Samsung Foundry to Announce HPC Data Center SF4X Process Technology 2023

Representatives from Samsung Foundry will present “Highly Reliable/Manufacturable 4nm FinFET Platform Technology (SF4X) for HPC Application with Dual-CPP/HP-HD Standard Cells” at the 2023 Symposium on VLSI Technology and Circuits.

In spite of the fact that Samsung’s current and forthcoming nodes will not be able to compete with TSMC’s, the company anticipates surpassing the Taiwanese behemoth within five years.

In the coming years, Samsung Foundry’s novel SF4X will vie with rival foundry TSMC’s N4P/N4X node technology. Due to the absence of early performance scores, it is uncertain which process technology will benefit HPC clients in terms of performance, cost, and efficiency.

Samsung Foundry will unveil its HPC SF4X process technology for data center CPUs and GPUs at this year’s Symposium on VLSI Technology and Circuits.

The company asserts, however, that the new SF4X will offer superior power and performance for HPC data center applications.

Given that HPC is a prevalent topic in an industry that requires high-performance computing levels, energy, and other factors, the company hopes that its vast clientele will implement the new process technology. Companies such as AMD, Intel, and NVIDIA, as well as Ampere, Graphcore, and IBM, require HPC-based process nodes at present.

Datacenter processors and graphics processing units (GPUs) are in high demand for high-performance computing applications that require the processing of large quantities of computation. Artificial intelligence and machine learning are synonymous when discussing HPC applications.

To meet the requirements of their consumers, Samsung and TSMC endeavor to develop new and more intricate fabrication processes. Samsung is presenting a technology that will provide “significant performance +10% boosting with power -23% reduction” and will be in high demand because it has the potential to reduce administrative expenses and environmental risks.

Leave a Reply