Solution Performance

FWI2D-inversion* is a 2D time-domain full waveform inversion that uses finite differences for both forward and backward problems and a quasi-Newton (with L-BFGS algorithm) optimization scheme for the model parameters update. It is based on a 2D isotropic acoustic wave equation with variable density and P-wave velocity and uses a 8th order in space (symmetric coefficient) and 4th order in time scheme with absorbing boundary conditions.

Product and Performance Information

1

Fwi2d modeling & Fwi2d inversion: version dev1.6 from  june16.
OS: Centos 7.3.
Compiler: Intel® Parallel Studio XE Cluster Edition 2017 update 2
Run config:  MPI+OMP . One MPI per socket. Same workload for all IA.
Workload: Extended Marmousi-1 , single-node problem-size:  600x1423
Testing by Intel, April 2017
Baseline: 2S Intel® Xeon® processor E5-2680 v3, 2.5GHz, 24 cores (12 each socket), turbo and HT OFF. DDR4 2133 Mhz
Next gen: 2S Intel® Xeon® processor E5-2680 v4, 2.4GHz, 28 cores (14 each socket), turbo and HT OFF. DDR4 2400 Mhz
New: 2S Intel® Xeon® Gold 6148 processor, 2.4GHz, 40 cores (20 each socket), turbo and HT OFF. DDR4 2667 Mhz

2

Software and workloads used in performance tests may have been optimized for performance only on Intel® microprocessors. Performance tests, such as SYSmark* and MobileMark*, are measured using specific computer systems, components, software, operations and functions. Any change to any of those factors may cause the results to vary. You should consult other information and performance tests to assist you in fully evaluating your contemplated purchases, including the performance of that product when combined with other products. For more complete information about performance and benchmark results, visit http://www.intel.com/benchmarks.