A simple model for portable and fast prediction of execution time and power consumption of GPU Kernels

Characterizing compute kernel execution behavior on GPUs for efficient task scheduling is a non-trivial task. We address this with a simple model enabling portable and fast predictions among different GPUs using only hardware-independent features. This model is built based on random forests using 18...

Full description

Saved in:
Bibliographic Details
Main Authors: Braun, Lorenz (Author) , Nikas, Sotirios (Author) , Song, Chen (Author) , Heuveline, Vincent (Author) , Fröning, Holger (Author)
Format: Article (Journal)
Language:English
Published: 2021
In: ACM Transactions on architecture and code optimization
Year: 2021, Volume: 18, Issue: 1, Pages: 1-25
ISSN:1544-3973
DOI:10.1145/3431731
Online Access:Verlag, lizenzpflichtig, Volltext: https://doi.org/10.1145/3431731
Get full text
Author Notes:Lorenz Braun (Institute of Computer Engineering, Heidelberg University, Germany), Sotirios Nikas, Chen Song, and Vincent Heuveline (Engineering Mathematics and Computing Lab, Heidelberg University, Germany), Holger Fröning (Institute of Computer Engineering, Heidelberg University, Germany)

MARC

LEADER 00000caa a2200000 c 4500
001 175296330X
003 DE-627
005 20220819152353.0
007 cr uuu---uuuuu
008 210331s2021 xx |||||o 00| ||eng c
024 7 |a 10.1145/3431731  |2 doi 
035 |a (DE-627)175296330X 
035 |a (DE-599)KXP175296330X 
035 |a (OCoLC)1341402022 
040 |a DE-627  |b ger  |c DE-627  |e rda 
041 |a eng 
084 |a 28  |2 sdnb 
100 1 |a Braun, Lorenz  |d 1990-  |e VerfasserIn  |0 (DE-588)1230497145  |0 (DE-627)1752968506  |4 aut 
245 1 2 |a A simple model for portable and fast prediction of execution time and power consumption of GPU Kernels  |c Lorenz Braun (Institute of Computer Engineering, Heidelberg University, Germany), Sotirios Nikas, Chen Song, and Vincent Heuveline (Engineering Mathematics and Computing Lab, Heidelberg University, Germany), Holger Fröning (Institute of Computer Engineering, Heidelberg University, Germany) 
264 1 |c 2021 
300 |a 25 
336 |a Text  |b txt  |2 rdacontent 
337 |a Computermedien  |b c  |2 rdamedia 
338 |a Online-Ressource  |b cr  |2 rdacarrier 
500 |a Publication date: December 2020 
500 |a Gesehen am 31.03.2021 
520 |a Characterizing compute kernel execution behavior on GPUs for efficient task scheduling is a non-trivial task. We address this with a simple model enabling portable and fast predictions among different GPUs using only hardware-independent features. This model is built based on random forests using 189 individual compute kernels from benchmarks such as Parboil, Rodinia, Polybench-GPU, and SHOC. Evaluation of the model performance using cross-validation yields a median Mean Average Percentage Error (MAPE) of 8.86-52.0% for time and 1.84-2.94% for power prediction across five different GPUs, while latency for a single prediction varies between 15 and 108 ms. 
650 4 |a cross-validation 
650 4 |a Execution time prediction 
650 4 |a GPGPU 
650 4 |a GPU computing 
650 4 |a portable performance prediction 
650 4 |a power prediction 
650 4 |a profiling 
650 4 |a random forest 
700 1 |a Nikas, Sotirios  |e VerfasserIn  |0 (DE-588)1148482083  |0 (DE-627)1008751863  |0 (DE-576)496317229  |4 aut 
700 1 |a Song, Chen  |d 1988-  |e VerfasserIn  |0 (DE-588)1124811346  |0 (DE-627)879205520  |0 (DE-576)48315671X  |4 aut 
700 1 |a Heuveline, Vincent  |d 1968-  |e VerfasserIn  |0 (DE-588)1046579266  |0 (DE-627)776691880  |0 (DE-576)399904727  |4 aut 
700 1 |a Fröning, Holger  |d 1976-  |e VerfasserIn  |0 (DE-588)133209466  |0 (DE-627)538678658  |0 (DE-576)299696189  |4 aut 
773 0 8 |i Enthalten in  |a Association for Computing Machinery  |t ACM Transactions on architecture and code optimization  |d New York, NY : [Verlag nicht ermittelbar], 2004  |g 18(2021), 1, Artikel-ID 7, Seite 1-25  |h Online-Ressource  |w (DE-627)385612966  |w (DE-600)2142607-7  |w (DE-576)111418801  |x 1544-3973  |7 nnas 
