Parallel computing has taken many forms over the past two decades. Mainframe shared-memory devices have evolved to RISC-based systems with both distributed and shared memories. Early versions of massively-parallel systems attempted to use commodity chips such as the Intel 386 and 486 as processing elements. Although low in cost, these early Intel CPU's lacked the performance and support for large memory sizes required for effective use in parallel simulation. However, recent architectural advances have overcome these obstacles. With clock speeds exceeding 400 Megahertz and support for memory sizes up to several gigabytes, the new Intel-based platforms have exhibited the ability to handle substantial simulations in serial mode. Using these commodity CPU's in tandem to perform parallel simulation further exploits their low cost. The work described in this paper discusses the application of a parallel cluster of Windows NT-based workstations to reservoir simulation. Up to twenty-four processors with 512 megabytes of memory per processor were utilized to solve black-oil and compositional simulation problems with hundreds of thousands of cells each. Computational performance comparable to more sophisticated systems was obtained. The work focuses on reducing the overhead of message passing and interprocessor communication, increasing the inter-node bandwidth, and predicting the ultimate effect of these two factors on the scalability of the system. Several communication schemes are investigated, including an NT version of the Message Passing Interface (MPI) with off-the-shelf Ethernet hardware as well as a custom high-performance messaging layer constructed over the Virtual Interface Architecture (VIA) and a high-speed System Area Network (SAN).

You can access this article if you purchase or spend a download.