公开(公告)号：US20160179549A1

公开(公告)日：2016-06-23

申请号：US14580498

申请日：2014-12-23

Applicant: INTEL CORPORATION

Inventor： POLYCHRONIS XEKALAKIS ,  JAMISON D. COLLINS ,  SUMIT AHUJA

IPC: G06F9/38 ,  G06F9/30

CPC classification number: G06F9/3818 ,  G06F9/3005 ,  G06F9/3842 ,  G06F9/3846

Abstract: A processor includes a front end including a decoder to decode a branch instruction to perform a branch operation. The processor includes a loop stream unit with logic to identify from the branch instruction that the branch operation is a loop operation, determine whether the loop operation will include a fixed or effectively-infinite number of iterations, load decoded instructions of a loop iteration of the loop operation, and cyclically issue the decoded instructions of the loop iteration in a manner based upon whether the loop operation will include a fixed or effectively-infinite number of iterations. The processor also includes an execution unit to execute the branch instruction and a retirement unit including to retire the branch instruction.

Abstract translation: 处理器包括前端，其包括用于解码分支指令以执行分支操作的解码器。 处理器包括具有逻辑的循环流单元，用于从分支指令中识别分支操作是循环操作，确定循环操作是否包括固定或有效无限次数的迭代，将解码的指令循环迭代 循环操作，并且基于循环操作是否将包括固定或无限次迭代次数的方式循环地发布循环迭代的解码指令。 处理器还包括执行分支指令的执行单元和退出分支指令的退休单元。

公开(公告)号：US20160179538A1

公开(公告)日：2016-06-23

申请号：US14576915

申请日：2014-12-19

Applicant: Intel Corporation

Inventor： JAMISON D. COLLINS ,  Jayesh Iyer ,  Sebastian Winkel ,  Polychronis Xekalakis ,  Howard H. Chen ,  Rupert Brauch

IPC: G06F9/30

CPC classification number: G06F9/30134 ,  G06F8/41 ,  G06F9/3004 ,  G06F9/30072 ,  G06F9/30087 ,  G06F9/3013 ,  G06F9/30145 ,  G06F9/384 ,  G06F9/3859 ,  G06F9/3863

Abstract: Embodiments of a method and apparatus for implementing and maintaining a stack of predicate values with stack synchronization instructions. In one embodiment the apparatus is an out of order hardware/software co-designed processor including instructions to explicitly manage the predicate register stack to maintain stack consistency across branches of executing that push a variable number of predicate values onto the predicate stack. In one embodiment the stack-based predicate register implementation enables early branch calculation and early branch misprediction recovery via early renaming of predicate registers.

Abstract translation: 用于使用堆栈同步指令来实现和维护谓词值堆栈的方法和装置的实施例。 在一个实施例中，该装置是一种无序的硬件/软件协同设计的处理器，其包括明确地管理谓词寄存器堆栈以便在执行的分支之间保持堆栈一致性的指令，其将可变数量的谓词值推送到谓词堆栈。 在一个实施例中，基于栈的谓词寄存器实现能够通过早期重命名谓词寄存器来实现早期分支计算和早期分支错误预测恢复。