Feedback Directed Implicit Parallelism

Overview

This paper presents an automated way of using spare CPU resources within a shared memory multi-processor or multi-core machine. The approach is to profile the execution of a program; from this to identify pieces of work which are promising sources of parallelism; recompile the program with this work being performed speculatively via a work-stealing system and then to detect at run-time any attempt to perform operations that would reveal the presence of speculation.