چكيده به لاتين
Microprocessor production restrictions such as drastic reduction of semiconductor manufacturing technology(Physical limitations) ,Power consumption and high cost, lead scientists to try efficient application of processors And increase the efficiency of the processors performance. Previous methods were used to reduce a task’s time for processor. these methods are based on processor’s scheduling and Task management, will prevent hardware complexity and reduce Time and Energy overload. But there is no definition on acceleration in processor’s Computation. In some cases it’s about speedup, but that’s not true .So it’s crucial to define accelerating Computation.
The main goal of this thesis is to define computation acceleration. We need a definition to understand when we have speedup and when we have acceleration. then, we propose a general idea to use multithread processors more efficiently. The main parts of this idea are : a) pre-analysis the code b) analysis and pre-process specific part’s of the code and C) get the results to the main processor in the appropriate time.
To experiment the idea, we test the effect of the idea on GPU ‘s performance acceleration. according to the results, pre-processing specific parts of the code and then get it to GPU, will lead to GPU’s performance acceleration.it show’s that with pre-processing a specific part of the code, we can save the time equivalent as the time of the whole process.
Keywords: processor , multithread processing , acceleration, pre-proccess
