âACTIV solutions already drive many of the marketâs leading participants including specialists, market makers, hedge funds, algorithmic-driven trading firms and exchanges,â said Frank Piasecki, President of ACTIV Financial. âACTIV feed handlers are amongst the fastest available in the industry. Our new silicon based solution will take us to the next level in terms of both latency and throughput to enable us to deliver greater business value to our clients.â
ACTIV Financial is porting its highly successful market data ticker plant solution directly onto silicon using an FPGA co-processor and AMD64 technology with Direct Connect Architecture and HyperTransportâ¢ technology. âWe have been tracking this technology in our labs since we started the company and architected our software systems to enable hardware acceleration,â noted Mike Dunne, CTO of ACTIV Financial. âOur efforts have led to a system that not only offers key latency and TCO benefits, but will extend the world-class reliability and advanced operations management features of ACTIVâs systems. AMD64 technology enables us to efficiently integrate both the traditional AMD Opteronâ¢ processor as well as a co-processor card which plugs directly into the CPU socket. The performance improvements for processing market data are astounding.â
âACTIV Financial continues to provide innovative solutions to financial services customers with this advanced trading technology,â said Doug OâFlaherty, division manager for acceleration strategy, AMD (NYSE:AMD). âThis is an excellent example of the achievements possible through AMDâs Torrenza initiative, which leverages the low-latency and high-bandwidth benefits of HyperTransportâ¢ technology to allow close coupling of co-processors and AMD64 technology.â
ACTIV was founded with the product strategy of offering the lowest possible systemic latency to institutions with high time-value priority, in an environment of skyrocketing market data rates. This performance will be critical to addressing the OPRA message rates that are projected to double from todayâs rates to 716,000 messages per second by January 2008.
Mike Dunne, CTO, said, âThe amount of general purpose hardware needed to handle projected message rates for top-of-range systems is growing. The new MPU approach could potentially reduce an 8 to 12 server installation into a single hardware-accelerated device.â
Dunne continued, âThe key issue for most of our clients is latency. Our software systems were among the first to break the millisecond barrier. We are now looking at order-of-magnitude improvement on that performance metric, and that is just a starting point.â