Search results for UC San Diego
AI & Enterprise
Dinotisia unveils KV cache compression technology to ease AI computing bottlenecks
AI infrastructure company Dinotisia said on Wednesday it has released a paper and source code for KV cache compression technology that stores previously read context on GPUs to avoid recalculation by large language models. The company said its STAR-KV technology, a joint effort involving UC San Diego’s VVIP Lab and Dinotisia researchers, was selected as an ICML 2026 Spotlight paper. It said tests showed up to 75 percent reduction and up to 20-fold compression.
AI & Enterprise
Google to build data centre from 2,000 discarded Pixel phones in low-cost server test
Google has begun testing a low-cost data centre built from retired Pixel smartphones, aiming to cut costs and electronic waste by turning discarded devices into computing resources. Researchers stripped phones to mainboards and replaced Android with a Linux-based environment to run orchestration platforms such as Kubernetes. Tests showed clusters of dozens of phones could support applications used by students. The team is also pursuing a 2,000-phone facility designed to support about 100 classes.
AI & Enterprise
Nvidia unveils Ising open-source quantum AI models to tackle errors
Nvidia has unveiled an open-source artificial intelligence model family called Ising to calibrate quantum computers and correct errors. The company aims to use AI to address key bottlenecks that have held back commercial quantum computing. It chose decoding and calibration as initial applications. Ising includes an error-correction decoding model and a calibration model that automates ongoing tuning work. Nvidia also released guides, training data and Nvidia NIM microservices.
