For Release: April 17, 2008
HPC Platform has been working since 2005 on ultra-low latency market
data processing, for feed-handling and algorithm-trading, as well as complex financial
products valuation acceleration.
Working closely with Altera, HPC Platform has designed a unique FPGA
based hardware system optimized for these applications.
Now, thanks to work conducted with leading investment banks, HPC
Platform can announce that its systems, using full hardware processing, has
achieved an outstanding performance in market data processing with a
latency of 1.1 to 2.0 microseconds (see below).
For more information please contact marc.battyani@hpcplatform.com
For Release: December 17, 2007
HPC Platform has been working since 2005 on ultra-low latency market
data processing, for feed-handling and algorithm-trading, as well as complex financial
products valuation acceleration.
Working closely with Altera, HPC Platform has designed a unique FPGA
based hardware system optimized for these applications.
On a benchmark conducted with leading investment banks, an HPC
Platform system has been able to compute, in double precision, the price of a basket
option composed of 3 underlyings at the record speed of 1400 Millions
simulation paths/s. For instance the computation takes 715
microseconds for 1M simulation paths.
This performance results from the association of a dedicated hardware with specialized compilers and libraries.
For more information please contact marc.battyani@hpcplatform.com
For Release: November 20, 2007
HPC Platform has been working since 2005 on ultra-low latency market
data processing, for feed-handling and algorithm-trading, as well as complex financial
products valuation acceleration.
Working closely with Altera, HPC Platform has designed a unique FPGA
based hardware system optimized for these applications.
On a benchmark conducted with leading investment banks, an HPC
Platform system has been able to compute, in double precision, the price of an european
option at the record speed of 4000 Millions simulation paths/s. For instance the computation takes 250 microseconds for 1M
simulation paths.
This performance results from the association of a dedicated hardware with specialized compilers and libraries.
For more information please contact marc.battyani@hpcplatform.com