The world’s first, Google Tensor G3 chip supports mobile phone hardware accelerated AV1 encoding but lacks software adaptation!
On February 24 that Google’s self-developed chip Tensor G3 was unveiled at Google’s new hardware product launch on October 4 last year and was used in the Pixel 8 and Pixel 8 Pro phones.
Mishaal Rahman discovered that although Google did not mention it, the code shows that it supports 4K 60FPS AV1 encoding (but in fact there is currently no software support at all). This is also the world’s first smartphone SoC to support hardware-accelerated AV1 encoding (other smartphone SoCs None support AV1 encoding).
Rahman also pointed out that the default camera app on Google Pixel 8 and Pixel 8 Pro also does not support advanced AV1 encoding options, which is a complete waste for Tensor G3.
We noticed that a user in his comment area tried to correct it, saying that the A17 Pro of Apple iPhone 15 Pro /Max also supports AV1 video encoding, but in the end it was discovered that the A17 Pro actually only supports AV1 decoding rather than encoding.
AV1 (AOMedia Video 1) is an open, patent-free video encoding format designed for network streaming. It was developed by the Alliance for Open Media (AOMedia) and aims to replace its VP9.
In 2018, Facebook (now Meta) conducted tests based on simulation conditions close to the real environment and found that the AV1 data compression rate could be improved by 50.3%, 46.2% and 34.0% respectively compared with x264 main profile, x264 high profile and libvpx-vp9.
Compared with the existing H.264/AVC and HEVC formats, AV1 adopts a licensing model similar to VP9, which can be used free of license fees, so it will not cause obstacles to software development like formats such as H.264, which calculate licensing fees based on sales volume.
Previous Post Up to 18,000 characters, Microsoft Copilot chatbot opens “notebook” function