Perceptron Branch Prediction with Separated Taken/Not-Taken Weight Tables
نویسندگان
چکیده
Perceptron predictor is well known as its ability to exploit long history information. A limitation of perceptron predictor is that it cannot effectively predict linearly inseparable correlations. In this paper, we present a revised perceptron predictor with separated Taken/Not-Taken weight tables (SWP). It can treat the taken/not taken correlation separately and therefore predict some linearly inseparable branches effectively. Compared to the existing perceptron branch predictors, it requires less number of bits in each weight to achieve the same accuracy, and thus less computational latency of making a prediction.
منابع مشابه
Low-Power Perceptron Branch Predictor
Perceptron based predictors are highly accurate. This high accuracy is the result of exploiting long history lengths [1] and is achieved at the expense of high complexity. The dot product of two vectors is used as prediction. The first vector is the branch outcome history where the second vector is composed of per branch weights, which represent the correlation between branch outcome and previo...
متن کاملA3PBP: A Path Traced Perceptron Branch Predictor Using Local History for Weight Selection
In this paper, we present a new perceptron branch predictor called Advanced Anti-Aliasing Perceptron Branch Predictor (APBP), which can be reasonably implemented as part of a modern micro-processor. Features of the predictor are twofold: (1) Local history is used as part of the index for weight tables; (2) Execution path history is effectively used. Contrary to global/local perceptron branch pr...
متن کاملBranch Prediction Techniques and Optimizations
Branch prediction is one of the ancient performance improving techniques which still finds relevance into modern architectures. While the simple prediction techniques provide fast lookup and power efficiency they suffer from high misprediction rate. On the other hand, complex branch predictions – either neural based or variants of two-level branch prediction – provide better prediction accuracy...
متن کاملPerformance Analysis of Branch Prediction Unit for Pipelined Processors
The branch predictor plays a crucial role in the achievement of effective performance in microprocessors with pipelined architectures. This paper analyzes performance of branch prediction unit for pipelined processors. A memory of 512 bytes is designed for storing instructions. A 32 byte memory is designed for branch target buffer (BTB). This memory is utilized for storing history of the branch...
متن کاملRevisiting the Perceptron Predictor
The perceptron branch predictor has been recently proposed by Jiménez and Lin as an alternative to conventional branch predictors. In this paper, we build upon this original proposal in three directions. First, we show that the potential accuracy that can be achieved by perceptron-like predictors was largely underestimated. The accuracy of the redundant history skewed perceptron predictor (RHSP...
متن کامل