Description: GPGPU acceleration of the reduced eigensolution phase of the AMS eigensolver delivers improved performance for large-scale models requiring more than 10,000 eigenmodes. In addition, the parallel scaling of the reduced eigensolution phase is improved compared to Abaqus 6.13 for the same kind of models. Table 11–1 illustrates the performance improvements of the AMS eigensolver due to GPGPU acceleration for two industrial models: Model 1 is a 2.5 million degree-of-freedom vehicle body model, and Model 2 is a 9.4 million degree-of-freedom vehicle body model. Both analyses were run on a Linux machine with dual 2.6 GHz Intel Sandybridge processors (2 8 cores), 128 GB memory, and 2 Nvidia Kepler cards (K20X). In all cases 16 cores are used.