Supercomputers and their performance in computational fluid dynamics

Meer over het boek

Supercomputer technologies have advanced significantly since the introduction of the CRAY-1 in 1976. In the early 1980s, three Japanese supercomputers emerged alongside Cray Research's X-MP series, including the CRAY-2 and NEC SX series, marking a notable generation of supercomputers with peak performances exceeding 1 GFLOPS and impressive compiler improvements. Numerous articles and books detailed their architectures and performance on various benchmark problems. The late 1980s introduced a new wave of supercomputers, with the CRAY Y-MP and Hitachi S-820 debuting in 1988, followed by NEC's SX-3 and Fujitsu's VP2000 series in 1990, and Cray's Y-MP C-90 in 1991. These machines achieved peak performances ranging from several to tens of GFLOPS, but their practical performance remains under-documented. Computational Fluid Dynamics (CFD) has significantly progressed alongside supercomputer development, making it essential for future supercomputers to meet the specific needs of CFD. While many benchmark reports exist, they primarily focus on kernels; fluid dynamics researchers require benchmarks based on real fluid dynamic codes for accurate assessments.