BADGR: A practical GHR implementation for TAGE branch predictors

BADGR: A practical GHR implementation for TAGE branch predictors
复制标题

DOI:
10.1109/iccd.2016.7753338
复制
发表时间:
2016-10
期刊:
2016 IEEE 34th International Conference on Computer Design (ICCD)
影响因子:
--
通讯作者:
David J. Schlais;Mikko H. Lipasti
David J. Schlais;Mikko H. Lipasti
中科院分区:
其他
文献类型:
--
作者:
David J. Schlais;Mikko H. Lipasti

文献摘要

被引文献

相似文献

在这项工作中,我们探索了带有推测性更新的标记几何长度(TAGE)风格分支预测器的全局历史寄存器(GHR)实现。我们分解了在正常操作期间和错误计算之后更新和恢复TAGE预测器历史寄存器的需求,讨论了各种设计在哪些地方显示出较大的检查点和/或操作开销。为了降低这些低效率,我们引入了BADGR,这是一种用于TAGE预测器的新型GHR设计,与原始检查点技术相比,它的功耗和芯片面积分别降低了90%和85%。
In this work, we explore global history register (GHR) implementations for Tagged Geometric length (TAGE) style branch predictors with speculative updates. We break down the requirements to both update and recover TAGE predictors' history registers during normal operation and after mispeculation, discussing where various designs exhibit large checkpoint and/or operation overheads. To reduce these inefficiencies, we introduce BADGR, a novel GHR design for TAGE predictors that lowers power consumption and chip area over naive checkpointing techniques by 90% and 85%, respectively.