- Compal Electronics introduces high-performance server platform SG720-2A/OG720-2A at AMD Advancing AI 2025 and ISC 2025.
- The platform features AMD Instinct MI355X GPU architecture with dual cooling configurations for enhanced efficiency.
- Designed for generative AI and large language model training, it supports major AI frameworks and rack standards.
Compal Electronics (2324.TW), a global leader in IT and computing solutions, has unveiled its latest high-performance server platform, the SG720-2A/OG720-2A, at AMD Advancing AI 2025 in the U.S. and the International Supercomputing Conference (ISC) 2025 in Europe. The server platform features the AMD Instinct MI355X GPU architecture and offers both single-phase and two-phase liquid cooling configurations, highlighting Compal's innovation in thermal management and system integration.
Specifically designed for next-generation generative AI and large language model (LLM) training, the SG720-2A/OG720-2A offers substantial flexibility and scalability, capturing widespread interest within the industry. Key features include support for up to eight AMD Instinct MI350 Series GPUs, a dual cooling architecture optimized for high thermal density workloads, and advanced architecture with 288GB HBM3E memory and 8TB/s bandwidth.
The platform is equipped with PCIe Gen5 connectivity and AMD Infinity Fabric, enabling high-speed interconnect performance for multi-GPU orchestration and reducing latency to boost AI inference efficiency. It is fully compatible with mainstream open-source AI stacks such as ROCm, PyTorch, and TensorFlow, facilitating streamlined AI model integration and accelerated time-to-market.
With support for EIA 19" and ORv3 21" rack standards and a modular design, Compal's SG720-2A/OG720-2A allows for simplified upgrades and maintenance, making it ideal for diverse data center environments. The series' debut at Advancing AI 2025 and ISC 2025 marks Compal's commitment to expanding its global presence and partnerships in the AI and HPC domains.
