Abstract: An embedded genetic allocator system uses genetic algorithms to generate trial solutions to the allocation of data buffers among various kinds of memory banks and measures the quality of each solution. The quality is determined through the use of a monitoring system that accurately measures application execution times while maintaining low intrusiveness in the software execution. Because the allocation process is highly automated and based upon accurate performance data derived from actual system operation, optimal solutions to the allocation problem are reached much more quickly and efficiently than can be done using current, largely manual, approaches.