Gerry Fan serves as the Chief Executive Officer at XConn Technologies, a company at the forefront of innovation in next-generation interconnect technology tailored for high-performance computing and AI applications.
Established in 2020 by a team of seasoned experts in memory and processing, XConn is dedicated to making Compute Express Link™ (CXL™), an industry-endorsed Cache-Coherent Interconnect for Processors, Memory Expansion, and Accelerators, accessible to a broader market.
In pursuit of expediting the adoption of CXL, Gerry and his teams have successfully introduced the world’s inaugural hybrid CXL and PCIe switch – with a strategic approach that will make computers faster, smarter, and better for the environment.
M.R. Rangaswami: What barriers are being faced by AI and HPC applications that you are looking to
solve?
Gerry Fan: Next generation applications for artificial intelligence (AI) and high-performance computing
(HPC) continue to face memory limitations. The exponential demand these applications place on memory bandwidth has become a barrier to their further innovation and widespread adoption.
The CXL specification has been developed to alleviate this challenge by offering unprecedented memory capacity and bandwidth so that critical applications, such as research for drug discovery, climate modeling or natural language processing, can be delivered without memory constraints. By applying CXL technology to break through the memory bottleneck, XConn is helping to advance next-generation applications where a universal interface can allow CPUs, GPUs, DPUs, FPGAs and other accelerators to share memory seamlessly.
M.R.: How are you looking to solve the challenge with the industry’s first and only hybrid CXL
and PCIe switch?
Gerry: While CXL technology is poised to alleviate memory barriers in AI and HPC, a hybrid approach that combines CXL and PCIe on a single switch provides a more seamless pathway to CXL adoption. PCIe (Peripheral Component Interconnect Express) is a widely used interface for connecting hardware components, including GPUs and storage devices. Many traditional applications only need the interconnect capability offered by PCIe. Yet, emergingly, next-generation applications need the higher bandwidth enabled by CXL. System designers can be stuck with what approach will be the highest need.
XConn is meeting this challenge by offering the industry’s first and only hybrid CXL 2.0 and PCIe Gen 5 switch. Combining both interconnect technologies on a single 256-lane SoC, the XConn switch is able to offer the industry’s lowest port-to-port latency and lowest power consumption per port in a single chip – all at a low total cost of ownership. What’s more, system designers only have to design once to achieve versatile expansion, heterogeneous integration for a mix of accelerators, and fault tolerance with the redundancy mission critical applications require for true processing availability.
M.R.: In your view, how will XConn revolutionize the future of high-performance computing and AI
applications?
Gerry: Together with other leading CXL ecosystem players, XConn is delivering on CXL’s promise to support faster, more agile AI processing. This will deliver the performance gains AI and HPC applications
needs to accelerate research and innovation breakthroughs. It will also support greater energy efficiency and sustainability while helping to proliferate the “AI Everywhere” paradigm for smarter and more autonomous systems.
By helping to foster innovation and accelerate application use cases, XConn is delivering the missing link that will pave the way for unprecedented computing performance needed for tomorrow’s breakthroughs and technology advancements.
M.R. Rangaswami is the Co-Founder of Sandhill.com