覆盖具身智能训练全流程的解决方案

Search documents
紫光股份:公司X10000系列分布式存储可应对仿真数据、动捕数据的大规模存储需求
Ge Long Hui· 2025-08-04 07:59
Core Viewpoint - The company has launched a comprehensive solution for embodied intelligence training, leveraging its expertise in storage, computing power, and networking to provide high-performance ICT infrastructure for the implementation of embodied intelligence [1] Storage Solutions - In the data storage phase, the company's X10000 series distributed storage can handle large-scale storage needs for simulation and motion capture data, ensuring that foundational data is both stored securely and reliably [1] - During the training phase, the X20000 series high-performance storage features high throughput and low latency, enabling rapid read and write operations for 1 billion frames of motion data, allowing models to efficiently access quality data for learning [1] Computing Power Support - The company integrates dispersed computing resources through high-performance networking equipment and diverse heterogeneous computing servers, forming a stable and reliable computing cluster that not only provides massive computing power but also avoids resource waste through efficient scheduling [1]
紫光股份(000938.SZ):公司X10000系列分布式存储可应对仿真数据、动捕数据的大规模存储需求
Ge Long Hui· 2025-08-04 07:08
Core Viewpoint - The company has launched a comprehensive solution for embodied intelligence training, leveraging its expertise in storage, computing power, and networking to provide high-performance ICT infrastructure for the implementation of embodied intelligence [1] Group 1: Data Storage - The company’s X10000 series distributed storage addresses the large-scale storage needs for simulation and motion capture data, ensuring that foundational data can be stored reliably and securely [1] - During the training phase, the X20000 series high-performance storage enables rapid read and write of 1 billion frames of motion data, allowing models to efficiently access high-quality data for learning [1] Group 2: Computing Power Support - The company integrates dispersed computing resources through high-performance networking equipment and diverse heterogeneous computing servers, forming a stable and reliable computing cluster [1] - This setup not only provides massive computing power but also optimizes resource scheduling to prevent waste [1]