3 ms·
No one would make WiFi router chips or even low-end smartphone processors on these processes.
by stefan_ 5y ago
No one would make WiFi router chips or even low-end smartphone processors on these processes.
- kllrnohj 5y agoNot sure why you're being downvoted, you're absolutely correct. The low-end 400 series is a good 1-2 processes behind, being 11nm currently with the Snapdragon 480 5G (2021) on what looks to be Samsung's 8nm (aka, Samsung 10nm+), and the Qualcomm 215 (2019) is on 28nm even. Qualcomm isn't going to buy bleeding-edge foundry space for a low-end product, and definitely not for a wifi router which has almost no power or significant performance constraints.
- sempron64 5y agoThey never said which process the foundry services will provide, it's entirely possible they'll use the extremely mature 14nm++ process.
- paulmd 5y agothe article did say it, Qualcomm will be using the Intel 20A process. here's a better article, but the Reuters article did include the point about Qualcomm using Intel 20A. https://www.businesswire.com/news/home/20210726005136/en/ https://www.businesswire.com/news/home/20210726005136/en/
- jrockway 5y agoMy (limited) understanding is that a lot of beamforming / MIMO techniques are limited by the power budget of the WiFi chips. So if you could use the fanciest processes, the WiFi chips could be better. I don't know if that's worth the price (everyone expects WiFi to suck 24/7), or if the process helps there, but it does sound like manufacturing technology, not math, is the limiting factor for WiFi right now.
- stefan_ 5y agoHaving extra chains is a ton of power, but it's also analog stuff - that's already not what cutting edge processes are for.