About Globalsilicon

Globalsilicon Technologies is dedicated to rapid delivery of cutting-edge ASIC solutions to There customers. They have a proven track record of first silicon successes at the 16nm, 12nm, and 7nm nodes. There total engineering and business solution strategy ensures that customers receive quality SoC product solutions on-time or ahead of schedule.

Globalsilicon builds an internal IT environment in its own IDC, and designs and verifies related products with the support of local servers. However, the rapid increase in quantity of the transistors and customer volume has caused the local IT infrastructure to encounter serious bottlenecks, which are as follows:


  • the storage capacity of a single project may reach more than 500TB. Local storage mainly depends on SAN/NAS storage, which is expensive to expand.
  • In the process of chip simulation, hundreds of computing nodes read and write NAS at the same time, which requires extremely high disk IO performance and network throughput performance.
  • Data backup relies on third-party tools, it has high cost, poor flexibility and complicated to manage.
  • At the peak of business, due to resource exhaustion, partial design and verification, data backup can only wait in line, which leads to customer complaints and loss.

Solutions & Architecture

To meet the challenges of basic IT architecture that customers are currently facing, Wangsu suggests that customers using the unlimited computing power and storage resources on Amazon Web Services (AWS) to shorten the design and delivery cycle of chips.

The local computer room is connected with AWS Ningxia node through AWS Direct Connect, which realizes the dynamic scheduling of resources between the local computer room and AWS. This scheduling method of obtaining computing and storage resources on demand can not only improve the design efficiency, but also optimize the cost.

Outcomes of Project

  1. 1. After adopting the hybrid networking mode, Alchips began to try to take orders on AWS that exceeded the workload of on-premises. Every month, it will increase the company's income by RMB 2 million to RMB 3 million
  2. 2. Compared with the local DC expansion computing and storage capacity, EDA projects in AWS can save more than 60% of the cost
  3. 3. Only one maintenance engineer is added on AWS. However, The same scale in the local requires 4 engineers
  4. 4. Compared with on-premises, the delivery time has been shortened from few weeks to less than one hour
  5. 5. The EDA physical simulation stage will take up a lot of network throughput, which is limited by the local 10 Gigabit network, and some projects cannot be expanded to a larger cluster scale. With AWS can easily scale to 25Gbps