TurboQuant Squeezes LLMs to 3 Bits, Claims Zero Accuracy Loss
Google Research says its new TurboQuant family can squeeze key-value caches down to 3 bits without retraining or accuracy drops. The catch: every benchmark comes from Google’s own labs.
GPU
NVIDIA Just Open-Sourced the Keys to Your GPU Kingdom
NVIDIA handed control of its GPU scheduling software to an open-source foundation at KubeCon Europe. Eight companies including AWS, Google Cloud, and Microsoft are contributing. The tools are free and available today.
Meta and AMD: A 6GW GPU Gamble to Power the Next AI Era
Meta and AMD announced a 6GW GPU partnership to scale AI infrastructure, leveraging AMD's Instinct GPUs and Meta's Helios architecture. The collaboration aims to diversify compute suppliers and integrate with Meta's silicon program.
Meta and NVIDIA’s AI Power Play: Scaling Efficiency or Just Another Tech Arms Race?
Meta and NVIDIA’s AI partnership targets energy efficiency and privacy through cutting-edge hardware and networking solutions, with Vera CPUs set for 2027 deployment.