Exploration of Parallelization Frameworks for Computational Finance
Exploratory Systems Lab, IBM Systems and Technology Group, Poughkeepsie, NY, USA
The 2012 International Conference on Parallel and Distributed Processing Techniques and Applications (PDPTA’12), 2012
This paper presents a comparison of parallelization frameworks for efficient execution of computational finance workloads. We use a Value-at-Risk (VaR) workload to evaluate OpenCL and OpenMP parallelization frameworks on multi-core CPUs as opposed to GPUs. In addition, we study the impact of SMT on performance using GCC (4.4) and IBM XLC (11.01) compilers for both single-precision and double-precision codes. We use an 8-core, 4-way SMT IBM Power7 with Linux (RHEL 6.0, 2.6.32 kernel) to evaluate OpenCL and OpenMP. Using the IBM XLC compiler, 2-way SMT is able to provide over 30% average improvement as compared to 1 SMT thread per core, whereas, 4-way SMT is able to provide over 50% average improvement as compared to 1 SMT thread per core.
September 22, 2012 by hgpu
