This study will generate a true “apples-to-apples” comparison of an industrial scientific computing code across both major leadership-class HPC architectures, as well as the more commodity Linux cluster. As GE, and the greater world, looks to increase simulation-based science, larger simulations will be required, and knowing the right platform upon which these simulations should be run will be very valuable. The scalability study includes both Strong and Weak scaling of TACOMA simulating the fluid flow through two blade-rows of a turbine, representing an unsteady, sliding mesh calculation. The scalability will be compared to identical scaling studies done on IBM BlueGene/P and x86 Linux clusters at other sites.
|Source||Hours||Start Date||End Date|