Harsh, B. (CSE) – SUPERSCALAR, MULTIPLE TAKEN BRANCH PREDICTOR

This work addresses improvements in branch prediction mechanism to support high perfor-
mance processors. The state of the art aims to balance the prediction latency and prediction
accuracy using multi level correcting predictors [27]. Prior published work focusses on scalar
designs and prediction accuracy improvement for hard to predict branches employing tailor
made, non generic and non transferrable solutions [8]. Recent work also proposes ahead pre-
diction [42–44] to solve the problem of low accuracy of L0 predictor.

This work proposes efficent, generic and transferrable solutions to reduce mispredic-
tions and to use the fetch bandwidth more efficiently. This includes a biased overriding multi-
level hierarchy with three predictor levels (L0, L1, L2). L0 uses a High-Confidence-Only Taken
(HOTP) predictor that only predicts high-confidence taken control-flow instructions. This work
further uses L1-L2 biased training to decrease mispredictions by L2 while it trains on branches
on which L1 has reached high confidence. This work proposes a superscalar predictor built
using the state of the art scalar predictor. Superscalar predictor is implemented by sizing a su-
perscalar TAGE variant (BATAGE) using Optuna-based search. with varying table sizes and
aspect ratios. The work further proposes a branch predictor frontend design (nTakenBP) to de-
liver multiple taken branch predictions per cycle. Unlike prior work, nTakenBP achieves this by
extending the existing BTB and TAGE tag-comparison logic rather than deepening lookahead.

Event Host: Bhawandeep Singh Harsh, Ph.D. Candidate, Computer Science & Engineering

Advisor: Jose Renau

Zoom: https://ucsc.zoom.us/j/4166778865?pwd=cS9NcnVjRjArYlRRcDcrY3d5N0ZKQT09