Accelerating seismic computations using customized number representations on FPGAs

Reverse time migration

The primary bottleneck of reverse time migration is applying the finite-difference stencil. In addition to the large operation count (5 to 31 samples per cell) the access pattern has poor cache behavior for real size problems. Beyond applying the 3-D stencil the next major cost is implementing damping boundary conditions. Methods such as Perfectly Matched Layers (PML) can be costly (Berenger, 1994). Finally, if you want to use reverse time migration for velocity analysis, subsurface offset gathers need to be generated. The same cost profile that exists in downward continued based migration exists for reverse time migration.

Last summer the focus was on implementing the 2-D elastic modeling convolutional kernel. We achieved a speed up of 8-16x, again depending on data precision. This summer we concentrated on 3-D acoustic modeling kernel.

Accelerating seismic computations using customized number representations on FPGAs