zetalex Posted June 21 Share Posted June 21 Hello: I am tyring to build the latest version of moonlight-qt (6.0.0). Up until now, I succeded by removing the disable-libdrm and disable-wayland to enable hardware acceleration and use H264, H265 and AV1 all of them through hardware acceleration. However, I get huge decoding times and they are even variable, ranging from 40ms to 300ms depending on what is going on in the streaming. The 4.3.1 version provided by @amazingfate works very good at 4ms decoding time but I would like to test the new AV1 encoder and features from the latest version. What steps should I follow to get it working correctly? Thank you in advance 0 Quote Link to comment Share on other sites More sharing options...
amazingfate Posted Saturday at 12:51 AM Share Posted Saturday at 12:51 AM Decoding time is ok if there is no frame drop. 0 Quote Link to comment Share on other sites More sharing options...
zetalex Posted Sunday at 08:17 AM Author Share Posted Sunday at 08:17 AM Hello again @amazingfate: I did some more tests to check what you said. First, there are no frame drops, video is smooth using any decoder. Regarding decoding time, the behaviour is a little bit strange: the higher the framerate that moonlight renders, the lower the decoding time. When I am running 120fps, the decoding time is 35 ms which is as low as I managed to get it to. In your 4.3.1 moonlight that you have in your ppa at launchpad I get 4ms decoding time. Sorry for the dumb question, I am just trying to understand: did you apply any patch or did anything in the source code to make the decoding time lower? Thanks in advance 0 Quote Link to comment Share on other sites More sharing options...
Solution amazingfate Posted Sunday at 08:53 AM Solution Share Posted Sunday at 08:53 AM 4.3.1 moonlight is linked to ffmpeg 4.4, while new moonlight is linked to ffmpeg 6 because libavcode-dev in my ppa is updated to 6.0. Decoding time caculation is ffmpeg releated. I think moonlight can't distinguish decoding time and rendering time with new ffmpeg. 0 Quote Link to comment Share on other sites More sharing options...
zetalex Posted Sunday at 09:22 AM Author Share Posted Sunday at 09:22 AM Thanks for answering so quickly. I didnt know that the decoding time measurement was performed this way. So, there shouldnt be any difference between both versions performance then. I didnt try to play much and latency didnt seem as noticeable. Thanks for sharing your knowledge again! PS: I put another post about linux kernel 6.1 not booting after updating some libraries from your ppa, could you take a look or tell me how can I debug it? 0 Quote Link to comment Share on other sites More sharing options...
q-bert Posted Monday at 03:44 PM Share Posted Monday at 03:44 PM @zetalex Were you able to use 4k120? If so, how did you enable 4k120 in Linux? I am not able to set the display (my TV) to 4k120, only 4k60... My TV's HDMI port is definitely a 4k120 port (verified with my PC and Nvidia GPU). Have you seen this tutorial? I lists all required steps if you want to use hardware acceleration on RK3588 with 6.1 kernel. https://jellyfin.org/docs/general/administration/hardware-acceleration/rockchip 0 Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.