➀ Intel plans to launch the Core Ultra 200HX mobile CPU lineup with up to 24 cores and a 5.5GHz boost speed; ➁ The lineup includes the Core Ultra 9 285HX as the flagship model, followed by five other models; ➂ Detailed specifications of each model, including core counts, boost speeds, and iGPU configurations, have been leaked.

➀ Beelink launches the SER9 mini-PC with AMD Ryzen AI 9 HX 370 processor; ➁ Priced at $999 with 32GB LPDDR5X-7500 RAM and 1TB SSD; ➂ Features 12 cores, four Zen 5 cores, eight Zen 5c cores, and 24 threads.

➀ The AMD Ryzen AI 300 Pro series is expected to be released soon, with rumors suggesting October 10th as the release date. The Ryzen AI 7 Pro 360, the first SKU in the lineup, will feature an 8-core configuration. ➁ The Ryzen AI 7 Pro 360 is expected to have eight cores and sixteen threads, with a configuration of 4x Zen5 and 4x Zen5c cores. ➂ The Ryzen AI 7 Pro 360 has shown strong performance in Geekbench AI benchmark tests, outperforming Intel Core Ultra 9 185H in single and half precision, and by about 20% in quantized score.

➀ Software-first approach to simplify development and maximize hardware utilization; ➁ Programmable Assembly Line Architecture with data 'conveyor belts' for efficient instruction and data movement; ➂ Deterministic Compute & Networking ensuring predictable execution times; ➃ On-chip Memory integration for high-speed data access and reduced latency.

➀ Efficient announced a strategic partnership with GlobalFoundries; ➁ The new processor is up to 166x more energy-efficient than standard embedded CPUs; ➂ Early access and customer sampling are planned for summer 2025

➀ Researchers at Peking University have developed the world's first tensor processing unit (TPU) based on carbon nanotubes, significantly enhancing AI processing efficiency. ➁ The TPU utilizes carbon nanotube field-effect transistors (FETs) in a systolic array architecture, enabling parallel execution and reducing energy consumption. ➂ The chip demonstrated high energy efficiency and low power consumption in image recognition tasks, paving the way for future energy-efficient AI computing.

1. Quadric introduces the third generation of general-purpose neural processing units (GPNPUs) with pre-integrated clusters of two, four, or eight cores, delivering up to 108 TOPS. 2. The QC series includes single-core and multi-core cluster versions, optimized for generative AI and safety-enhanced for automotive applications. 3. The QC-M family scales to support small to high-compute applications, providing up to 864 TOPS for Level 4 central ADAS applications.