Performance predictors for graphics processing units applied to dark-silicon-aware design space exploration
Research output: Journal contributions › Journal articles › Research › peer-review
Standard
In: Concurrency and Computation: Practice and Experience, Vol. 35, No. 17, e6877, 01.08.2023.
Research output: Journal contributions › Journal articles › Research › peer-review
Harvard
APA
Vancouver
Bibtex
}
RIS
TY - JOUR
T1 - Performance predictors for graphics processing units applied to dark-silicon-aware design space exploration
AU - Sonohata, Rhayssa
AU - Arigoni, Danillo Christi A.
AU - Fernandes, Eraldo Rezende
AU - Ribeiro dos Santos, Ricardo
AU - Dessandre Duenha, Liana
N1 - Special Issue: WSCAD 2020. PDCAT 2020/PDCAT‐PAAP 2020 This study was financed in part by FUNDECT and Coordenação de Aperfeiçoamento de Pessoal de Nível Superior ‐ Brasil (CAPES) ‐ Finance Code 001.
PY - 2023/8/1
Y1 - 2023/8/1
N2 - The limitations on the scalability of computer systems imposed by the dark-silicon effects are so severe that they support the extensive use of heterogeneity such as the GP-GPU for general purpose processing. Performance simulators of GP-GPU heterogeneous systems aim to provide performance accuracy at the cost of execution time. In this work, we handle time-consuming simulations of design space exploration systems based on GPUs. We have developed performance predictors based on machine learning (ML) algorithms and evaluated them in accuracy and throughput (number of predictions per second). We measure model accuracy through the mean absolute percentage error (MAPE) and the model efficiency through a throughput metric (millions of predictions per second). Our experiments revealed that decision trees predictors are the most promising regarding accuracy and efficiency. We applied the best predictors into the MultiExplorer, a dark silicon-aware design space exploration tool that allows designers to explore the architecture and microarchitecture of multicore/manycore system design.
AB - The limitations on the scalability of computer systems imposed by the dark-silicon effects are so severe that they support the extensive use of heterogeneity such as the GP-GPU for general purpose processing. Performance simulators of GP-GPU heterogeneous systems aim to provide performance accuracy at the cost of execution time. In this work, we handle time-consuming simulations of design space exploration systems based on GPUs. We have developed performance predictors based on machine learning (ML) algorithms and evaluated them in accuracy and throughput (number of predictions per second). We measure model accuracy through the mean absolute percentage error (MAPE) and the model efficiency through a throughput metric (millions of predictions per second). Our experiments revealed that decision trees predictors are the most promising regarding accuracy and efficiency. We applied the best predictors into the MultiExplorer, a dark silicon-aware design space exploration tool that allows designers to explore the architecture and microarchitecture of multicore/manycore system design.
KW - dark-silicon
KW - design space exploration
KW - GPUs
KW - heterogeneous computing
KW - performance predictors
KW - Business informatics
UR - http://www.scopus.com/inward/record.url?scp=85125564653&partnerID=8YFLogxK
U2 - 10.1002/cpe.6877
DO - 10.1002/cpe.6877
M3 - Journal articles
AN - SCOPUS:85125564653
VL - 35
JO - Concurrency and Computation: Practice and Experience
JF - Concurrency and Computation: Practice and Experience
SN - 1532-0626
IS - 17
M1 - e6877
ER -