Correlating Branch Prediction with Two-Level Predictors
What does GHR stand for in the context of a correlating branch predictor?
In a two-level correlating predictor, what is the GHR used for?
What is stored in the Pattern History Table (PHT)?
What key problem do correlating predictors solve that simple predictors do not?
A (2,2) correlating predictor has a GHR with the value '10'. What index in the PHT will it use?
After a prediction is made and the actual outcome is known, which two components are updated?
In a (2,2) predictor, the GHR is '01'. The branch is actually 'Taken' (T). What will the new state of the GHR be?
Why is a GAg predictor called a 'two-level' predictor?
Consider a (2,2) predictor where GHR = '11' and PHT[3] = '01' (Weak NT). The prediction is 'Not Taken'. The branch is actually 'Taken'. What are the new states of GHR and PHT[3]?