寅谱计算携手AMD发布Infplane Mini AI工作站：Hilbert

Core Insights - Infplane Computing has launched the Mini AI workstation named Hilbert, equipped with its KLEENE motherboard control technology, aimed at providing a stable edge computing solution for local large model inference, scientific computing, and vertical industry AI applications [1] - The Mini AI workstation is the first product in Infplane's "desktop supercomputing" series, developed in collaboration with several ecosystem partners including AMD, aoni, and abee, and is expected to enter the North American market by 2026 [1] - The workstation is powered by the AMD Ryzen AI Max+395 processor, which features an innovative CPU, GPU, and NPU architecture, delivering powerful heterogeneous computing capabilities [1] Technical Specifications - The unified memory architecture (UMA) of the workstation offers up to 96GB of large video memory, suitable for MoE architecture large language models, and is compatible with various multimodal models such as machine vision, automatic speech recognition, and image understanding [1] - The KLEENE technology focuses on the underlying control layer of the computer, optimizing power consumption, performance, and workload distribution of the processor, memory, and SSD without altering hardware design or software compatibility [2] - This system provides a reusable and transferable methodology that can serve various terminal forms, including laptops, desktops, mobile devices, automotive systems, and robotic controllers [2] Market Impact - AMD's Vice President of Marketing for Greater China, Ji Chaohui, highlighted the unique advantages of the Mini AI workstation, including high computing power and large memory, which have garnered widespread industry attention and recognition [2] - The integration of edge AI with vertical industry scenarios is gradually being realized, expanding the applicability of Mini AI workstations in various fields [2] - Infplane's CTO, Xiong Wei, emphasized that as AI moves into inference and engineering stages, computing power is becoming a new foundational resource, with system-level scheduling and energy efficiency management being key variables for personal computing experiences [2]