'AV2 decoding is roughly five times more complex than AV1 decoding. In practice, that means software running on today’s hardware will struggle to decode AV2 in real time without careful, architecture-specific optimization'
AV1 software decoding is already very intensive so AV2 decoding benchmarks are the next thing that would be really interesting (or mortifying) to see.
I think you might be misunderestimating how incredible the dav1d AV1 decoder is. Not only does it require less total time than the reference decoder to decode the same video, but it can spread that out over far more threads. I was unable to watch 4k 60fps av1 video on my media center PC (it's from 2019, so predates hardware av1 decoding, and, well, the CPU was a little long in the tooth) until I switched to dav1d. With dav1d I am now able to watch 4k 60fps av1 using software decoding, and my machine uses 10% CPU while doing so. Really amazing piece of software.
With any luck, the dav2d 5x claim will hold true, and 10% CPU usage will scale to 50% CPU usage, meaning I'm still able to watch 4k 60fps video on my media center without a hardware upgrade. (that machine doesn't have hyperthreading, so 50% cpu is actually 50%, not 100% in a fancy suit)
Sorry if this sounds naive, but does it make sense to write a codec library in C/ASM considering how well Rust is progressing, especially when, as the author puts it, AV2 decoding is roughly five times more complex than AV1 decoding?
When AV1 was first announced, I got the impression the name was chosen partly as a pun/reference/homage to AVI, the classic but outdated format with used to be popular. Then when I saw Dav1d, OK, good way to continue the pun.
But now with AV2 and Dav2d, that completely breaks. Are we eventually going to get AV3/Dav3d and AV4/Dav4d, which will read like Ave/Daved and Ava/Davad? Seems a bit awkward. Was the idea from the start to have the 1 be the version number, and have it specifically be part of the name?
> Make it fast on older desktop, by writing asm for SSSE3+ chips
I guess 5 years ago (around the time when Intel stopped making SSE-only chips) is technically "older", but I wouldn't prioritize avx2 when devices intended for consuming media definitely experience much less pressure to upgrade than workstations…
> Lets see AI take the AV2 spec and Dav1d code and try to make a working high performance AV2 decoder.
That sounds like one of these high-risk, high-reward things that are great for people / projects / companies who have nothing to lose, but is not a great baseline strategy for an established market player. AV2 is here with support from aomedia and its members. AV2 will be used, and we need a production-grade decoder regardless of where AI is at, so it makes much more conservative business sense to use established approaches (language: c/asm, devteam: ffmpeg/dav1d) as a starting point. While that's happening, we can dabble in AI and other risky stuff and see if it helps. If so, great, and if not, nothing lost.
What's the high risk? That you end up with useless code?
I didn't mean that the Dav1d people should yolo vibe code Dav2d. My point was this this is a very interesting possible experiment since there is no existing Dav2d contamination in the training data.
Seems like the blog succumbed to the HN hug of death (`Actioning this file would cause "jbkempf.com//blog/2026/dav2d/" to exceed the per-day file actions limit of 160000 actions, try again later`), is there a copy available somewhere?
A codec spec isn't done until there is at least one decoder developed in the field. So reference + 1. The field implementations often become the de facto spec.
Reading the MPEG1 specs back in the 90s as a child opened my eyes to how to define complex systems. For a media coding standard, they spent most of their time saying how to interpret encoded bytes, which I realized is genius. Be descriptive about decoding and you don't have to be prescriptive about encoding. Encoding is where you can apply all the creativity, but you need to provide a way to have a shared understanding of the encoded bytes.
> The page you have tried to access is not available because the owner of the file you are trying to access has exceeded our short term bandwidth limits. Please try again shortly.
I'm not quite convinced a 25% reduction in size is worth effectively obsoleting all devices that have hardware decoders for AV1 but will struggle to decode AV2
A new codec doesn't obsolete old devices. At least, not right away.
Studios still release new dvds with mpeg2 video. Online videos tend to be available in many codecs. Video conferencing tends to negotiate to best available or has settled on ancient codecs and won't change quickly.
