Indirect addressing and load balancing for faster solution to Mandelbrot set on SIMD architectures

The authors present a method for using local indirect addressing to achieve faster solutions for some problems with data-dependent convergence rates on SIMD (single-instruction-stream, multiple-data-stream) architectures. A class of problems characterized by computations on data points where the computation is identical but the convergence rate is data dependent is examined. In the absence of indirect addressing, algorithm time is governed by the maximum number of iterations. An algorithm using indirect addressing allows a processor to proceed to the next data point upon convergence. Thus the overall number of iterations will approach the mean convergence rate for a sufficiently large problem. Load-balancing techniques can be applied for additional performance improvement. These techniques are used for solving Mandelbrot sets on the MP-1 massively parallel computer