How artificial intelligence is reshaping edge computing

How much computing power is needed at the edge? How much memory and storage is enough for edge AI? As AI opens the door to innovative applications that require more and faster processing, storage, and memory, the minimum requirements are growing. How do today's memory and storage technologies meet the stringent requirements of these challenging new edge applications?

What do we mean by "edge"

The edge includes any distributed application where specific processing occurs outside of the server, even though the data ultimately is sent to the data center. Its main idea is to avoid sending all data over the Internet to a server for processing, but instead allows the data to be processed closer to the collection location, avoiding the latency issues caused by long data round-trips and achieving near real-time On-site response.

Edges are roughly divided based on the distance from the server to the endpoint. The so-called near edge can include applications close to the data center, perhaps even within the same building. In applications such as self-driving cars, this trend goes to the other extreme. The overlapping feature is that edge systems process data that is traditionally sent to the data center, which has practical applications in many industries.

Data Latency and Bandwidth at the Industrial Edge

In industrial applications, edge computers are often designed to receive input from sensors or other devices, and perform corresponding operations on the input. For example, preventive maintenance takes acoustic, vibration, temperature or pressure sensor readings and analyzes them to identify anomalies that indicate minor machine malfunctions.

Machines can be taken offline immediately, or if needed, for maintenance before a catastrophic failure occurs. Response times must be fast but data volumes low. However, artificial intelligence is putting pressure on these edge systems.

The impact of artificial intelligence on edge processing load

Artificial intelligence brings different loads to computer systems. AI workloads require faster processors, more memory and powerful GPUs. For example, AOI has been widely used in PCB inspection, using video input from high-speed cameras to identify missing components and quality defects. In fact, similar visual inspection technology is being widely used in different industries such as agriculture, where it can be used to identify defects and discoloration in products.

Executing complex algorithms on video input requires the parallel processing capabilities of power-hungry GPU cards, more memory for efficient and accurate AI inference, and more storage space for efficient and accurate AI inference. for additional data, but this already exists in the data center.

Bringing the power of the data center to the edge

Essentially, in order to process AI tasks at the edge, we are bridging the edge and gaps between data centers. Servers hidden away in temperature-controlled data centers have terabytes of memory and massive storage to handle specific high-volume loads and keep systems working quickly.

But when it comes to inference far away from the data center, the situation is different. Edge computers don't like this idyllic environment and must be able to withstand harsh environments. The edge requires hardware that strives for maximum performance while accounting for less-than-ideal conditions.

Edge Hardware

Adding artificial intelligence to the industrial edge requires hardware suitable for the task. An industrial computer that can handle extreme temperatures, vibrations, and space constraints is a must. In particular, three things are needed for a vision system, which is by far the most prolific AI application, memory to support efficient AI inference, storage for input data, and PoE to support adding cameras.

The latest DDR5 can get more memory in a smaller space. It delivers higher memory capacity at the edge, with twice the speed and four times the capacity of DDR4, making more efficient use of available space and resources in the same footprint.

Edge applications need to expand capacity because data must reach the server or stay at the edge for a period of time, so SSDs are needed as temporary storage. The shift from SATA to NVMe has opened the door to higher speeds and performance, and the upcoming NVMe PCIe G4X4 SSD is the latest SSD in Cervoz's product line, delivering industrial performance for these applications.

Visual systems require cameras. PoE is the simplest and most efficient way to add high-speed cameras to a system, providing power and data transmission over a single cable. Cervoz's PoE Ethernet modular PCIe expansion card adds this functionality with a small power plug-in.

Get a head start on artificial intelligence at the edge

For businesses looking to gain an edge, industrial computers coupled with industrial-grade memory and The combination of storage provides the reliability to withstand harsh edge environments and the capabilities needed to enable next-generation AI technologies at the edge of the network.

The above is the detailed content of How artificial intelligence is reshaping edge computing. For more information, please follow other related articles on the PHP Chinese website!