• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>International Conference on Cognitive Informatics and Soft Computing >Functional Link Artificial Neural Network (FLANN) Based Design of a Conditional Branch Predictor
【24h】

Functional Link Artificial Neural Network (FLANN) Based Design of a Conditional Branch Predictor

机译：功能链路人工神经网络（FLANN）条件分支预测器的设计

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Conditional branch predictor (CBP) is an essential component in the design of any modern deeply pipelined superscalar microprocessor architecture. In the recent past, many researchers have proposed varieties schemes for the design of the CBPs that claim to offer desired levels of accuracy and speed needed to meet the demand for the architectural design of multicore processors. Among various schemes in practice to realize the CBPs, the ones based on neural computing--i.e., artificial neural network (ANN) has been found to outperform other CBPs in terms of accuracy. Functional link artificial neural network (FLANN) is a single layer ANN with low computational complexity which has been used in versatile fields of application, such as system identification, pattern recognition, prediction and classification, etc. In all these areas of application, FALNN has been tested and found to provide superior performance compared to their multilayer perceptron (MLP) counterpart. This paper proposes design of a novel FLANN-based dynamic branch predictor and compares the performance against a perceptron-based CBP. The proposed FALNN-based CBP has been implemented in C++. The performance of the proposed CBP has been evaluated using standard benchmark trace data files and is found to have performance comparable to the existing perceptron-based predictions in terms of speed and accuracy.
机译：条件分支预测器（CBP）是任何现代深层流水线上卡微处理架构的重要组成部分。在最近的过去，许多研究人员提出了用于设计CBP的品种方案，要求提供满足多核处理器建筑设计需求所需的所需精度和速度的所需的精度和速度。在实践中的各种方案中，以实现CBPS，基于神经计算 - 即，人工神经网络（ANN），在准确性方面越优越其他CBP。功能链接人工神经网络（FLANN）是一个具有低计算复杂度的单层ANN，它已被用于多功能的应用领域，例如系统识别，模式识别，预测和分类等。在所有这些应用领域，FALNN具有经过测试，发现与其多层的感知（MLP）对应相比提供出色的性能。本文提出了一种设计基于小型的Flann的动态分支预测因子，并比较了对基于Perceptron的CBP的性能。拟议的基于FALNN的CBP已在C ++中实施。已经使用标准基准跟踪数据文件进行了评估了所提出的CBP的性能，并且发现在速度和准确性方面具有与现有的Perceptron的预测相当的性能。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号