by Jack Watchmaker, with Dan Mandell | 07/09/2024

 

High-end edge servers are revolutionizing operational environments and creating substantial value for both OEMs and end users. Most notably, this includes their ever-growing capacity to process large volumes of data locally through leveraging hardware acceleration technologies. Further advances in connectivity, storage, and software development have also amplified the capabilities of modern edge server platforms. With unprecedented scalability and flexibility to support a wide range of applications and workloads, high-end edge servers enable solution builders/providers and end users to optimize their operations, improve decision making, and drive more differentiation for OT solutions and services.

One feature of high-end edge servers deployed beyond the datacenter today is that of GPU acceleration, as this technology enables advanced applications and offloads compute-intensive workloads to the GPU, enhancing overall performance. One of many examples of an edge server incorporating GPU acceleration is Supermicro Computer's IoT A+ Server (AS-1115S-FWTRT) with its NVIDIA A2 GPU acceleration support. The server’s NVIDIA’s A2 GPUs support allows for accelerated workloads that can be offloaded to its GPU, enabling advanced virtualization applications that enhance the server’s performance as well as integrated AI/ML applications. Additionally, the NVIDIA’s A2 GPU also includes Tensor Cores supporting int4 operations, second-generation RT Cores for ray tracing, and dedicated hardware encoding and decoding engines. Coupled with this server’s sophisticated AMD EPYC 8004 Series processors, these capabilities make it well suited for use cases such as virtualization, firewall applications, edge/cloud computing services, CDN/vCDN/Cloud CDN, 5G/Telco/NEBS environment, and vRAN/o-RAN.

Types of Computer system Components Purchased: Workload Accelerator Cards

(Percentage of Respondents) Note: 2024 Voice of the Engineer Data Available Soon!

Another high-end server benefiting from its hardware acceleration is HPE’s Edgeline EL8000 server, with its Intel N3000v FPGA and NVIDIA Tesla T4 support under its e910t blade. Particularly, the Intel N3000v FPGA enables network function virtualization, multi-access edge computing, and cyber security acceleration applications. Specifically, it leverages Intel's Arria 10 GT FPGA with hardened floating point arithmetic blocks delivering up to 1.5 TFLOPS of performance to optimize workload acceleration. The FPGA integrates programmable logic, embedded processors, transceivers, and memory controllers on a single chip, thereby enhancing its versatility, efficiency, and overall performance. Due to its powerful FPGA-driven GPU acceleration, which is ideal for data plane and control plane offload, and its ruggedized design, this server supports numerous use cases in the public sector, retail, industrial manufacturing, energy, transportation, and video analytics.

Lenovo’s ThinkEdge SE360 V2 is also setting a new standard for servers in the edge computing space. This server offers compatibility with various hardware accelerators. However, its support for the Qualcomm Cloud AI 100 provides the gateway to support multiple generative AI large language models. The Qualcomm Cloud AI 100 includes an AI Chatbot driven by on-premises enterprise data, which features retrieval augmented generation via 4 simultaneous LLM workloads. This Qualcomm accelerator also provides unique AI capabilities including text-to-image generation, multi-stream simultaneous transcription, and language translation. Accordingly, its primary use cases include process automation, video surveillance, virtual desktop computer, and tracking capabilities. Notably, this server is seeing traction among retailers, with its AI applications assisting in rapid store-side data processing, identifying organized retail crime activity, optimizing labor, generating direct cost savings, and consumer behavioral insights not available through other technologies.

Edge server platforms are becoming more intelligent, needing to support AI/ML, vision, security, and other edge computing applications. As a result of the industry’s rapidly increasing competitiveness, suppliers of edge servers are required to provide robust computing power, advanced connectivity, and integration with powerful but efficient hardware accelerators.

Learn more about VDC’s studies investigating embedded hardware acceleration trends, such as Embedded & Edge AI Hardware and Embedded Boards, Modules & Systems by downloading the IoT & Embedded Technology 2024 Roadmap.