Differences
This shows you the differences between two versions of the page.
| Both sides previous revision Previous revision Next revision | Previous revision | ||
| hardware:en-decoder [2020/11/16 19:10] – derpeter | hardware:en-decoder [2026/07/26 22:17] (current) – external edit 127.0.0.1 | ||
|---|---|---|---|
| Line 1: | Line 1: | ||
| - | = Video Acceleration | + | # Video Acceleration |
| * https:// | * https:// | ||
| * https:// | * https:// | ||
| * https:// | * https:// | ||
| - | = APIs | + | # APIs |
| - | == VAAPI | + | ## VAAPI |
| VAAPI was specified by intel but can be used with hardware of different vendors. It is well supported in gstreamer and ffmpeg. In most cases this is you best option. Only for some hardware / driver combinations you may get better results with one of the more proprietary APIs / drivers. | VAAPI was specified by intel but can be used with hardware of different vendors. It is well supported in gstreamer and ffmpeg. In most cases this is you best option. Only for some hardware / driver combinations you may get better results with one of the more proprietary APIs / drivers. | ||
| Line 68: | Line 67: | ||
| - | == VDPAU | + | ## VDPAU |
| * https:// | * https:// | ||
| - | == OMX / OpenMAX | + | ## OMX / OpenMAX |
| * https:// | * https:// | ||
| - | = Hardware | + | # Hardware |
| Hardware en/decoder can be found in a number of chips. The probably most common type are GPUs or GPU components of CPUs and SOCs. | Hardware en/decoder can be found in a number of chips. The probably most common type are GPUs or GPU components of CPUs and SOCs. | ||
| Besides that there are also some dedicated de/encoder cards available. You probably can also use FPGAs but we didn't look into that (yet) | Besides that there are also some dedicated de/encoder cards available. You probably can also use FPGAs but we didn't look into that (yet) | ||
| - | == Intel | + | ## Intel |
| Intel implement the video acceleration in the GPU parts of their CPUs. Its mostly available in Desktop and Tablet CPUs but also in some Xeon Server CPUs. | Intel implement the video acceleration in the GPU parts of their CPUs. Its mostly available in Desktop and Tablet CPUs but also in some Xeon Server CPUs. | ||
| The hardware feature is called QuickSync an can be used via different APIs / Libs. Intel makes it not easy to user to figure out which CPU/GPU can do what, | The hardware feature is called QuickSync an can be used via different APIs / Libs. Intel makes it not easy to user to figure out which CPU/GPU can do what, | ||
| Line 106: | Line 105: | ||
| To see the load of your GPU, which also will tell you how much headroom you have for more streams, you can use intels < | To see the load of your GPU, which also will tell you how much headroom you have for more streams, you can use intels < | ||
| - | == Nvidia | + | ## Nvidia |
| As usual NVIDIA provides the worst user experience possible by limiting the feature not based on hardware capabilities but on the price of the card. | As usual NVIDIA provides the worst user experience possible by limiting the feature not based on hardware capabilities but on the price of the card. | ||
| This means you will not be able to en- / decode much stream in parallel if you don't pay a an absurd amount of money. But never the less you might already own a card .... | This means you will not be able to en- / decode much stream in parallel if you don't pay a an absurd amount of money. But never the less you might already own a card .... | ||
| Line 128: | Line 127: | ||
| To see the load of your GPU use nvidia' | To see the load of your GPU use nvidia' | ||
| < | < | ||
| - | == AMD / ATI | + | ## AMD / ATI |
| - | === Radeon | + | ### Radeon |
| The AMD stand alone GPUs provide also video en/ decoder capabilities. With recent Linux Kernels e.g. the NAVI Gpus (Radeon GT 5xxx) can be used without binary drivers. | The AMD stand alone GPUs provide also video en/ decoder capabilities. With recent Linux Kernels e.g. the NAVI Gpus (Radeon GT 5xxx) can be used without binary drivers. | ||
| The first AMD/ATI GPU with video accelerator was R600 architecture from 2007. AMD developed different accelerator suites over time: | The first AMD/ATI GPU with video accelerator was R600 architecture from 2007. AMD developed different accelerator suites over time: | ||
| Line 170: | Line 169: | ||
| while gstreamer seems not support the usage jet. | while gstreamer seems not support the usage jet. | ||
| - | == Broadcom | + | ## Broadcom |
| - | === Christal HD | + | ### Christal HD |
| Broadcoms Christal HD cards where mostly used for decoding video on devices not capably to to it in CPU. It can only do decoding so it is not that relevant here. | Broadcoms Christal HD cards where mostly used for decoding video on devices not capably to to it in CPU. It can only do decoding so it is not that relevant here. | ||
| * https:// | * https:// | ||
| - | === Raspberry Pi 1-4 | + | ### Raspberry Pi 1-4 |
| The SOCs used in the raspberrypis can be used for de/encoding of video. It can be accessed with OMX api which is present in gstreamer and ffmpeg | The SOCs used in the raspberrypis can be used for de/encoding of video. It can be accessed with OMX api which is present in gstreamer and ffmpeg | ||
| * https:// | * https:// | ||
| Line 194: | Line 193: | ||
| omxhdmiaudiosink: | omxhdmiaudiosink: | ||
| </ | </ | ||
| - | = Software | + | # Software |
| - | == FFMPEG | + | ## FFMPEG |
| * https:// | * https:// | ||
| * https:// | * https:// | ||
| - | == Gstreamer | + | ## Gstreamer |
| To see which de/encoder gstreamer can use on your system you can run e.g.: | To see which de/encoder gstreamer can use on your system you can run e.g.: | ||
| < | < | ||