Logo image
低功率超大型積體電路之資源配置
Thesis

低功率超大型積體電路之資源配置

江鳳儀
Masters, National Tsing Hua University
1998

Abstract

低功率高階合成,變數固定,暫存器分割,時脈,運算器配置。 high level synthesis, variable binding, register partitioning, functional unit binding.
We present in this project two new algorithms for the synthesis of low power VLSI circuits. One is for binding variables to registers and partitioning the registers into groups so that a group of registers can be shut down when all the registers in the group are inactive during some of the clock cycles. The other is for binding operations to functional units to reduce the circuit activities of the units. By holding the selection lines of the multiplexors which are connected to a functional unit unchanged until the functional unit is ready to execute a new operation, we can reduce substantially the circuit activities of the functional units. We integrated the two algorithms in a synthesis. Our experimental results show that (i) Our register binding and partitioning algorithm reduces 40-65% of the total number of active steps in all registers when compared with the left edge algorithm. (ii) Our functional unit binding algorithm improve both area (substantial) in terms of the number of functional units and power consumption (small,~3%) when compared with the integrated allocation algorithm.

Metrics

1 Record Views

Details

Logo image