https://medium.com/@chrishantha/linux-performance-observability-tools-19ae2328f87f

For Linux, QSV will use the libmfx library (MFX should be the proper term for QSV, MFX uses a modified VAAPI, see notes below). Normal VAAPI will use the libva library. The Debian link I added shows a list of supported HW for VA-API, we could incorporate this into the guide.

grep -rnw '/path/to/somewhere/' -e 'pattern'

Intel Media Driver https://github.com/intel/media-driver

AMF is supported on Linux but x265 coding is not. https://www.phoronix.com/scan.php?page=news_item&px=FFmpeg-AMD-AMF-Vulkan

Found this ffmpeg has AMF enabled for x264/5 https://www.reddit.com/r/Amd/comments/8eirp4/ffmpeg_40_released_includes_amf_hardware_encoding/

On Ubuntu, the guide for setting up QSV is actually just using VAAPI.
https://wiki.ubuntu.com/IntelQuickSyncVideo

ffmpeg page for QSV states only Kaby Lake can do HEVC. "Encode a 1080p HEVC main10 profile (supported since Kaby Lake platform)" http://trac.ffmpeg.org/wiki/Hardware/QuickSync

As always, the Arch Wiki lol. It does not mention MSDK or libmfx. https://wiki.archlinux.org/index.php/Hardware_video_acceleration

Information on MSDK and libmfx https://software.intel.com/en-us/articles/quick-start-on-integrating-ffmpeg-libraries

AMF official support. https://github.com/GPUOpen-LibrariesAndSDKs/AMF/blob/master/README.md

Enabling VAAPI for Intel Processors https://gist.github.com/Brainiarc7/eb45d2e22afec7534f4a117d15fe6d89

Here's the Programmers Manual for MFX https://01.org/sites/default/files/documentation/intel-gfx-prm-osrc-kbl-vol08-media_vdbox.pdf

Shows a nice breakdown for ffmpeg. 01.org is Intel's Open Source Technology Center https://01.org/linuxmedia

https://stackoverflow.com/questions/43898097/how-to-determine-the-specific-h264-encoder-in-use-from-ffmpeg-library

https://ffmpeg.org/doxygen/2.8/demuxing__decoding_8c_source.html

https://github.com/FFmpeg/FFmpeg/blob/aff8cf18cb0b1fa4f2e3d163c3da2f25aa6d1906/libavcodec/mpeg12dec.c#L1189

https://trac.ffmpeg.org/wiki/Encode/AAC

https://www.reddit.com/r/jellyfin/comments/dh89fu/nvidia_hardware_transcoding/

https://stackoverflow.com/questions/42688491/how-do-you-change-the-docker-container-tz-in-spring#42699704

Here is a note on the Ubuntu page about libmfx (MSDK)

Future Approach #2: Media SDK ("MSDK")
(This is the approach that the FFmpeg documentation rather confusingly refers to as "Intel QSV")

Intel's Media SDK is the company's professional offering with the aim of providing the highest performance and the most video features all accelerated on Intel chips.

It sounds great. You can download the SDK for free, and they have also published a first open source release of it too. But there are hidden reasons why we can't use any of this yet in Ubuntu. The problem is that pieces are missing. Both of them have a dependency on a custom proprietary fork of LibVA (VA-API). While this "missing" code is actually provided in the commercial download, it requires that you:

Overwrite official Ubuntu files, possibly breaking your system; and
Recompile all your video players yourself; and
Don't redistribute what you've made.
That won't work for us right now. And we don't recommend you try. It's very complicated and can easily break your Ubuntu installation.

I haven't checked, but is ffmpeg running with --enable-libmfx --enable-nonfree This might be required to enable QSV/MFX, testing is needed. https://software.intel.com/en-us/articles/accessing-intel-media-server-studio-for-linux-codecs-with-ffmpeg

List of supported codecs https://github.com/Intel-Media-SDK/MediaSDK

Supported video encoders: HEVC, AVC, MPEG-2, JPEG, 
VP9 Supported video decoders: HEVC, AVC, VP8, VP9, MPEG-2, VC1, JPEG 
Supported video pre-processing filters: Color Conversion, Deinterlace, Denoise, Resize, Rotate, Composition