2001, when UnRisk started, quant finance practitioners in banks utilized almost exclusively Wintel platforms. Consequently, UnRisk was optimized for Wintel.
But now, we use MacBook Pro with 8 cores and an AMD GPU for presentations and workshops. Our customers utilize, say, DELL servers with, say, 128 CPUs and four NVIDIA Tesla C 20 series GPU cards. My iPad is a powerful little machine, as well as Android based tablets..
More than a year ago, we decided to overhaul the UnRisk engines code base to make it compatible with other target operating systems than Windows and to support modern processor architectures and utilize the latest computing muscles. See The UnRisk Agenda 2011
The first outcome of this refactoring effort will be a Linux and OS X version - UnRisk-Q version 5.1 to be released quite soon. But our spade-work embraces the radical redesign of core algorithms to be implemented in OpenCL. Yes, this is a step back into programming in the small, but the benefit of platform-independence pays back.
This blazingly fast one-for-all code will run on current and future technologies. Transparent to users it will utilize the power of the CPU cores, combined with massive parallel architectures. The level of parallelization granularity is defined by the architecture. OpenCL code is driven over the grid.
Double benefit from a high level of platform independence
Quant developers, UnRisk-Q users, enjoy the double benefit of the blazingly fast engines that can be programmatically manipulated in a declarative financial language framework.
There are two great principles in UnRisk's platform independence. The first is that the messy plumbing of the utilization of hybrid platforms is automated. The second is that most that can be done in UnRisk-Q can be done optimized without change - from a small device to a large HPC farm.
