Our trusty desktop PCs and laptops have not seen any great innovation over the years. But that may be about to change, with almost all PC makers set to debut their AI PCs. Imagine not having to connect to the net to run a Gen AI chatbot like ChatGPT \u2013 that and more are what AI PCs will usher in.<\/p>\n
One of the tricks to turning your average PC into an AI-capable one lies in integrating a neural processing unit (NPU) into the hardware. NPUs are specialised hardware accelerators designed to efficiently perform tasks related to artificial neural networks, particularly deep learning tasks.<\/p>\n
Intel has been an early mover in this space and in 2016 acquired a company called Movidius that was building a discrete computer vision chip. That IP has been utilised in a number of places. \u201cBut what we also did with that IP was we embedded it into a CPU that can go into a PC or into an edge form factor. This chip now has a CPU, a NPU which is built from the Movidius IP, and a GPU, all in a single package,\u201d says Alexis Crowell, VP of sales, marketing & communications group and general manager of technology solutions, software & services for Asia Pacific Japan at Intel.<\/p>\n
Alexis says AI PCs will have tonnes of advantages over their traditional counterparts. \u201cIf you\u2019re an Excel junkie, you\u2019ll have better Excel capabilities, you\u2019ll also have better Microsoft Copilot capabilities. You can envision a creator, that\u2019s taken a video and they want to do auto-stabilisation, or they\u2019ve got a video of a skier and they want to do an auto zoom in and crop but keep that skier in the centre. So instead of the skier going up and down a mountain, the only thing you see is the skier centred the whole way, that\u2019s all AI. But now you can do it on that single form factor without shipping it all to the cloud,\u201d she says.<\/p>\n
Not having to rely on the cloud to process your AI needs has a few advantages.<\/p>\n
\u201cWith cloud, you need the bandwidth of shipping data back and forth, you\u2019ve got the latency cost that go with that. You also have the costs of running that somewhere else, whether it\u2019s your own data centre or one of the big public clouds. Whereas if you can bring more of that compute down into a form factor like an AI PC, then you\u2019re reducing your total cost of ownership pretty dramatically,\u201d Alexis says.<\/p>\n
Michael Collins, president and GM of Dell Technologies\u2019 global consumer and small business, says an AI PC would allow users to have bespoke AI systems that are trained on the users\u2019 own data. \u201cNow you\u2019ve got this large language model running on your data, able to go and put emails or presentations together without being connected to anything.\u201d This also means you needn\u2019t have any privacy concerns about your personal data.<\/p>\n
But obviously, you need horsepower, and a lot of it. Michael says you\u2019ll need a PC capable of running 40 TOPS (trillion operations per second) to run on-the-edge AI tools. That will mean the PCs you buy may not come cheap.<\/p>\n","protected":false},"excerpt":{"rendered":"
Our trusty desktop PCs and laptops have not seen any great innovation over the years. But that may be about to change, with almost all PC makers set to debut their AI PCs. Imagine not having to connect to the net to run a Gen AI chatbot like ChatGPT \u2013 that and more are what AI PCs will usher in. One of the tricks to turning your average PC into an AI-capable one lies in integrating a neural processing unit (NPU) into the hardware. NPUs are specialised hardware accelerators designed to efficiently perform tasks related to artificial neural networks, particularly […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[7],"tags":[],"class_list":["post-800737","post","type-post","status-publish","format-standard","hentry","category-it-2"],"acf":[],"_links":{"self":[{"href":"https:\/\/telecomlive.in\/web\/wp-json\/wp\/v2\/posts\/800737","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/telecomlive.in\/web\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/telecomlive.in\/web\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/telecomlive.in\/web\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/telecomlive.in\/web\/wp-json\/wp\/v2\/comments?post=800737"}],"version-history":[{"count":0,"href":"https:\/\/telecomlive.in\/web\/wp-json\/wp\/v2\/posts\/800737\/revisions"}],"wp:attachment":[{"href":"https:\/\/telecomlive.in\/web\/wp-json\/wp\/v2\/media?parent=800737"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/telecomlive.in\/web\/wp-json\/wp\/v2\/categories?post=800737"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/telecomlive.in\/web\/wp-json\/wp\/v2\/tags?post=800737"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}