| Sumario: | The article provides information about speculative parallelization, a technique used in parallelizing loops when the computer system fails to determine dependences at a compile time. Also known as thread-level speculation, this technique assumes that the system can execute all iterations of a given loop in parallel. Once the system discovers a dependence violation during runtime, speculative parallelization halts the incorrectly computer work and restarts it with the exact values. The effectiveness of speculative parallelization depends on the number of loops or iterations the system can execute in parallel.
|
|---|