ASPEED Software has announced that its parallel
distributed computing solution, ACCELLERANT, has successfully enabled fixed
income and credit derivative trading applications to perform on the IBM®
Blue Gene® System. With ASPEED’s support for Blue Gene, financial
applications can now rapidly and seamlessly exploit the scalability and
economics of the world’s fastest supercomputer.
ACCELLERANT now becomes the first solution to enable computational intensive
financial services applications to leverage the immense operational and
super-charged performance benefits of Blue Gene. Using ACCELLERANT, traders,
quants and other decision-makers can take their financial application
capabilities to new heights with analysis flexibility, in terms of increased
precision, volumes and considerations in less time. ACCELLERANT
transparently injects distributed parallel processing capabilities into
existing applications without re-engineering the underlying algorithms and
risking the integrity of results.
Blue Gene, IBM's cutting-edge economic supercomputing platform, propels
qualitative analysis to the next level. It is the only system commercially
available optimized for bandwidth, scalability and computational
performance, while consuming a fraction of the electricity and floor space
demanded by most high-performance systems. Proven in deployments for
leading-edge research in hydrodynamics, astronomy and space research,
quantum chemistry and climate modeling, Blue Gene is now being adopted,
through ACCELLERANT, for mission-critical processes such as risk
management, securities and portfolio pricing and hedging strategies.
In a recent benchmark, ASPEED demonstrated the ease of adaptation of
financial applications to Blue Gene and the potential performance impact to
trading applications. It took a single thread American Options portfolio
pricing application (AMBook), adapted it for the parallel environment and
ran it showing the reduced processing time. In a test case, a 240,000-option
portfolio pricing run was reduced from an hour to seconds using Blue Gene.
Variants of AMBook are generally run to determine the overall price and risk
of an options portfolio. The calculation examines each instrument. Each
calculation is stochastic – nondeterministic in some manner - usually by
using random numbers and therefore requires significant compute power.
There are two primary considerations for such runs, time and precision.
Ideally traders prefer more precision, usually characterized as time steps
or paths, but they often have to limit the analysis because of the overall
elapsed time to price. Blue Gene with ACCELLERANT enables any trader to
select more precision and variables while being more responsive with their
existing applications.
“Financial services companies stand to gain significant cost, competitive
and time advantages through using advanced software accelerators on their
critical and most computationally intense applications,” said Dave Turek,
vice president of Deep Computing at IBM. “The combination of Blue Gene and
ACCELLERANT in this performance result is indicative of the emerging
opportunity clients have to leverage the Blue Gene platform even further in
financial services applications.”
This is the latest enhancement for ACCELLERANT, which is widely used by
leading financial firms, investment banks, pharmaceutical drug developers,
life sciences firms, energy providers and the Department of Defense to
optimize application performance within grid, cluster, multi-core and other
distributed computing enviornments.