Recent News

AEWIN’s small footprint rackmount edge computing server integrated with double-width GPU boosts performance and reliability – DIGITIMES

The SCB-1932C and SCB-1937C high-performance edge computing servers. Credit: AEWIN

The emergence of 5G, Artificial Intelligence (AI) and Internet of Things (IoT) technologies can unlock digital transformation to meet the needs of various industries such as manufacturing, healthcare, urban infrastructure, etc. While Cloud solutions are becoming indispensable technology, challenges around the system performance, privacy, and complexity stimulate connected technology innovation for Edge Computing. AEWIN provides innovative hardware to handle evolving edge computing needs to keep ahead of the competition. The most recent releases of SCB-1932C and SCB-1937C purpose-built servers add speed, flexibility, reliability and maximal computing power to support diverse edge computing for multi-access edge computing (MEC) and AIoT applications.

AEWIN’s solutions provide better data security, and efficient control over continuous operation in AI-based applications such as in-vehicle systems, facial recognition, and manufacturing equipment. The distance between system computing units and the edge must be as close as possible to reduce delays caused by excessive load on the upper layer computing platform and mid layer network bandwidth. However, the physical space inside edge equipment is limited, which means the dimensions of boards and components must be drastically reduced. Furthermore, cooling issues can also impact the performance of computing units. These factors have caused challenges in product development SCB-1932C and SCB-1937C addressed these issues with small footprints and double-width GPU designs.

In terms of dimensions, both the SCB-1932C and SCB-1937C are 600mm in length, a mere 3/4 of the 800mm units commonly found in other servers. The smaller footprint allows for better space utilization while also catering to expandability, thereby achieving an excellent balance between flexibility and performance. On top of that, the frontal access design has 2 key advantages:

*No need to dismantle the entire case for maintenance, which increases convenience.
*Increased product reliability due to the cooling facilitated by the frontal configurations of network ports and accelerator cards.

The double-width GPU massively increases AI performance for edge computing. Compared to the current single-width GPUs used to perform AI inference on edge equipment, double-width GPUs provide both functions of inference and training. After sufficient data has been captured through a period of operation, double-width GPUs can retrain models in real-time and continuously improve the AI accuracy of edge equipment.

Nevertheless, AEWIN has both Intel and AMD CPU options as integration benefits to match users’ selections.

*SCB-1932C integrates with 3rd gen Intel Xeon scalable processors. which has been on the market quite a while; it boasts high market share in the high-performance computing market as well as comprehensive support for peripheral technologies.
*SCB-1937C integrates with AMD EPYC 7002/7003 series processors; it promotes advanced manufacturing processes and aggressive pricing strategy.

These two major brands bring benefits to the table with different key competitive positioning. Ultimately, users’ needs will determine which CPU brand is right for them by taking the budget and applications into considerations.

SCB-1932C and SCB-1937C can be applied in MEC, networking equipment, and AIoT systems; AEWIN also provides comprehensive services to support these applications. We boast industry leading designs in system architecture, platform support, accelerator card modules, and NVMe modules focused on networking applications. The BIOS, firmware, and mechanism designs by AEWIN are customizable to meet different customers’ requirements and needs.

In an increasingly competitive market environment, availability and product launch schedule become critical factors that impact competitiveness. However, the manufacturers/customers only discover their products’ performance falls short of expectations during the development phase. This results in extended development stage of the projects or launching underperforming products. An off the shelf (OTS) solution offered by AEWIN can ease the problem by allowing customers to conduct testing prior to model and solution selection. By verifying the products in advance, products can go-to-market on schedule successfully with controlled quality. Aside from the advantages, OTS can also be combined with AEWIN’s customization services to expand into a series of products. While other OEM/ODM designed manufacturers offer value-added services, they do not necessarily accept projects with low volume and high diversity as orders must achieve a minimum quantity threshold. AEWIN provides OTS customizable firmware, hardware, and mechanisms to satisfy the customization requirements and launch schedule expectations, thereby improving the product competitiveness and taking advantage of new business opportunities in both edge computing and AI markets.

This UrIoTNews article is syndicated fromGoogle News

About Post Author