773 1 8 |g volume:18  |g year:2021  |g number:1  |g elocationid:7  |g pages:1-25  |g extent:25  |a A simple model for portable and fast prediction of execution time and power consumption of GPU Kernels 
856 4 0 |u https://doi.org/10.1145/3431731  |x Verlag  |x Resolving-System  |z lizenzpflichtig  |3 Volltext 
951 |a AR 
992 |a 20210331 
993 |a Article 
994 |a 2021 
998 |g 133209466  |a Fröning, Holger  |m 133209466:Fröning, Holger  |d 700000  |d 720000  |e 700000PF133209466  |e 720000PF133209466  |k 0/700000/  |k 1/700000/720000/  |p 5  |y j 
998 |g 1046579266  |a Heuveline, Vincent  |m 1046579266:Heuveline, Vincent  |d 700000  |d 708000  |e 700000PH1046579266  |e 708000PH1046579266  |k 0/700000/  |k 1/700000/708000/  |p 4 
998 |g 1124811346  |a Song, Chen  |m 1124811346:Song, Chen  |d 700000  |d 708000  |e 700000PS1124811346  |e 708000PS1124811346  |k 0/700000/  |k 1/700000/708000/  |p 3 
998 |g 1148482083  |a Nikas, Sotirios  |m 1148482083:Nikas, Sotirios  |p 2 
998 |g 1230497145  |a Braun, Lorenz  |m 1230497145:Braun, Lorenz  |d 700000  |d 720000  |e 700000PB1230497145  |e 720000PB1230497145  |k 0/700000/  |k 1/700000/720000/  |p 1  |x j 
999 |a KXP-PPN175296330X  |e 3899514025 
BIB |a Y 
SER |a journal 
JSO |a {"recId":"175296330X","title":[{"title":"A simple model for portable and fast prediction of execution time and power consumption of GPU Kernels","title_sort":"simple model for portable and fast prediction of execution time and power consumption of GPU Kernels"}],"name":{"displayForm":["Lorenz Braun (Institute of Computer Engineering, Heidelberg University, Germany), Sotirios Nikas, Chen Song, and Vincent Heuveline (Engineering Mathematics and Computing Lab, Heidelberg University, Germany), Holger Fröning (Institute of Computer Engineering, Heidelberg University, Germany)"]},"type":{"media":"Online-Ressource","bibl":"article-journal"},"language":["eng"],"physDesc":[{"extent":"25 S."}],"note":["Publication date: December 2020","Gesehen am 31.03.2021"],"person":[{"given":"Lorenz","role":"aut","display":"Braun, Lorenz","family":"Braun","roleDisplay":"VerfasserIn"},{"role":"aut","display":"Nikas, Sotirios","given":"Sotirios","family":"Nikas","roleDisplay":"VerfasserIn"},{"family":"Song","roleDisplay":"VerfasserIn","role":"aut","display":"Song, Chen","given":"Chen"},{"family":"Heuveline","roleDisplay":"VerfasserIn","role":"aut","display":"Heuveline, Vincent","given":"Vincent"},{"roleDisplay":"VerfasserIn","family":"Fröning","given":"Holger","display":"Fröning, Holger","role":"aut"}],"origin":[{"dateIssuedDisp":"2021","dateIssuedKey":"2021"}],"id":{"doi":["10.1145/3431731"],"eki":["175296330X"]},"relHost":[{"physDesc":[{"extent":"Online-Ressource"}],"note":["Gesehen am 15.06.2020","Fortsetzung der Druck-Ausgabe"],"origin":[{"publisherPlace":"New York, NY","dateIssuedDisp":"2004-","publisher":"[Verlag nicht ermittelbar]","dateIssuedKey":"2004"}],"titleAlt":[{"title":"ACM TACO"}],"id":{"eki":["385612966"],"issn":["1544-3973"],"zdb":["2142607-7"]},"recId":"385612966","disp":"Association for Computing MachineryACM Transactions on architecture and code optimization","title":[{"title":"ACM Transactions on architecture and code optimization","subtitle":"ACM TACO","title_sort":"ACM Transactions on architecture and code optimization"}],"part":{"extent":"25","volume":"18","issue":"1","year":"2021","text":"18(2021), 1, Artikel-ID 7, Seite 1-25","pages":"1-25"},"corporate":[{"roleDisplay":"VerfasserIn","display":"Association for Computing Machinery","role":"aut"}],"type":{"bibl":"periodical","media":"Online-Ressource"},"language":["eng"],"pubHistory":["1.2004 -"]}]} 
SRT |a BRAUNLORENSIMPLEMODE2021