Financial Management Challenge

Fund and asset management relies upon rapid access to real-time information and then comprehensive analysis of its impact on individual assets, asset classes and portfolios. Point in time analysis provides invaluable insights and aids allocation decision making however it is predictive analysis and intelligent projections based on historical trends that separates good financial management from excellence. One of the biggest headaches in modern finance is the knowledge that the data needed to make decisions is out there floating in the ether but it is not in a form in which it can be readily analysed and decisions made. This is the fundamental point of professional fund management and its this 'information arbitrage' that separates relatively trivial marking of the market growth from exceptional performance. There are few other mission critical environments other than defence and health where quite literally a split second difference in decision making constitutes a significant opportunity cost.

The financial derivatives market is constantly evolving and innovating new products to meet the need to manage risk in differing environments. It's a discipline where even the slightest error in contract valuation can compound and lead to a huge amount of downside risk and potential losses. Traders rely on complex mathematical models (Black-Scholes pricing and Monte Carlo simulations for example) to arrive at the value and risk sensitivity of the contract. The variability and rapid pace nature of the financial markets makes it imperative that derivative valuations be fast and accurate. Monte Carlo models for example can simulate millions of scenarios for underlying contract variables and contextual factors such as stock and commodity prices, interest rates, bond prices etc., but ordinarily would have computation times that are exceedingly long. The sheer number of calculations and complexity of derivative contracts, and the need for both rapid model development and fast, accurate valuations, highlights some of the challenges faced by the derivatives market. Sophisticated software packages (such as SciFinance by SciComp for e.g.) leverage the power of our personal supercomputers to deliver this need faster than has ever previously been possible.

How Much Information is Enough?

We are in the information age where every country and in the not so distant future everybody is connected by several devices to the global Internet. Using the Internet as the best guide, the amount available information available now is estimated to be more than all the information put together in the previous years of the Internets existence. It's expected that information availability and growth will continue to at least double every year. Could there be information out there that could be important to you but you just don't have the resources to gather, filter and properly analyse it? In the words of the now famous quote 'What are your Unknown, Unknowns?'

Information Analysis and Decision Making

There are now some excellent software tools on the market that can leverage the power of 'personal supercomputer' class workstations to produce quite exceptional results. The multi-dimensional decision trees and massive in memory data sets require all round exceptional capability from compute resources. Many solutions have been experimented with including GRID and MESH computing but all have their bottlenecks in delivering results when they are needed. The power of the supercomputer serves one purpose and 100% of its resources are available on demand. Intel's new Nehalem platform together with nVidia Tesla provide a multi-Teraflop personal supercomputer desktop PC. Alternatively the PC can be rack mounted in server form and housed in supercomputing server farm in a traditional data centre or machine room. The high cycle speeds and multiple cores that the Intel Nehalem processor is capable of together with the memory bandwidth make it Intel's most powerful platform to date and a real step change from previous architectures. When combined with the mid-speed massive parallel processing capacity of the nVidia GPU and CUDA platform you have a platform with up to 24 main threads of execution, over 1000 parallel processing cores, over 100GB of main high speed memory and a resilient SAS RAID0/1/5/10 storage array with approaching 1GB/s bandwidth. All at the cost of a mid-range server.

Harness the power of the Personal Supercomputer

The Cryo PC desktops and workstations leverage Nehalem's tremendous muscularity to realise its full potential. The unprecedented memory bandwidth combined with at least four main cores with HyperThreading gives it record breaking performance to churn through scenarios and calculations and present conclusions in an instantly intelligible way so that rapid decisions can be made by Fund Managers, Traders and decision makers. The platform excels with the typical deep algorithms, simulations and formula used such as Monte Carlo simulations, Black-Scholes pricing, Chaos and game theory.

SciFinance is an example of a product which leverages the platform. It is a derivative model development environment that automatically generates serial C/C++ source code from concise, high-level model specifications. In order to dramatically speed up the execution time of the Monte Carlo pricing and risk models, SciComp has added a feature that automatically generates of NVIDIA? CUDA?-enabled source code (in standard C language). This automatic code generation allows critical sections of the pricing code to execute on the highly parallel architecture of the NVIDIA GPU's. To trigger the new code style, customers need simply add the keyword 'CUDA' to a model specification to produce CUDA-enabled, compiler-ready parallel code. The result: speed increases from 30X to over 100X with one NVIDIA Tesla C1060 GPU. Further increases are nearly linear with the number of GPU's due to the way such models can scale across the GPU architecture. "The speed-ups our users can achieve with CUDA and GPU's are astonishing," says Curt Randall, executive vice president of SciComp. "With parallel processing on GPU's, pricing a large portfolio of exotic contracts can be accomplished in minutes instead of hours. Because of the ease with which a financial institution can implement the NVIDIA solutions with our software, doing so is an easy choice for them."

Take a look at this Derivative Pricing and Risk Model video clip for more information.

The unique way in which our Cryo Boost process and technology unlocks an additional 50% or more power from the core of the Nehalem PC architecture further dramatically extends the performance envelope.

Simple Management and Support

The Cryo PC supercomputer is designed to live in the office as comfortably as it could in the data centre. It requires no special working environment or support. It can be maintained and managed like a standard PC it can also usually have a standard corporate server or desktop build installed on it. It's exceptionally quiet, a lot quieter than a standard PC due to its over engineered cooling solutions and has entirely standard connectivity and interfaces such as USB2, IEEE1394 Firewire, multi Gigabit Ethernet and eSATA. It uses all standard PC components to ensure its easily and cheaply upgradable and maintainable. The power, cooling and chassis of the machine will last many years of good service with only the core processor and support components requiring periodic upgrades as an when you want to benefit from further enhancements in processor, memory and storage power that have been released to market.