2005•Journal of the Korea Society of Computer and InformationRequires access

Performance Evaluation of Value Predictor in High Performance Microprocessors

Byoung-Chan Jeon, Kim Hyeock-Jin, RU Dae-Hee

Open publisher page 0 citations

Abstract

value prediction in high performance micro processors is a technique that exploits Instruction Level Parallelism(ILP) by predicting the outcome of an instruction and by breaking and executing true data dependences. In this paper, the mean Performance improvements by predictor according to a point of time for update of each table as well as prediction accuracy and Prediction rate are measured and assessed by comparison and analysis of value predictor that issues in parallel and run by predicting value, which is for Performance improvements of ILP in micro Processor. For the verification of its validity the SPECint95 benchmark through the simulation is compared by making use of execution driven system.

About this research paper

What this paper is about

value prediction in high performance micro processors is a technique that exploits Instruction Level Parallelism(ILP) by predicting the outcome of an instruction and by breaking and executing true data dependences. In this paper, the mean Performance improvements by predictor according to a point of time for update of each table as well as prediction accuracy and Prediction rate are measured and assessed by comparison and analysis of value predictor that issues in parallel and run by predicting value, which is for Performance improvements of ILP in micro Processor. For the verification of its validity the SPECint95 benchmark through the simulation is compared by making use of execution driven system.

Why it matters

A significance statement is not available in the OpenAlex record.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

value prediction in high performance micro processors is a technique that exploits Instruction Level Parallelism(ILP) by predicting the outcome of an instruction and by breaking and executing true data dependences. In this paper, the mean Performance improvements by predictor according to a point of time for update of each table as well as prediction accuracy and Prediction rate are measured and assessed by comparison and analysis of value predictor that issues in parallel and run by predicting value, which is for Performance improvements of ILP in micro Processor. For the verification of its validity the SPECint95 benchmark through the simulation is compared by making use of execution driven system.

Key concepts: Benchmark (surveying), Computer science, Parallelism (grammar), Value (mathematics), Table (database), Branch predictor, Exploit, Parallel computing

Related papers

Back to paper searchBrowse research topicsOriginal source
Performance Evaluation of Value Predictor in High Performance Microprocessors — Research Paper | ScholarLens