4 ms·

Video compression is a calculus of IO capacity, memory, and algorithmic complexity. Take the MPEG-1 codec for instance, it was new about 30 years ago. While tod

Video compression is a calculus of IO capacity, memory, and algorithmic complexity. Take the MPEG-1 codec for instance, it was new about 30 years ago. While today most people think of MPEG-1 videos as low quality the spec provides the ability to handle bit rates up to 100Mb/s and resolutions up to 4095x4095. That was way higher than the hardware of the time supported. One of MPEG-1's design goals was to get VHS-quality video at a bitrate that could stream over T1/E1 lines or 1x CD-ROMs. The limit on bitrate led to increased algorithmic complexity. It was well into the Pentium/PowerPC era until desktop systems could play back VCD quality MPEG-1 video in software. Later MPEG codecs increased their algorithmic complexity to squeeze better quality video into low bit rates. A lot of those features existed on paper 20-30 years ago but weren't practical on hardware of the time, even custom ASICs. Even within a spec features are bound to profiles so a file/stream can be handled by less capable decoders/hardware. There's plenty of video codecs or settings for them that can choke modern hardware. It also depends on what you mean by "modern hardware". There's codecs/configurations a Threadripper with 64GB of RAM in a mains powered jet engine sounding desktop could handle in software that would kill a Snapdragon with 6GB of RAM in a phone. There's also codecs/configurations the Snapdragon in the phone could play using hardware acceleration that would choke a low powered Celeron or Atom decoding in software.

cellularmitosis 6y ago
I wonder why all of the MPEG1 encoders of the day enforced a maximum of 320x240?
- giantrobot 6y ago
  While the spec allowed for outrageous settings playback wouldn't have been possible. Most hardware decoders were meant for (or derived from) VCD playback. The VCD spec covered CIF video which mean QVGA would fall into the supported macroblock rate for hardware decoders. In MPEG-1's heyday there would haven't been a lot of point in encoding presets producing content common hardware decoders couldn't handle. There were several other video codecs in the same era that didn't have hardware decode requirements. Cinepak was widely used and could be readily played on a 68030, 486, and even CD-ROM game consoles. As I recall Cinepak encoders had more knobs and dials since the output didn't need to hit a hardware decoder limitation.
- anthk 6y ago
  PAL/NTSC resolutions.
AareyBaba 6y ago
Are there codecs that require high compute (Threadripper) for encode but can be easily decoded on a Snapdragon ?
- monocasa 6y ago
  A lot of video codecs are NP hard to encode optimally, so rely on heuristics. So you could certainly say that some approaches take a lot of compute power to encode, but are much more easily decodable.
  - cogman10 6y ago
    The codecs aren't NP hard. Rather, the "perfect" encode is. That's where the heuristics are coming into play. The codec just specifies what the stream can look like, the encoders have to pick out how to write that language and the decoders how to read it. Decoders are relatively simple book keepers/transformers. Encoders are complex systems with tons of heuristics. This is also why hardware decoders tend to be in everything and are relatively cheap with equal quality to software counterparts. On the flip side, hardware encoders are almost always worse than their software counterparts when it comes to the quality of the output (while being significantly faster).
    - monocasa 6y ago
      > The codecs aren't NP hard. Rather, the "perfect" encode is. That's what I meant by my first sentence. And I'll throw out there that the vast majority of 'hardware codecs' are in fact software codecs running on a pretty general purpose DSP. You could absolutely reach the same quality as a high quality encoder given the right organizational impetus of the manufacturer; they simply are focused on reaching a specific real time bitrate for resolution rather than overall quality. By the time they've hit that, there's a new SoC with it's own DSPs and it's own Jira cards that needs attention. If these cores were more open, I'm sure you'd see less real time focused encoder software targeting them as well.
- giantrobot 6y ago
  Pretty much all of them. Encode complexity for most codecs is way higher than decode complexity (on purpose). This has been an issue with AV1, it's got relatively high decode complexity and there's not a lot of hardware acceleration available. The encode complexity is fantastic though and is very slow even on very powerful hardware, less than 1fps so ~30 hours to encode a one hour video. Even Intel's highly optimized AV1 encoder can't break 10fps (three hours to encode an hour of video) while their h.265 encoder can hit 300fps on the same hardware.

cellularmitosis 6y ago

I wonder why all of the MPEG1 encoders of the day enforced a maximum of 320x240?

giantrobot 6y ago
While the spec allowed for outrageous settings playback wouldn't have been possible. Most hardware decoders were meant for (or derived from) VCD playback. The VCD spec covered CIF video which mean QVGA would fall into the supported macroblock rate for hardware decoders. In MPEG-1's heyday there would haven't been a lot of point in encoding presets producing content common hardware decoders couldn't handle. There were several other video codecs in the same era that didn't have hardware decode requirements. Cinepak was widely used and could be readily played on a 68030, 486, and even CD-ROM game consoles. As I recall Cinepak encoders had more knobs and dials since the output didn't need to hit a hardware decoder limitation.
anthk 6y ago
PAL/NTSC resolutions.

AareyBaba 6y ago

Are there codecs that require high compute (Threadripper) for encode but can be easily decoded on a Snapdragon ?

monocasa 6y ago
A lot of video codecs are NP hard to encode optimally, so rely on heuristics. So you could certainly say that some approaches take a lot of compute power to encode, but are much more easily decodable.
- cogman10 6y ago
  The codecs aren't NP hard. Rather, the "perfect" encode is. That's where the heuristics are coming into play. The codec just specifies what the stream can look like, the encoders have to pick out how to write that language and the decoders how to read it. Decoders are relatively simple book keepers/transformers. Encoders are complex systems with tons of heuristics. This is also why hardware decoders tend to be in everything and are relatively cheap with equal quality to software counterparts. On the flip side, hardware encoders are almost always worse than their software counterparts when it comes to the quality of the output (while being significantly faster).
  - monocasa 6y ago
    > The codecs aren't NP hard. Rather, the "perfect" encode is. That's what I meant by my first sentence. And I'll throw out there that the vast majority of 'hardware codecs' are in fact software codecs running on a pretty general purpose DSP. You could absolutely reach the same quality as a high quality encoder given the right organizational impetus of the manufacturer; they simply are focused on reaching a specific real time bitrate for resolution rather than overall quality. By the time they've hit that, there's a new SoC with it's own DSPs and it's own Jira cards that needs attention. If these cores were more open, I'm sure you'd see less real time focused encoder software targeting them as well.
giantrobot 6y ago
Pretty much all of them. Encode complexity for most codecs is way higher than decode complexity (on purpose). This has been an issue with AV1, it's got relatively high decode complexity and there's not a lot of hardware acceleration available. The encode complexity is fantastic though and is very slow even on very powerful hardware, less than 1fps so ~30 hours to encode a one hour video. Even Intel's highly optimized AV1 encoder can't break 10fps (three hours to encode an hour of video) while their h.265 encoder can hit 300fps on the same hardware.