Industry Expert Blogs

In a world where AI chatbots answer complex queries in seconds and high-definition video streams from our smartphones, data centers are providing the essential backbone. With accelerating demands for bandwidth, data centers are undergoing a transformation, trending toward disaggregated architectures and workloads running on accelerators. Making these approaches possible is Compute Express Link™ (CXL™), the CPU-to-device, cache-coherent interconnect for processors, memory expansion, and AI accelerators.

XConn Technologies is well versed in the critical role that CXL plays in enabling next-generation AI and high-performance computing (HPC) applications. Founded in 2020 in San Jose, California, XConn’s mission is to accelerate AI computing in data centers and HPC via its high-performance, power-efficient, scalable, and cost-effective interconnect solutions.

Recently, XConn achieved first-pass silicon success for its XC50256 CXL 2.0 data center switch SoC using Synopsys CXL 2.0 Controller and PCI Express 5.0 PHY IP on a FinFET process. By implementing CXL, with its built-in support for PCIe in a complex switch, the company is looking to provide a more seamless avenue to CXL adoption. Read on to learn how XConn developed its industry-first CXL switch SoC for data center and memory pooling applications, delivering the highest throughput, lowest latency, and lowest power.

Click here to read more ...