Dismiss
Surety is performing system maintenance this weekend. Electronic date stamps on new Prior Art Database disclosures may be delayed.
Browse Prior Art Database
Quick Links

# Predicting Individual Branches with a Joint Recursive Branch Predictor

IP.com Disclosure Number: IPCOM000106166D
Original Publication Date: 1993-Oct-01
Included in the Prior Art Database: 2005-Mar-20
Document File: 2 page(s) / 92K

IBM

## Related People

R. Rechtschaffen, K. Ekanadham: AUTHOR

## Abstract

A Joint Recursive Branch Predictor (JRBP) has the following set of properties:

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Predicting Individual Branches with a Joint Recursive Branch Predictor

A Joint Recursive Branch Predictor (JRBP) has the following set
of properties:

o   It should continue with the prediction scheme if the prior
prediction is correct.

o   It should wrap around so that a finite length predictor can
continue to predict indefinitely as long as it is correct.

o   It need not synch with addresses of the branches, except at a
error, because

-   A recursive predictor of a branch is defined as a sequenceof
the actions of branches and can be used topredict the next
action of the next branch in conjunction with a BHT which
will supply therelated target address.

The ability to operate within a joint target set of a
set of branches with the same predictor format usedfor a
single branch is a result of the observation that each branch
will specify a targetthat dictates the next branch that is
encountered if thetarget is correct.  Just as joint
distributions containadditional information than marginal
distributions, the use of the joint recursive predictors of
prespecified length can be superior to the operation of
individual recursive predictors on the marginal sequences of
individual branch actions.

o   The predictor needs to have at least one occurrence of each
action of each branch within the loop so that the recovery
mechanism can recover to the position specified by the correct
action following an error.

A surprising property of a JRBP is that it is not affected by
the correctness or lack of correctness of its prediction.  What is
affected is the position within the predictor that is used to predict
the next branch.  As such, a JRBP can be used to predict a single
branch action within a set of branches while the remaining branches
are predicted by other means.  For this single branch the prediction
is made based on the action(s) of prior branches and as such a branch
which is predictable on this basis can be most correctly predicted
with a JRBP.

EXAMPLE - Consider three branches A, B, and C, that interrelate as
part of a JRBP set.  Let the actions of each of these branches be
denoted by {1, 2}.  So A has actions {A1, A2}, B has...