A MODEL AND COMPILATION STRATEGY FOR OUT-OF-CORE DATA-PARALLEL PROGRAMS

It is widely acknowledged in high-performance computing circles that p arallel input/output needs substantial improvement in order to make sc alable computers truly usable. We present a data storage model that al lows processors independent access to their own data and a correspondi ng compilation strategy that integrates data-parallel computation with data distribution for out-of-core problems. Our results compare sever al communication methods and I/O optimizations using two out-of-core p roblems, Jacobi iteration and LU factorization.