- Title
- Electronic poster: a massively parallel Lattice Monte Carlo algorithm in CUDA for thermal conduction simulations
- Creator
- Wang, Michael; Mignone, Paul J.; Riley, Daniel P.; Franks, George V.; Fiedler, Thomas; Murch, Graeme E.
- Relation
- International Conference for High Performance Computing Networking, Storage and Analysis (SC' 11). SC'11 Companion: Proceedings of the 2011 companion on High Performance Computing Networking, Storage and Analysis (Seattle, WA 12-18 November, 2011) p. 93-94
- Publisher Link
- http://dx.doi.org/10.1145/2148600.2148648
- Publisher
- Association for Computing Machinery (ACM)
- Resource Type
- conference paper
- Date
- 2011
- Description
- We present a highly parallel CUDA kernel based on the Lattice Monte Carlo (LMC) method for transient thermal conduction, which achieves a peak acceleration of more than 100x over a single-threaded Fortran version. A number of memory and branching optimizations for the Graphic Processing Unit (GPU) architectures are described. Combining all tweaks, a fully-optimized kernel is able to outperform the initial speed-up of around 13x observed for a naive CUDA implementation by another order of magnitude, to reach the peak performance reported (on a single NVIDIA Tesla C2050). Comparison benchmarks are also provided for the Tesla C1060, whereas the Fortran code was executed on an Intel i5 CPU running at 3.6 GHz.
- Subject
- Lattice Monte Carlo; diffusion; C1060; thermal conduction; transient; composites; parallel; GPU; Tesla; CUDA; C2050
- Identifier
- http://hdl.handle.net/1959.13/1057841
- Identifier
- uon:16283
- Identifier
- ISBN:9781450310307
- Language
- eng
- Reviewed
- Hits: 5584
- Visitors: 5538
- Downloads: 1
Thumbnail | File | Description | Size | Format |
---|