Dupros, MPI-OpenMP hybrid simulations using bound- ary integral equation and finite difference methods for earthquake dy- namics and wave propagation: Application to the 2007 Niigata Chuetsu- Oki earthquake (Mw6.6), International Conference on Computational Science, 2011 (accepted)....
I'm running into a difference in the behavior between using ifx -g and ifort -g. The old code works as expected with ifx when the -g flag is not used but throws an integer divide by zero when the -g flag is used. With ifort the code works as expected...
By encapsulating the performance-portability aspects (e.g. whether to parallelise with OpenMP or OpenACC), these scripts enable a separation of concerns between the scientific implementation and the optimisation choices. This allows each aspect to be explored and developed largely independently. ...
a B-scan where only the position of simple sources and receivers, moved using #src_steps and #rx_steps, changes between models. --opt-taguchi flag runs a series of models using an optimization process based on Taguchi's method. For further details see the user libraries section of the ...
In the previous two Laplacian posts, we developed a HIP implementation of a finite-difference code designed around a Laplace operator and applied two possible code optimizations to optimize memory movement between the L2 cache and global memory. This third part will cover some additional optimizations...
MPIOpenMP2007 Chuetsu-Oki earthquakeFor the past two decades, the combination between the boundary integral equation method (BIEM) and the finite difference method (FDM) has provided a powerful approach for simulating the dynamic rupture process along a complex fault system and the resultant seismic ...
I'm running into a difference in the behavior between using ifx -g and ifort -g. The old code works as expected with ifx when the -g flag is not used but throws an integer divide by zero when the -g flag is used. With ifort the code works as expecte...
Support for boundary layers and staggered-grid stencils. Vector-folding to increase data reuse via non-traditional data layout. Multi-level OpenMP parallelism to exploit multiple CPU cores and threads. OpenMP offloading to GPUs. MPI scaling to multiple sockets and nodes with overlapped communication an...
I'm running into a difference in the behavior between using ifx -g and ifort -g. The old code works as expected with ifx when the -g flag is not used but throws an integer divide by zero when the -g flag is used. With ifort the code works as e...
I'm running into a difference in the behavior between using ifx -g and ifort -g. The old code works as expected with ifx when the -g flag is not used but throws an integer divide by zero when the -g flag is used. With ifort the code works as ex...