#asahi-gpu · 2022-09-16 17:37
bluetailthere but not in the same order. Increase frequency => flicker cause bandwidth is not high enough. Running it at a lower frequency seems to degrade the image quite a bit. But what can I say, these are old TN panels and I'm sensitive to change...#asahi-gpu · 2022-09-16 17:37
bluetaildo. (Difference being macintosh system would flicker with something from the framebuffer / cache or whatever that is). Anyhow, then forcing the GPU to use the second higher value in idle ~ 300mhz fixed the whole issue. However on my laptop, through a USB c to HDMI adapter, I can feel it's off cause bandwidth doesnt seem to fit. And similar thing#asahi-gpu · 2022-09-16 17:37
bluetail I had this talk before, but back then I didnt quite knew what was going on. Until I replicated it on a system that does not have the limitations of a macintosh system. And then, the other day I found myself with the solution. As I reduced the blanking, my GPU clocked down to 98Hz and it would flicker similarly but not same a macintosh system would#asahi-gpu · 2022-09-16 17:29
bluetail(reduced blanking) down. I still feel weird when I use my M1 mac mini no matter if its macOS or the Asahi Linux implementation. Feels as if Pixelclock is not quite high enough. I could not possibly capture it with a capture card. It's not quite the same but similar as if you would use a 240hz monitor on a 60hz setting. But what it doesn't do#asahi-gpu · 2022-09-16 17:29
bluetailfortunately, is flicker like it would on macOS...#asahi-gpu · 2022-09-16 17:29
bluetailalyssa modelines had helped me with random flicker on quite a few macintosh systems system I had. It usually is that the 'usual' gpu's memclock sit at around ~ 100MHz-300MHz at their idle clock and when you have a Monitor thats > 60Hz then you get flicker cause the bandwidth is no more given. More Pixelclock usually drives gpu memclock up and less#asahi-gpu · 2022-09-16 16:14
alyssayeah...#asahi-gpu · 2022-09-16 16:03
jannaumore like 70GB if you want to be able to upgrade mac os#asahi-gpu · 2022-09-16 16:01
alyssaMaybe I did want the 512gb model after all, if a macos partition with nothing on it is 40gb...#asahi-gpu · 2022-09-16 16:01
alyssaI don't have that much storage, though#asahi-gpu · 2022-09-16 16:01
alyssaah, sure.#asahi-gpu · 2022-09-16 15:29
jannaufirmware upgrades. you could keep your current setup and add new mac os 12.3 installation. the boot picker is annoyingly slow but works#asahi-gpu · 2022-09-16 15:01
alyssato this day I chainload over USB :p#asahi-gpu · 2022-09-16 15:01
alyssalinux kernel upgrades?#asahi-gpu · 2022-09-16 15:01
alyssawhich upgrades?#asahi-gpu · 2022-09-16 14:01
jannaudo you have enough storage on the device? you could install the upgrades as new mac os/linux installations#asahi-gpu · 2022-09-16 13:57
alyssanod#asahi-gpu · 2022-09-16 13:57
jannauI haven't yet looked into dcp on m2 either#asahi-gpu · 2022-09-16 13:56
jannauI can related, I finally switched to the macbook pro 14" as main device and testing on it is now annoying#asahi-gpu · 2022-09-16 13:54
alyssaand if I get an M2 device I'll get stressed about the GPU#asahi-gpu · 2022-09-16 13:54
alyssai really need to uprev but with only a single m1 device it's too painful#asahi-gpu · 2022-09-16 13:53
alyssajannau: :+1:#asahi-gpu · 2022-09-16 13:53
jannauresolution / refresh rate changing still works#asahi-gpu · 2022-09-16 13:49
alyssa(I.e. the Displays settings for GNOME, or your sway config)#asahi-gpu · 2022-09-16 13:49
alyssaswitching resolutions with DRM works, or at least worked when I first spun up DCP#asahi-gpu · 2022-09-16 13:49
alyssawhy do you want modelines?#asahi-gpu · 2022-09-16 13:38
jannaulooking into it is not a priority for the initial dcp/display driver#asahi-gpu · 2022-09-16 13:37
jannauthere is probably an interface to use custom modes but if I saw something related I forgot#asahi-gpu · 2022-09-16 13:36
jannaumaybe never. we will communicate with the dcp firmware which offers a number of modes for the display#asahi-gpu · 2022-09-16 13:30
bluetailFun fact - I still use them even in 2022. Specifically I had to use it to reduce blanking.. on a x86 system in order to make my gpu not run fullspeed memclock, but able to clock down...#asahi-gpu · 2022-09-16 13:28
bluetailAt what stage of development do we have access to XFree86 Modeline? You know, the thing you need to use a custom resolution.#asahi-gpu · 2022-09-15 21:11
alyssaHahaha#asahi-gpu · 2022-09-15 21:11
robherGoogle Summer of Complaining. Sign me up.#asahi-gpu · 2022-09-15 21:09
alyssalooks like the project was KUnit, maybe I'm mixing up my memories about floats#asahi-gpu · 2022-09-15 21:08
alyssaboth good things#asahi-gpu · 2022-09-15 21:08
alyssaeither way#asahi-gpu · 2022-09-15 21:08
alyssaor just complaining about them#asahi-gpu · 2022-09-15 21:07
alyssarobher: IIRC there's a GSoC project right now cleaning up amdgpu's floating points#asahi-gpu · 2022-09-15 19:40
phirereally the linux kernel should have a softfloat implementation for in-kernel use#asahi-gpu · 2022-09-15 19:30
Sobek[m]I think rust has soft floats lying around too, it might be possible to see with rust on Linux if there’s a way to get that enabled.#asahi-gpu · 2022-09-15 18:06
linaI think that makes sense if it's just going into sysfs, yeah...#asahi-gpu · 2022-09-15 18:06
linaSo it does support arbitrary floats, and even arbitrary invalid strings, since it doesn't try to parse them ^^#asahi-gpu · 2022-09-15 18:05
linaSee iio_show_mount_matrix()#asahi-gpu · 2022-09-15 18:05
linaLooks like that just gets exposed via sysfs, the kernel never parses it?#asahi-gpu · 2022-09-15 17:56
robherBTW, 'mount-matrix' is the case of float strings, but it looks like all the users don't support fractions so far.#asahi-gpu · 2022-09-15 17:55
robherYeah, that would probably be more consistent and avoid dealing with '/bits/ 64 <1.0f>'#asahi-gpu · 2022-09-15 17:53
linaI guess since /bits/ is already a thing, float vs. double would have to be specified that way anyway, so if it isn't an explicit keyword then anything with a . would be parsed as a float of whatever the current bitness#asahi-gpu · 2022-09-15 17:52
robherC-style is fine with me, but that's a little detail David will care about.#asahi-gpu · 2022-09-15 17:48
linaCould also be a keyword prefix in the style of `/bits/ 64`, like `/float/ 32 <1.0>` or just `/f32/ <1.0>`#asahi-gpu · 2022-09-15 17:47
linaWhat syntax would you prefer? C-style, like 1.0f? (Trailing f means f32, I don't need f64 but C style would be decimal without the f for that)