ABSTRACT
In the present scenario, nobody wants to compromise on computational speed when it comes to competitive business, transferring information quickly in a communication link, or even imparting education from a distance, and many more. These requirements lead to high carbon emissions in the environment from a computing machine. The high carbon emissions are one of the main reasons behind green computing. This problem can be solved by speculative parallelism operating procedure. Speculative parallelism is the process of parallel execution aggressively of sequential applications into multiple parallel threads without relying on a compiler in advance. This technique generates different fragments of threads without considering any data and control inter-thread dependence. So, this concept has a substantial fast speed in execution as compared with ILP with multicore chip architecture (MCA). But if any data and control inter-thread dependencies found during execution, 4the whole execution of the sequential work is aborted and re-executed. These aborts are the victim of total lower performance in this concept. One has to deal with various overheads to limits the frequency of aborts leading to high performance and less power consumption. Also, speculative parallelism has less chip’s energy consumption hence promoting green computing. In this chapter, a survey is presented highlighting the working of speculative parallelism techniques, overheads due to mis-speculation of parallel threads, performance, and finally efficient power consumption, which will encourage green computing.
