AI researchers found a way to run LLMs at a lightbulb-esque 13 watts with no loss in performance

Tom's Hardware
Jun 26, 2024, 7:01 am103 pts
Elimination of matrix multiplication from LLM processing can massively increase performance-per-watt with the correct optimizations, researchers from UC Santa Cruz demonstrate. It remains to be seen how applicable this approach is for AI in general.

Read Article Share
Share Article
- email
- x.com
- facebook
- pocket
- reddit
- tumblr
- linkedin
- pinterest

Trending Today on Tech News Tube

Musk moseys into encyclopedia arena as Wikipedia multiplies AI data partnerships

Digitimes

Musk moseys into encyclopedia arena as Wikipedia multiplies AI data partnerships 128

Taiwan Ministry of National Defense releases major drone procurement order, increasing domestic IC design investment

Digitimes

Taiwan Ministry of National Defense releases major drone procurement order, increasing domestic IC design investment 128

Apple, Qualcomm, MediaTek align on 2nm chips in 2026 as specs lose consumer appeal

Digitimes

Apple, Qualcomm, MediaTek align on 2nm chips in 2026 as specs lose consumer appeal 127

56% of Companies Have Seen Zero Financial Return From AI Investments, PwC Survey Says

Slashdot

56% of Companies Have Seen Zero Financial Return From AI Investments, PwC Survey Says 121

For the price of Netflix, crooks can now rent AI to run cybercrime

The Register

For the price of Netflix, crooks can now rent AI to run cybercrime 121

Sony’s TV business is being taken over by TCL

The Verge

Sony’s TV business is being taken over by TCL 120

cURL Removes Bug Bounties

Slashdot

cURL Removes Bug Bounties 118

Amazon CEO Andy Jassy goes wobbly on AI bubble possibility

The Register

Amazon CEO Andy Jassy goes wobbly on AI bubble possibility 114

About Tech News Tube

Tech News Tube is a real time news feed of the latest technology news headlines.

Follow all of the top tech sites in one place, on the web or your mobile device.

Featured

oToBrite Electronics expands full-domain…

oToBrite Electronics expands full-domain visual AI solutions as automotive-grade…

OpenAI and ServiceNow Strike Deal to Put…

OpenAI and ServiceNow Strike Deal to Put AI Agents in Business Software

Setapp Mobile To Close in February as…

Setapp Mobile To Close in February as Alternative iOS App Store Economics Prove…

AI Agents 'Perilous' for Secure Apps…

AI Agents 'Perilous' for Secure Apps Such as Signal, Whittaker Says

Palantir CEO Says AI To Make Large-Scale…

Palantir CEO Says AI To Make Large-Scale Immigration Obsolete

Ransomware gang claims it hacked into…

Ransomware gang claims it hacked into Hyatt systems, says it has stolen data for…

New botnet targets HPE OneView…

New botnet targets HPE OneView vulnerability, so patch now

Ingram Micro reveals ransomware attack…

Ingram Micro reveals ransomware attack hit 42,000 people - here's how to find…

Key Apple, Nvidia, and Tesla supplier…

Key Apple, Nvidia, and Tesla supplier sees confidential files allegedly exposed…

Samsung reportedly to test EUV tools at…

Samsung reportedly to test EUV tools at Taylor fab in March