How Fast is Fast? MoSys Embedded Search and Classification Acceleration Using Xilinx FPGA and MoSys Programmable HyperSpeed Engine

By Gus Lignos
Vice President Sales, MoSys

There is a great saying from inventor and businessman Dean Kamen that goes, Every once in a while, a new technology, an old problem, and a big idea turn into an innovation.” I love this quote because it basically sums up the spirit of high tech and Silicon Valley. It also speaks quite literally to what MoSys has quietly been working on behind the scenes. We are making some major moves and to that end, we will be taking our news public, starting with demonstrating exciting new accelerator technology. We will be showcasing new packet filtering capability on a 200G PCIe Card at the upcoming Xilinx Developer Forum. What we will show is Layer 2 forwarding and packet classification firmware running on the 32 RISC cores embedded in the MoSys PHE device to perform classification of packet header data as an alternative to TCAM silicon. The search and classification solution is targeted at the growing number of 2 x 100G Ethernet systems and is ideal for SmartNICs and acceleration cards. Our interactive demonstrations will feature a TCAM compiler, classifier and a Layer 2 forwarding database running on a MoSys PHE connected to a Xilinx VU9P UltraScale+ FPGA on a PCIe card plugged into a standard server box.

So, just how fast can we accelerate? The performance improvement represents an order of magnitude over standalone FPGAs and 100x over current software solutions running on host CPUs that are bottlenecked by random accesses of memory. The demonstration will take place at the Xilinx Developer’s Forum, Oct. 1-2, at the Fairmont Hotel in San Jose, California.  You can find us at booth number 36, right outside the session rooms, so come talk to us and learn more about how we can help accelerate your applications now and in the future.