00:00 zfnmxt: Does (for the high frequencies) here pertain to signal frequencies (e.g., guessing high bandwidth) or refresh rates?
00:00 imirkin: high pixclks
00:00 imirkin: above 340mhz
00:00 imirkin: the pix clock is htotal * vtotal * vrefresh
00:00 zfnmxt: Ah.
00:01 imirkin: so however you want to do it, anything above 340mhz requires hdmi 2.0 scrambling stuff
00:01 zfnmxt: So the theory is that 3840*2160*60 exceeds the threshold for that fancy hdmi2 stuff
00:01 imirkin: well, it definitely does :)
00:01 imirkin: that works out to 594mhz
00:01 zfnmxt: And 3840*2160*30 definitely does not?
00:01 zfnmxt: Ah right
00:01 imirkin: correct
00:01 zfnmxt: I see :D
00:01 imirkin: (htotal != 3840 btw ... there's hblank, etc)
00:02 zfnmxt: I don't know too much about this stuff as you can probably tell hehe
00:02 imirkin: yeah, no worries
00:02 imirkin: it's not entirely obvious either, which is why i'm pointing out some slightly subtle bits
02:01 mixfix41_: looks like getting 60fps on 460 nvidia, on nvidia binary blob. but i have a 670 and on the same ultra settings in albion online i get 20fps
02:01 mixfix41_: with nouveau
02:02 mixfix41_: i probably should get a newer radeon card mid range but tempted not to
04:26 mixfix41: hmm radeon driver is like nouveau at all?
04:26 mixfix41: i love playing on my hd6850 looks really good
04:27 mixfix41: ii guess being both open source it helps..
09:59 karolherbst: imirkin: would you mind if we would also submit our patches on gitlab directly? One can even subscribe to a label, so you get notifications every time a new MR was opened. I know that reviewing on gitlab is a bit annoying, and the only workaround I have is to open every patch in a new tab and leave the comments there as this is probably the less painful way, but 1. we get the CI testing and 2. we don't have random people telling
09:59 karolherbst: others to submit on gitlab
10:56 imirkin: you can do whatever you like
10:58 imirkin: mixfix41: if you reclock, should get higher framerates. those GPUs tend to boot to lowest speeds.
10:58 imirkin: (this should be possible with the 670 but not the 460)
11:06 karolherbst: imirkin: I also meant more as a general rule if others ask, but if it's fine for you both ways then I don't mind. I just don't want to start telling people to use gitlab, if there was no discussion on that before
11:06 imirkin: i mean, i don't plan on reviewing stuff in gitlab
11:06 imirkin: but people can do what they like
11:07 karolherbst: yeah well.. it would be kind of pointless for people to send out nouveau patches if you won't review them ;)
11:07 karolherbst: I mean, I could do that as well, but you know, generally
11:07 imirkin: which is why i recommend emailing them
11:07 karolherbst: okay
11:08 karolherbst: I am mainly asking because somebody replied to marks patches telling him to open an MR
11:08 imirkin: yeah
11:08 imirkin: i saw
11:08 imirkin: people say all kinds of things
11:09 imirkin: https://xkcd.com/386/
11:09 karolherbst: maybe we should tell devs to post to the nouveau list instead
11:09 karolherbst: would probably get rid of those comments
11:09 imirkin: that's fine
11:09 karolherbst: k
11:09 imirkin: i don't really get why you're so worried about those comments though
11:09 karolherbst: confuses contributors
11:10 karolherbst: and we would have to reply to clear that up
20:03 edgecase: more of a general question, but could the vulkan approach ... do all the details yourself... be used on
20:03 edgecase: older chips (not just radeon pre GCN) but with missing features, sort of like GLES is a subset of GL?
20:04 imirkin: in theory? sure. with vulkan? no.
20:04 imirkin: vulkan requires, essentially, for the GPU to have compute and a mmu
20:04 imirkin: one could design a different API which didn't require these things
20:10 edgecase: mmu I understand, what do you mean by compute
20:12 imirkin: compute shaders
20:42 edgecase: are they turing complete, loops/conditionals and what not, is that where they draw the line, so to speak?
20:43 imirkin: they are turing-complete
20:43 imirkin: but they're not used for rasterization