#asahi-dev · 2026-02-13 10:26
oliver_Wifi on my m1 works perfectly fine with lid open and closed, before, during and after. #asahi-dev · 2025-12-20 12:21
oliver_`eDP display without DP-HDMI conversion`#asahi-dev · 2025-12-20 12:20
oliver_according to the comment, it shouldnt be#asahi-dev · 2025-12-20 12:20
oliver_thank you, then i'll adjust it again#asahi-dev · 2025-12-20 12:20
oliver_got it#asahi-dev · 2025-12-20 12:19
oliver_so you would really prefer the resolution check#asahi-dev · 2025-12-20 12:18
oliver_ah i see#asahi-dev · 2025-12-20 12:18
oliver_iff main_display is already true for the internal display, that seems cleaner?#asahi-dev · 2025-12-20 12:16
oliver_jannau: regarding your comment on setting the timestamps only for the internal display: i am now checking for dcp->main_display, is valid too? #asahi-dev · 2025-12-19 06:01
oliver_Sven: thank you. Toying with the idea was nice, but my parental leave ends this year and I am going to spend the time with the family #asahi-dev · 2025-12-18 22:05
oliver_Only just briefly, I couldnt quickly make sense of it, and i didn't need it to get to 120hz. But I can check again #asahi-dev · 2025-12-18 21:52
oliver_jannau: i've updated the pr#asahi-dev · 2025-12-18 21:13
oliver_dcp_swap_submit is already called only when needed, as far as i see.#asahi-dev · 2025-12-18 21:10
oliver_yea, i'll do that and test it again#asahi-dev · 2025-12-18 21:08
oliver_ok#asahi-dev · 2025-12-18 21:07
oliver_jannau: set them to a fixed (meaningless value of) 120 you mean? or mimic what macos does, increment the timer by 8.33ms on each submit?#asahi-dev · 2025-12-18 20:58
oliver_well, after some more experimentation, it seems that the dcp does not care much about the values in the three timestamp fields as long as they are non zero. even if i set them all to 1 i get smooth 120hz.#asahi-dev · 2025-12-18 18:40
oliver_I was playing with the idea of coming too, I actually live in Hamburg. But looks like it is already sold out. #asahi-dev · 2025-12-18 18:37
oliver_sven: will your talk be recorded? #asahi-dev · 2025-12-18 11:59
oliver_I interpret the first one as 0x4000 bytes at 0x38bc00000, which are also the values i would use with RegMonitor, is that correct?#asahi-dev · 2025-12-18 11:59
oliver_regarding the mmio in the adt, are we talking about the `reg` definitions in here?: https://github.com/AsahiLinux/linux/blob/773401986194d732f8f8a3b18b50d46d469a5dee/arch/arm64/boot/dts/apple/t600x-die0.dtsi#L330#asahi-dev · 2025-12-18 11:37
oliver_if the dcp would keep time in sync with the cpu, and macos reads and uses that time for swap_submit requests, then i would not get a perfect 8.33ms between each submit#asahi-dev · 2025-12-18 11:31
oliver_okay, and where can i find some information about what memory regions belong to the dcp? #asahi-dev · 2025-12-18 11:27
oliver_how can i read out the mmio, you have any pointers for me?#asahi-dev · 2025-12-18 11:25
oliver_looks like it is keeping track of its own "virtual" time#asahi-dev · 2025-12-18 11:24
oliver_yes, i get that, i would assume that macos writes ts1=cntpct+8.33ms... but it doesnt, it just writes ts1=ts1_prev + 8.33ms, even if cntpct increased by 100ms#asahi-dev · 2025-12-18 11:22
oliver_haha, yea i know what 8.33ms means, my question is: why is it set totally independent of the real time#asahi-dev · 2025-12-18 11:09
oliver_what i do see is that the dcp ts1 is updated by exactly 8.33ms, even when we get a real frame only every 100ms while tracing. for ts2 and ts3 macos adds ~14ms and ~12ms on top, independent of the target frame rate. so maybe it does not even depend on the realtime clock. thats something i'll try out now#asahi-dev · 2025-12-18 10:23
oliver_I did verify that cntpct correlates to realtime values. any idea why this might be the case?#asahi-dev · 2025-12-18 10:23
oliver_interesting: when comparing cntpct against the timestamps that are submitted in dcp swap requests, the values are totally off when tracing. cntpct vs dcp ts value: https://files.narf.zone/156ad921f5d8432d/Screenshot%20From%202025-12-18%2011-16-39.png#asahi-dev · 2025-12-17 17:53
oliver_Thank you#asahi-dev · 2025-12-17 16:53
oliver_print(f"XXX Timestamps {ts3:0.2f}\t{ts1-ts3:0.2f}ms\t{ts2-ts3:0.2f}ms\t{ts1-ts2:0.2f}ms\t{fps}hz")#asahi-dev · 2025-12-17 16:51
oliver_CNTVCT_EL0 or the clock of the dcp if that can differ.#asahi-dev · 2025-12-17 16:50
oliver_the first timestamp is ts3, the third timestamp in the payload. i did not yet look into how to read cntvct_el0. if you could give me a hint i can try that tomorrow ... at least if my kid doesnt need to stay home#asahi-dev · 2025-12-17 16:46
oliver_i can send you the trace file if you're interested and dont want to trace yourself#asahi-dev · 2025-12-17 16:45
oliver_in the gist? yea, offsets from the first value, which is ts3 in swap_submit#asahi-dev · 2025-12-17 16:43
oliver_ldat?#asahi-dev · 2025-12-17 16:41
oliver_i did some more dcp tracing to make more sense of the timestamps. interestingly, it looks like it doesnt really matter. in my traces, macos always sets ts1 to 12-13ms and ts2 to ~2ms. https://gist.github.com/oliverbestmann/96f2831df41874b77fab1939a98fc654#asahi-dev · 2025-12-16 13:34
oliver_jannau: yes, ts1 and ts2 are often zero, I think whenever there is just a partial update or overlay, i. e. macos does not set the first surface, only the second surface with x/y non zero. In that case the timestamps are also zero. Only ts3 seems to be set always. #asahi-dev · 2025-12-16 12:12
oliver_thanks guys, i'll have a look tomorrow and will adjust the constants. i'll probably have some questions regarding the mmio regions.#asahi-dev · 2025-12-16 12:04
oliver_chaos_princess: the test you've proposed is to have 120hz content on screen on the mac?#asahi-dev · 2025-12-16 12:03
oliver_so i am halfway confident that it actually updates with 120hz#asahi-dev · 2025-12-16 12:03
oliver_and i have a second laptop with a 120hz screen here where i could reproduce the results of the test programm on a non-macbook#asahi-dev · 2025-12-16 12:02
oliver_chadmed: yea using the demo programm from the pr as j`ey said. #asahi-dev · 2025-12-16 12:00
oliver_login screen, maybe desktop already. needed to move things around to get submits. #asahi-dev · 2025-12-16 11:57
oliver_yea, i've just graphed the dcp trace values and looked for anything that looked like a pattern. #asahi-dev · 2025-12-16 11:55
oliver_chaos_princess: i see. is that already in the kernel? then i can have a look into that the the next few days#asahi-dev · 2025-12-16 11:54
oliver_j`ey: you can just select 60hz I guess#asahi-dev · 2025-12-16 11:53
oliver_j`ey: never heard that before ;p#asahi-dev · 2025-12-16 11:52
oliver_chaos_princess: i am new to the kernel stuff, but the 24mhz fits neatly with the conversion factor of 41.67 that i need to use. would the timesync thing be useful ensure that the dcp clock does not drift from the kernels ktime_get_ns clock?