The lab supports heterogeneous processing technologies from CPU and accelerator vendors who are connecting their hardware with IDT’s portfolio of RapidIO and PCIe interconnect semiconductors, advanced timing and memory interface products. Based at IDT’s facility in Ottawa, Canada, the Open HPAC Lab runs on a contribution model. It has been initially funded by IDT and partner companies to enable end users to develop application software for analytics and high-performance computing (HPC) requiring a variety of processor types with low latency, high throughput and energy efficiency. Target applications for the lab workload include real-time analytics, deep learning, pattern recognition, video analytics and image processing.
The technology at the centre of the lab was used for analysing the content of Twitter traffic during the 2014 FIFA World Cup Finals, and was recently adopted by CERN for its Large Hadron Collider and data centre analytics. The lab is based on industry-standard IT form factor solutions that align with the Open Compute Project (OCP) High-Performance Computing initiative, which IDT co-chairs.
“As analytics of large amounts of unstructured data become more important in the hyperscale cloud data centre, the need for using accelerators such as GPUs and FPGAs in conjunction with processors is paramount, and low-latency interconnect is key to keeping up with the workload in distributed multi-processor systems,” said Sailesh Chittipeddi, IDT's chief technology officer and vice president of Global Operations. “We kicked off the Open HPAC Lab to provide an area of collaboration for those requiring the low latency and energy efficiency that RapidIO interconnect delivers.”
“Highly programmable, ultra-high performance GPU technology will transform hyperscale cloud data centres in such novel areas as real-time analytics applied at the mobile edge,” said Duncan Poole, director of Platform Alliances at NVIDIA. “We look forward to working with IDT and the Open HPAC Lab to enable customers to develop advanced, next-generation applications on top of the RapidIO heterogeneous computing platform.”
“Low-latency energy-efficient computing is an important