Celoxica and Intel Collaborate on PCI Express Technology
Celoxica Holdings, a provider of low latency trading solutions, has announced that as a result of customer demand for its hardware-accelerated market data products on the latest server platforms, it is working with Intel to produce a PCI Express implementation of its Accelerator Series.
The collaboration will see Celoxica’s acceleration technology implemented in the Intel fasterLAB facility, which targets simulation of financial services industry (FSI) and high performance computing (HPC) workloads. The new platform has a high-speed 8-lane PCI Express interface to the latest generation of Xilinx FPGAs with 4Gb ethernet SFPs, either copper or fiber, connected directly to the FPGA to give the lowest latency acceleration of financial protocol handling.
The Celoxica Accelerator for PCI Express will be available on Intel platforms from October 2008, with the first supported high volume feeds being OPRA, ISE, NASDAQ ITCH, CME, ICE, NYSE Arca and Eurex. Recent tests for the Accelerator have proven latency of less than 10 microseconds, processing market data from ISE, OPRA and ITCH at over three million messages per second. A further seven feeds will be available in the first quarter of 2009, said Celoxica.
Intel fasterLAB is the engineering capability applied to the focus on FSI. Centred on the facilities in London, UK, the lab gives access to Intel engineering and technology plus major investments by partnering companies to enable simulation and measurement of FSI workloads – from infrastructure, risk, trading, market data and latency measurement. fasterLAB has access to Intel’s worldwide lab and engineering resource and is available to clients and partners.
Lee Staines, CEO of Celoxica, said: “Our collaboration with Intel has been strengthened since the Celoxica re-launch into financial services at the beginning of the year. By working with Intel and with the client feedback from our early adopter programs, we have been able to ensure the product road map would provide scale for our line handler solution without sacrificing speed, consistency or reliability.”