Thoughts on Windows HPC Server 2008 and F# Cashflows
There doesn’t seem to be a lot around on the web regards usage of HPC from a trading perspective. I know from previous projects I’ve been involved in that DataSynapse and Platform are the main grid platforms on the sell-side.
Why am I interested in HPC? There are a number of reasons for my HPC interest.
- I’ve only recently managed to get my hands on enough hardware to build a cluster so know is as good a time as any to see what Windows HPC offers
- Dryad has been of interest to me for some time. Which hence leads to the question of if it offers any “interesting” solutions in finance
- Given I have a Market Strategy Engine, and a Real-Time Market Risk Cube (RTC), the obvious piece joining the two Proof Of Concepts (POC) is a grid that leverage the F# trade cashflow/PV generation – previously blogged
Looking at the main Windows HPC Server 2008 resource kit site, it would appear that one possible avenue is to look at leveraging the service-oriented architecture (SOA) programming model for high-performance computing (HPC) and hence building a Market Risk SOA. Looking at this sample it looks like I need the equivalent of EchoSvc to host my F# cashflow code and then a client that listens to StreamInsight, and when a trade return from the Market (Market Strategy Engine code) submits the trade to the Market Risk SOA.
More interestingly, it might be possible to “best guess” the trade PV number in the StreamInsight pipeline, feed that into the RTC to provide an initial value, and in parallel submit the trade the HPC which once complete would override the initial RTC PV trade number.
The things get more and more interesting. Have you considered adding a tag for this series of posts?