GPU based sparse grid technique for solving multidimensional options pricing PDEs

Gaikwad, Abhijeet; Toke, Ioane Muni

doi:10.1145/1645413.1645419

Cited by 29 publications

(18 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…A Graphics Processing Unit (GPU) is a common component found in most desktop computers, but only recently has the potential of these devices for general computation been realized, in a range of disciplines from sciences to finance (Gaikwad & Toke 2009;Sainio 2010;Isborn et al 2011;Mashimo et al 2013). By utilizing the large scale parallelization ability of GPUs, increased performance can be achieved for some computational calculations if correctly incorporated.…”

Section: Gpu Architecturementioning

confidence: 99%

Microlensing observations rapid search for exoplanets: morse code for GPUs

McDougall

Albrow

2015

Mon. Not. R. Astron. Soc.

View full text Add to dashboard Cite

The rapid analysis of ongoing gravitational microlensing events has been integral to the successful detection and characterisation of cool planets orbiting low mass stars in the Galaxy. In this paper we present an implementation of search and fit techniques on Graphical Processing Unit hardware. The method allows for the rapid identification of candidate planetary microlensing events and their subsequent followup for detailed characterisation.

show abstract

Section: Gpu Architecturementioning

confidence: 99%

Microlensing observations rapid search for exoplanets: morse code for GPUs

McDougall

Albrow

2015

Mon. Not. R. Astron. Soc.

View full text Add to dashboard Cite

show abstract

“…As LAMMPS, GROMACS is a molecular dynamics simulator, but this application does not need GPU acceleration 4 . This multi-thread and multi-process application that contributes a higher degree of heterogeneity to our experimental workload.…”

Section: ) Gromacsmentioning

confidence: 99%

“…In addition to the favorable performance/cost ratio of GPUs, this evolution has been further stimulated by considerable advances in GPU programmability, with the introduction of frameworks such as CUDA [1], OpenCL [2] and OpenACC [3]. As a result, GPU computing (also known as GPGPU) is nowadays successfully exploited in areas as diverse as finance [4], chemical physics [5], computational algebra [6], health-care equipment [7], computational fluid dynamics [8], and image analysis [9], among others.…”

Section: Introductionmentioning

confidence: 99%

SLURM Support for Remote GPU Virtualization: Implementation and Performance Study

Iserte

Castelló

Mayo

et al. 2014

2014 IEEE 26th International Symposium on Computer Architecture and High Performance Computing

View full text Add to dashboard Cite

“…These interpolations take up 99% of the 80 computation time needed to solve the equation system. As they have a high arithmetic intensity, i.e., many arithmetic operations are performed for each byte of memory transfer and access, they are perfectly suited for GPUs [31,39,21]. We therefore offload parts of the interpolation tasks from the compute nodes to their attached accelerators,…”

mentioning

confidence: 99%

Scalable high-dimensional dynamic stochastic economic modeling

Brumm

Mikushin

Scheidegger

et al. 2015

Journal of Computational Science

View full text Add to dashboard Cite

We present a highly parallelizable and flexible computational method to solve high-dimensional stochastic dynamic economic models. Solving such models often requires the use of iterative methods, like time iteration or dynamic programming. By exploiting the generic iterative structure of this broad class of economic problems, we propose a parallelization scheme that favors hybrid massively parallel computer architectures. Within a parallel nonlinear time iteration framework, we interpolate policy functions partially on GPUs using an adaptive sparse grid algorithm with piecewise linear hierarchical basis functions. GPUs accelerate this part of the computation one order of magnitude thus reducing overall computation time by 50%. The developments in this paper include the use of a fully adaptive sparse grid algorithm and the use of a mixed MPI-Intel TBB-CUDA/Thrust implementation to improve the interprocess communication strategy on massively parallel architectures. Numerical experiments on "Piz Daint" (Cray XC30) at the Swiss National Supercomputing Centre show that high-dimensional international real business cycle models can be efficiently solved in parallel. To our knowledge, this performance on a massively parallel petascale architecture for such nonlinear high-dimensional economic models has not been possible prior to present work. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. AbstractWe present a highly parallelizable and flexible computational method to solve highdimensional stochastic dynamic economic models. Solving such models often requires the use of iterative methods, like time iteration or dynamic programming. By exploiting the generic iterative structure of this broad class of economic problems, we propose a parallelization scheme that favors hybrid massively parallel computer architectures.Within a parallel nonlinear time iteration framework, we interpolate policy functions partially on GPUs using an adaptive sparse grid algorithm with piecewise linear hierarchical basis functions. GPUs accelerate this part of the computation one order of magnitude thus reducing overall computation time by 50%. The developments in this paper include the use of a fully adaptive sparse grid algorithm and the use of a mixed MPI-Intel TBB-CUDA/Thrust implementation to improve the interprocess communication strategy on massively parallel architectures. Numerical experiments on "Piz Daint" (Cray XC30) at the Swiss National Supercomputing Centre show that highdimensional international real business cycle models can be efficiently solved in parallel. To our knowledge, this performance on a massively ...

show abstract

GPU based sparse grid technique for solving multidimensional options pricing PDEs

Cited by 29 publications

References 19 publications

Microlensing observations rapid search for exoplanets: morse code for GPUs

Microlensing observations rapid search for exoplanets: morse code for GPUs

SLURM Support for Remote GPU Virtualization: Implementation and Performance Study

Scalable high-dimensional dynamic stochastic economic modeling

Contact Info

Product

Resources

About