Many-core GPU computing with NVIDIA CUDA
Proceedings of the 22nd annual international conference on Supercomputing
High performance discrete Fourier transforms on graphics processors
Proceedings of the 2008 ACM/IEEE conference on Supercomputing
MATLAB®: a language for parallel computing
International Journal of Parallel Programming
Hi-index | 0.00 |
Several important laser-based medical treatments rest on the crucial knowledge of the response of tissues to laser penetration. Optical properties are often localised and are measured using optically active fluorescent microspheres injected into the tissue. However, the measurement process combines the tissue properties with the optical characteristics of the measuring device which in turn requires numerically intensive mathematical simulations for extracting the tissue properties from the data. In this paper, we focus on exploiting the algorithmic parallelism in the biocomputational simulation, in order to achieve significant runtime reductions. The entire simulation accounts for over 30,000 spatial points and is too computationally demanding to run in a serial fashion. We discuss our strategies of parallelisation at different levels of granularity and we present our results on two different parallel platforms. We also emphasise the importance of retaining a high level of code abstraction in the application to benefit both agile coding and interdisciplinary collaboration between research groups.