An alternative to branch prediction

Abstract

Through this paper we developed an alternative approach to the present -day two level dynamic branch prediction structures. Instead of predicting branches based on history information, we propose to pre -calculate the branch outcome. A pre -calculated branch prediction (PCB) determines the outcome of a branch as soon as all of the branch's operands are known. The instruction that produced the last branch's operand will trigger a supplementary branch condition estimation and, after this operation, it correspondingly computes the branch outcome. This outcome is cached into a prediction table. The new proposed PCB algorithm clearly outperforms all the classical branch prediction schemes, simulations on SPEC and Stanford HSA benchmarks, proving to be very efficient. Also, our investigations related to architectural complexity and timing costs are quite optimistic, involving an original alternative to the present-day in branch prediction approach.

Unbiased Branches: An Open Problem

“…Vintan [6] proposed pre-computing branches by determining a branch outcome as soon that branch's operands were available. The basis behind such precomputation was that the instruction that produced the last branch source operand would also trigger the branch condition estimation.…”

A Bypass Mechanism to Enhance Branch Predictor for SMT Processors

“…The past few years, some new methods are introduced such as Lucian N. Vintan's pre-computed branches [17] which compute the destination of conditional branches as early as the first operand is ready for superscalar processors, Robert S. Chappell's Difficult-path branch prediction that uses subordinate micro-threads [18], Craig Zilles's Execution-based prediction that uses speculative slices [19], Lucian Vintan introduced the Neural Branch Prediction [20], Renju Thomas et al studied dynamic dataflow-based identification of correlated branches from a large global history [21], Steven Swanson et al evaluated the importance of branches in modern deep pipelined processors [22], and David Tarjan introduced the hashed perceptron predictor, which merges the concepts behind the g-share, path-based and perceptron branch predictors [23].…”