Research Of Power Optimization Techniques Based On Scheduling And Binding

Along with the rapid development of electronic technology, the characteristic dimension of integrated circuits is getting smaller and smaller, the integrity and clock frequency are also increasing fleetly, the power optimization has become a main consideration in integrated circuit (IC) design. Low power VLSI designs can be achieved at various design levels, which rang from system, algorithmic (behavioral), architecture, and logic levels to circuit level, according to the Down-Top design flow. The effects of low power design are significantly different at different levels. The higher is the level of design abstraction, the more early low power designs are considered. So low power designs at the higher levels are able to achieve evident effect. In this paper, we optimize power at high level stage to achieve maximum optimization. The main contents of this paper are shown as follows:1. Multiple voltage schedule base on comparison of mobility and delay. Reduce the operating voltage is the most effective way to reduce dynamic power consumption, we propose a multiple voltage scheduling algorithm under timing constraints. By comparing the operating mobility and delay, reducing operating power supply voltage in meeting the conditions of time constraints, thereby reducing the power consumption of the circuit. Experimental results show that our method can effectively reduce power consumption and achieve the desired requirements.2. Multiple voltage scheduling based on Whole Annealing Genetic Algorithm. Theory has proven time constrained multiple voltage scheduling problem belongs to NP-hard problems, while the genetic algorithm is simulated in the natural environment of biological genetic and evolutionary processes shape the probability of an adaptive global optimization search algorithm, which is an effective way to solve NP-hard problem, so you can use genetic algorithms to solve multi-voltage scheduling problem. But the genetic algorithm has its inherent defects, so the genetic algorithm and simulated annealing algorithm for combining the whole annealing genetic algorithm is applied to multi-voltage scheduling technologies. Experimental results show that the proposed method can reduce power consumption significantly and achieve the desired requirements.3. Low power optimization algorithm based on multiple voltage scheduling and resource binding. High-level synthesis phase of the three tasks: operation scheduling, resource allocation and resource bindings, which are mutually affected. The proposed algorithm considers scheduling and binding stages, and synthetically use of multi-voltage scheduling technology and resources binding technique. Experimental results show that, algorithm considering the scheduling and binding can reduce more power consumption than algorithm only considering the scheduling.