Just like with AV1, Intel beats AMD and Nvidia to support H.266 (VVC) decoding

Back in March 2022, Intel beat AMD and Nvidia to become the first company to announce full support for AV1 codec, ie, Arc could not only decode AV1 videos but encode them too. Today, Intel has again managed to sort of replicate the feat as its new Xe2 graphics is the first with VVC decoding capability, ahead of AMD and Nvidia.

Xe2 is making its debut inside Intel"s new Lunar Lake mobile processor, and the media engine of Xe2 is certainly impressive to bring VVC support. VVC or Versatile Video Codec is the successor to HEVC or high efficiency video codec. Hence, the formal is also officially referred to as H.266 just like HEVC is referred to as H.265 and it was announced back in 2020.

Just like the H.265 codec, H.266 also promised to bring even better quality at lower sizes by up to 50%. Sadly, unlike H.264 and 265, H.266 just did not get market adoption and AV1 became much more popular choice. Hopefully, with Intel adding support for the codec, eventually, AMD and Nvidia would do the same too. H.266, according to Intel, has an additional 10% file size reduction compared to AV1.

Coming to the GPU design itself, Intel says that Xe2 is 1.5x or 50% faster than the Xe graphics inside Meteor Lake. This gain is achieved thanks to the architectural improvements it has. Intel claims that Xe2"s worst-case improvement is 20% which is in tessellation while in the best-case scenario, there is up to 12.5 times improvement.

The improvements have been highlighted by Intel in the slides below:

It will be interesting to see how Xe2 will stack up against AMD"s new RDNA 3.5 integrated graphics processor (iGP) which was announced yesterday with its new Ryzen AI 300 series parts. Unlike Intel Xe2"s 67 TOPS, AMD does not seem to specifically state what the AI performance of RDNA 3.5 iGP is.

You can check out our entire Computex 2024 coverage in these articles here.

Report a problem with article
Next Article

These are the new Apple Stickers for WWDC 2024

Previous Article

Intel: Here's how Windows gets best performance out of Lunar Lake, with no hyper threading