:22imirkin: karolherbst: do you have a kepler plugged in?
17:22imirkin: [or anyone else who can test a mesa patch + run a deqp test]
17:51karolherbst: mhhh, I have, but I'd prefer not to touch that laptop anymore, but yeah, I could run something sooner or later
17:55imirkin: ok, well it's not super-important