Faster Results by Multi-Threading Data Steps
Many SAS users run programs on systems with multiple processors. Starting with SAS9, certain intensive SAS procedures take can advantage of these multiple processors to reduce run-time, and make more effective use of system resources. Many users have long running data steps, which could also benefit from parallel execution. This paper discusses how, and when to multi-thread an intensive data step, along with all the considerations need to do so. Topics covered include partitioning, I/O parallelism and SPDE/SPDS, run-time tracking, inter-row dependencies, thread execution, and re-joining. An example parallel data step macro package will also be included.