It’s that time of year again. It can drop 20 degrees on any given day, and we’re stuck indoors watching people watch Qualcomm announce new chips and reference designs in sunny Hawaii. The Snapdragon Summit is the component-maker’s annual opportunity to map out its big plans for the next year, ahead of the holiday scrum and product deluge of CES and MWC.
It’s an ideal time to pepper the industry with some timeline news items. Many of the major manufacturers are effectively finished announcing hardware for the year, and things won’t really ramp up for another couple of months.
The big news is, naturally, Snapdragon 8 Gen 2. That’s the chip that’s going to power a majority of your flagship Android handsets next year — at least until the Snapdragon 8+ Gen 2 presumably starts rolling out at some point mid-2023.
It’s likely not surprising for those who have been following the space for the last several years that Qualcomm is positioning AI/ML as the centerpiece of its latest system on a chip. With the new Hexagon Processor (that’s a Qualcomm trademark, mind) at its center, the new system on a chip promises up to 4.35x gains for things like natural language processing.
“This is thanks to the industry’s only Micro Tile Inferencing technique so we can power features like real-time multi-language translation,” the company writes. “In other words, you can speak into a language translator and have it translated into multiple languages running these complex networks.”
Computational photography is the other big piece there. The system is able to recognize and segment different aspects of an image before the photo is taken. It uses a portrait as an example — breaking up hair, clothes, the background and a face into different segments. It’s a feature that will no doubt be present in imaging products like Portrait mode, in which depth sensing is important.
Read more about this at techcrunch.