DeepX claims 100B-parameter inference at under 5 W with second-gen chip
South Korean AI chip startup DeepX has reportedly taped out its second-generation processor, according to a report from EE Times. The company projects a 6.7x performance gain over its prior design by integrating LPDDR5X processing-in-memory technology. DeepX claims the chip can run inference on models of up to 100 billion parameters while drawing less than 5 watts of power. The announcement was shared on Hacker News, where it drew modest attention. The performance figures are the company's own projections and have not been independently verified.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.
Discussion (0)
Log in to join the discussion and vote.
Log in