#asahi-dev · 2026-07-14 19:57
sofusactually why is the slop policy in the docs?#asahi-dev · 2026-07-14 19:50
sofusthats nice, if we cant do the same for our for of linux, i feel like we should at least add a link to the contributing page#asahi-dev · 2026-07-14 19:12
sofusi feel like the slop policy should be more out in the open, unless im stupid its not easy to find#asahi-dev · 2026-07-14 13:37
sofusfor tracing?#asahi-dev · 2026-07-14 13:36
sofuswhy not target a newer dcp version?#asahi-dev · 2026-07-14 13:35
sofushow do you even make a mistake like that#asahi-dev · 2026-07-14 13:32
sofusim both jealous and very happy that avd does not talk that much :p#asahi-dev · 2026-07-14 13:31
sofusnice, getting the newer fw running is needed for dcp on m3 right?#asahi-dev · 2026-07-14 09:56
sofusSAODBLK_B_MainConcept_4 passed! To more left for the "main" tests to be done#asahi-dev · 2026-07-14 09:37
sofusheck yeah DSLICE_C_HHI_5 passed :p#asahi-dev · 2026-07-14 09:01
sofusi think bit 0 is set in the flags passed to set_coded_slice whenever its a new slice segment or the TileId changes#asahi-dev · 2026-07-13 12:45
sofusother than ^ only two more to go#asahi-dev · 2026-07-13 09:26
sofusSAODBLK_B_MainConcept_4 is the worst one, with that one most of the logic seems to break#asahi-dev · 2026-07-13 09:24
sofusbut i also feel like there should be something in the spec we can use instead of this. So far i have not been able to find it...#asahi-dev · 2026-07-13 09:23
sofusmanly its the codec slice flags and when to do the cm3_cmd_set_ctb_xy command thats wrong#asahi-dev · 2026-07-13 09:22
sofusDSLICE_C_HHI_5#asahi-dev · 2026-07-13 09:21
sofusa bitstream?#asahi-dev · 2026-07-13 09:18
sofusit works for almost all streams but not for a few one. And i cant figure out whats special about the ones where it fails#asahi-dev · 2026-07-13 09:17
sofusim really having trouble with this thing: https://github.com/eiln/avd/blob/e8dfcb70dcef9fbf6cc4f11443647d9d49c3d270/avid/h265/halv3.py#L389#asahi-dev · 2026-07-13 09:17
sofushevc main is gaining progress, 14 tests still failing on JCT-VC-HEVC_V1#asahi-dev · 2026-07-09 18:35
sofusyay for avd :p#asahi-dev · 2026-07-07 11:42
sofusideally it would deal with the vp's like the driver. We would use the piodma and try to keep as litle as possible in cm3 memory#asahi-dev · 2026-07-07 11:36
sofuswe could basicly drop the vp slot dance thing by not writing directly to the registers#asahi-dev · 2026-07-07 11:16
sofusin the adt's there is an decode-samples-per-seconds parameter. All SoC's have the same one expect for neo and base m1, and thoose have the least amount of vp's, so i would think some performance can be gained#asahi-dev · 2026-07-07 11:11
sofusthey are very usefull for multislice in h26{4,5}#asahi-dev · 2026-07-07 11:09
sofusi found them by testing, i have never seen macos use them#asahi-dev · 2026-07-07 11:09
sofusi think they are independent, they can be used in parallel, im not sure if they give any performance boost / penalty#asahi-dev · 2026-07-07 09:23
sofuschaos_princess: try sed -i 's/0x2b000000/0x2bfff000/g' drivers/media/platform/apple/avd/avd-{drv,vp9}.c#asahi-dev · 2026-07-07 09:20
sofusoops loks like i should have waited a second#asahi-dev · 2026-07-07 09:20
sofusdts nodes for t8103 are back and i updated the buffer :p #asahi-dev · 2026-07-07 08:49
sofuswhat?? Looks like i removed them?#asahi-dev · 2026-07-07 08:37
sofusvp9 seems very slow, i should also tests that macos is not faster#asahi-dev · 2026-07-07 08:36
sofuschaos_princess: one of the buffers just needed a bit more space, i will tests a few more bit streams before updating#asahi-dev · 2026-07-07 07:59
sofusnice! on t6000?#asahi-dev · 2026-07-07 07:46
sofusi cant get gstreamer to select the right format if i add more than one, so i disabled that for now#asahi-dev · 2026-07-07 07:45
sofuschaos_princess: i force-pushed a fix, please also enable the DEBUG_INST* in case it does not work, should be unrelated to t8103#asahi-dev · 2026-07-06 20:47
sofusall the yuv4* are actually lying, i found out it was software decoding#asahi-dev · 2026-07-06 20:46
sofusand macos for reference https://pastebin.com/wZfq4aqU#asahi-dev · 2026-07-06 20:46
sofusthanks, here are the ones i have testet (and the results) https://pastebin.com/f1LqPi3V#asahi-dev · 2026-07-06 20:41
sofuswhere did you get theese? I tested with some other ones#asahi-dev · 2026-07-06 20:38
sofus* zero#asahi-dev · 2026-07-06 20:38
sofusi assumed there was close to no userspace support for anything other than profile one#asahi-dev · 2026-07-06 20:33
sofusit should actually work, its just the bytes per line calculation being wrong#asahi-dev · 2026-07-06 20:33
sofusoh i see bd10 bit depth 10 maybe? That would make sense#asahi-dev · 2026-07-06 20:26
sofuswhoah ha#asahi-dev · 2026-07-06 20:11
sofuswith a 4k video#asahi-dev · 2026-07-06 20:11
sofusoh wow, looks like i save ~4 W#asahi-dev · 2026-07-06 20:09
sofushow did i not know you can just give mpv a yt link?? #asahi-dev · 2026-07-06 20:04
sofusforce-pushed now#asahi-dev · 2026-07-06 20:03
sofusoops forgot to add the updated header