Vectorization of computer programs with applications to computational fluid dynamics

This book provides efficient tools for vectorizing serial computer programs, adapting them to modern vector computer architectures to maximize performance. It targets users in computational physics, including scientists and programmers, but is also relevant for those interested in numerical algorithms. Chapter 9 features examples from Computational Fluid Dynamics, though the discussed numerical methods are widely applicable across various fields of computational physics. The text is organized into four parts. It begins with an introduction highlighting the limitations of conventional serial computers compared to the capabilities of vector machines. The second part focuses on key features of existing computer architectures, specifically the CRAY-1S and CDC-CYBER 205 vector computers. The discussion acknowledges the availability of numerous other vector, parallel computers, and array processors, including DENELCOR's Heterogeneous Element Processor, ICL's Distributed Array Processor, SPERRY UNIVAC's Array Processing System, STAR TECHNOLOGIES ST-100, FLOATING POINT SYSTEMS' Array Processor, FUJITSU's FACOM VP-100 and VP-200, and HITACHI's Integrated Array Processor and S 810 models.