Core Insights - The article highlights the launch of the new generation GPU architecture "Huagang" by Moer Thread, which supports full precision computing from FP4 to FP64, with a 50% increase in density and a 10-fold improvement in efficiency [1][2]. Group 1: Product Features - The new GPU architecture "Huagang" features significant performance enhancements, with a 50% increase in computing density and optimized energy efficiency, supporting full precision calculations from FP4 to FP64 [2]. - It integrates a new asynchronous programming model and supports large-scale interconnectivity, enabling the expansion of intelligent computing clusters with over 100,000 cards through the self-developed MTLink high-speed interconnect technology [2]. - The architecture includes a built-in AI generative rendering framework and enhances hardware ray tracing acceleration, fully supporting DirectX 12 Ultimate for high synergy between graphics rendering and intelligent computing [2]. Group 2: Future Developments - The company plans to launch high-performance AI training and inference chip "Huashan" and a chip specialized in high-performance graphics rendering named "Lushan" based on the MUSA architecture [2]. - Moer Thread introduced the "Kua'e Wanka" intelligent computing cluster, which provides full precision and general computing capabilities, achieving efficient and stable AI training and inference at a scale of 10,000 cards [2]. - The company also unveiled the AI computing device MTT AIBOOK, equipped with the intelligent SoC chip "Changjiang" [2]. Group 3: Strategic Vision - The CEO emphasized that the ecosystem is the core moat and value of the GPU industry, and the company is committed to increasing R&D investment to tackle core technological challenges from hardware to software [3]. - Moer Thread aims to deepen collaboration with ecosystem partners through open innovation to build a self-reliant and robust domestic computing industry ecosystem [3].
效能提升10倍！摩尔线程新一代全功能GPU架构“花港”正式发布