Overview

A data center is a facility established to store, manage, process and transmit large amounts of data. It is an important entity for large-scale high-performance computing and massive data storage. It is usually composed of servers, network equipment, storage, cooling systems and other facilities which provide users with cloud computing, big data, artificial intelligence and other services through high-speed network connections. As a foundation of digital and intelligent transformation, the construction of data centers has spread all over the world.Key specifications for measuring data centers include large capacity, large bandwidth, large computing power, low latency, high energy efficiency, high reliability, etc. With the development of semiconductor and communication network technologies, data centers are gradually moving from core areas to edge nodes. The computing, storage and other resource services provided will be much closer to user needs, and the operation and management of the data center itself will become more intelligent, efficient and energy-saving.

Background

The three core modules of the data center, like computing, storage and network, respectively put forward higher requirements for computing performance, capacity and speed. Due to the slowdown of Moore's Law in computing chips such as CPUs and GPUs, advanced processes are gradually approaching physical limits and the economic benefits are getting lower. It is impossible to significantly improve computing chip performance, power, area and other indicators through traditional SoC architecture. Meanwhile, chiplets based on advanced packaging technology became one of the important technologies to break through bottlenecks. Chiplet architecture shows obvious advantages, such as smaller size, higher yield, and flexible reuse of chiplets at different process nodes, which greatly shortened the time to market. Many leading companies have successfully implemented HPC chips based on Chiplet technology. Storage such as DRAM, NAND and NOR are facing upgrades. High-performance HBM based on 3D stacking technology has moved from HBM2 to HBM3\3E, and has evolved towards large bandwidth and low latency which has a significant impact on the signal\power integrity design of data transmission links. Increasingly demanding. The data processing capacity and forwarding rate of network switches\routers continue to increase for massive applications, but the signal integrity of the backplane high-speed interconnection structure and channels affects the data transmission of the data center. In addition, large-scale equipment in data centers operates with serious heat dissipation, which affects the overall data center energy efficiency. Therefore, how to consider thermal issues from the device and equipment  is an important topic for current research and development.

Solution

For the research and development of computing chips\servers, storage chips\equipment, network switching chips\equipment and energy management chips for data centers, Xpeedic provides full-stack EDA solutions to support high-performance computing, storage, network and energy products. Through the EDA design and simulation platform, we jointly realize high-performance computing & storage to promote energy conservation and emission reduction of infrastructure.