Same comment, same wording was had when going from H.264/AVC to H.265/HEVC and again to AV1. Across the board it's 30% better Size vs Quality and with an X amount increase in compute required to encode and decode.
Each time these standards were put into ink, they were years from being practical.
If decode is becoming so complicated and expensive the hardware can't handle it, why not just go full neural, send latents, and run decode on tensor cores?
The answer is probably the same as for why not AV2 everything; a lot of hardware couldn't support it today. But in 10 years?
It seems we're running up against fundamental limits of human-engineered video codecs at this point. There might be a lesson in there.
39 comments
[ 3.0 ms ] story [ 64.1 ms ] threadAV1 software decoding is already very intensive so AV2 decoding benchmarks are the next thing that would be really interesting (or mortifying) to see.
I think you might be misunderestimating how incredible the dav1d AV1 decoder is. Not only does it require less total time than the reference decoder to decode the same video, but it can spread that out over far more threads. I was unable to watch 4k 60fps av1 video on my media center PC (it's from 2019, so predates hardware av1 decoding, and, well, the CPU was a little long in the tooth) until I switched to dav1d. With dav1d I am now able to watch 4k 60fps av1 using software decoding, and my machine uses 10% CPU while doing so. Really amazing piece of software.
With any luck, the dav2d 5x claim will hold true, and 10% CPU usage will scale to 50% CPU usage, meaning I'm still able to watch 4k 60fps video on my media center without a hardware upgrade. (that machine doesn't have hyperthreading, so 50% cpu is actually 50%, not 100% in a fancy suit)
But now with AV2 and Dav2d, that completely breaks. Are we eventually going to get AV3/Dav3d and AV4/Dav4d, which will read like Ave/Daved and Ava/Davad? Seems a bit awkward. Was the idea from the start to have the 1 be the version number, and have it specifically be part of the name?
I guess 5 years ago (around the time when Intel stopped making SSE-only chips) is technically "older", but I wouldn't prioritize avx2 when devices intended for consuming media definitely experience much less pressure to upgrade than workstations…
Like we had weird examples like C compilers and Bun. This is a much more interesting example because its highly nontrivial.
AV1 exists, Dav1d exists. Lets see AI take the AV2 spec and Dav1d code and try to make a working high performance AV2 decoder.
That sounds like one of these high-risk, high-reward things that are great for people / projects / companies who have nothing to lose, but is not a great baseline strategy for an established market player. AV2 is here with support from aomedia and its members. AV2 will be used, and we need a production-grade decoder regardless of where AI is at, so it makes much more conservative business sense to use established approaches (language: c/asm, devteam: ffmpeg/dav1d) as a starting point. While that's happening, we can dabble in AI and other risky stuff and see if it helps. If so, great, and if not, nothing lost.
I didn't mean that the Dav1d people should yolo vibe code Dav2d. My point was this this is a very interesting possible experiment since there is no existing Dav2d contamination in the training data.
- https://web.archive.org/web/20260531130034/https://jbkempf.c...
- https://archive.md/ln5UE
Reading the MPEG1 specs back in the 90s as a child opened my eyes to how to define complex systems. For a media coding standard, they spent most of their time saying how to interpret encoded bytes, which I realized is genius. Be descriptive about decoding and you don't have to be prescriptive about encoding. Encoding is where you can apply all the creativity, but you need to provide a way to have a shared understanding of the encoded bytes.
HN hug of death
The AV2 Video Standard Has Released (Final v1.0 Specification)
https://news.ycombinator.com/item?id=48340910
Studios still release new dvds with mpeg2 video. Online videos tend to be available in many codecs. Video conferencing tends to negotiate to best available or has settled on ancient codecs and won't change quickly.
Each time these standards were put into ink, they were years from being practical.
The answer is probably the same as for why not AV2 everything; a lot of hardware couldn't support it today. But in 10 years?
It seems we're running up against fundamental limits of human-engineered video codecs at this point. There might be a lesson in there.
This is an odd signoff. Are people having a go at dav2d?