以信号处理中非递归型Finite Impulse Response(FIR)滤波器作为研究对象,针对硬件电路设计中存在的面积、功耗和计算性能的问题,通过利用Catapult C高层次综合工具先优化单个FIR滤波,再级联滤波算法的方法以改善滤波的能耗.同时FIR滤波算法在STxP70专用微处理器中采用单指令多数据Single Instruction Multiple Data(SIMD)VECx技术以加速电路的计算性能,分析与评估硬件与软件对实现FIR滤波算法的影响.实验结果表明,采用硬件等效替代的方法能使单个FIR滤波能耗降低6倍,级联滤波能耗降低7倍,满足了绿色环保的电路设计理念.采用软件的SIMD技术可实现数据向量化加速计算,计算加速比提高4倍以上.%This paper discussed the non-recursive FIR filter in signal processing, the face to the problems on area, power consumption of the electronic circuit. Through the use of Catapult C High-Level synthesis tool, we optimized a single non-recursive FIR filter algorithm, then cascade filter algorithm, in order to achieve improving the energy consumption. Through the use of dedicated microprocessor STxP70, we used the SIMD VECx technique to improve the computing performance. Finally we analyzed and evaluated the impacts of the FIR filter algorithms implementation. The experimental results showed the optimizations algorithms are very efficient, it can be reduced 6 times for the single filter and 7 times for the fusion filter on energy consumption. The circuit is ecologic. And software optimizations algorithms with SIMD technique can be increased 4 times for the computation acceleration.
展开▼
