News Overview
-
Product Launch: One Stop Systems (OSS) has introduced the Ponto system, a PCIe Gen5 expansion platform supporting up to 16 dual-width GPUs or accelerator cards, each with a power capacity of up to 675 W.
-
High-Power Density: The system delivers an unprecedented 11 kW of power to its expansion slots, equating to 1.83 kW per rack unit, setting a new standard in power density for such systems.
Original article link: One Stop Systems introduces ultra-dense 16-way GPU expansion system for AI/ML/edge computing
In-Depth Analysis
System Specifications
-
Expansion Capacity: The Ponto system accommodates up to 16 dual-width add-in cards, each supporting up to 675 W, such as the 600 W H200 NVL. Alternatively, it can be configured to support 32 single-width add-in cards, offering flexibility based on computational needs.
-
Cooling Mechanism: To manage the substantial heat output, the system employs high-voltage fans, specialized ducting, and dynamic fan controls, ensuring optimal thermal management and consistent performance.
Management Features
- Unified Baseboard Management Controller (U-BMC): OSS’s U-BMC provides advanced system monitoring, integrating seamlessly with the host server’s management interface. It offers features like dynamic fan controls, GPU power throttling, and telemetry for add-in cards, enhancing operational efficiency and reliability.
Target Applications
Designed for data center environments that demand high-density compute performance, the Ponto system is particularly suited for high-data-throughput applications. Its architecture aligns well with PCIe expansion-focused composable infrastructure, making it ideal for AI, machine learning, and sensor processing tasks at the edge.
Commentary
The introduction of the Ponto system by One Stop Systems represents a significant advancement in GPU expansion technology. By delivering an unparalleled power density and accommodating the latest high-wattage GPUs, OSS addresses the escalating computational demands of modern data centers. The system’s robust cooling and management features ensure reliability and efficiency, which are critical for AI and machine learning applications. This innovation positions OSS competitively in the market, catering to organizations seeking scalable and efficient solutions for edge computing challenges.