BibTex format
@article{CMS:2024:10.1007/s41781-024-00124-1,
author = {CMS, Collaboration and Hayrapetyan, A and Tumasyan, A and Adam, W and Andrejkovic, JW and Bergauer, T and Chatterjee, S and Damanakis, K and Dragicevic, M and Hussain, PS and Jeitler, M and Krammer, N and Li, A and Liko, D and Mikulec, I and Schieck, J and Schöfbeck, R and Schwarz, D and Sonawane, M and Templ, S and Waltenberger, W and Wulz, C-E and Darwish, MR and Janssen, T and Mechelen, PV and Bols, ES and D'Hondt, J and Dansana, S and De, Moor A and Delcourt, M and Faham, HE and Lowette, S and Makarenko, I and Müller, D and Sahasransu, AR and Tavernier, S and Tytgat, M and Onsem, GPV and Putte, SV and Vannerom, D and Clerbaux, B and Das, AK and De, Lentdecker G and Favart, L and Gianneios, P and Hohov, D and Jaramillo, J and Khalilzadeh, A and Khan, FA and Lee, K and Mahdavikhorrami, M and Malara, A and Paredes, S and Thomas, L and Bemden, MV and Velde, CV and Vanlaer, P and De, Coen M and Dobur, D and Hong, Y and Knolle, J and Lambrecht, L and Mestdach, G and Amarilo, KM and Rendó},
doi = {10.1007/s41781-024-00124-1},
journal = {Comput Softw Big Sci},
title = {Portable Acceleration of CMS Computing Workflows with Coprocessors as a Service.},
url = {http://dx.doi.org/10.1007/s41781-024-00124-1},
volume = {8},
year = {2024}
}
RIS format (EndNote, RefMan)
TY - JOUR
AB - Computing demands for large scientific experiments, such as the CMS experiment at the CERN LHC, will increase dramatically in the next decades. To complement the future performance increases of software running on central processing units (CPUs), explorations of coprocessor usage in data processing hold great potential and interest. Coprocessors are a class of computer processors that supplement CPUs, often improving the execution of certain functions due to architectural design choices. We explore the approach of Services for Optimized Network Inference on Coprocessors (SONIC) and study the deployment of this as-a-service approach in large-scale data processing. In the studies, we take a data processing workflow of the CMS experiment and run the main workflow on CPUs, while offloading several machine learning (ML) inference tasks onto either remote or local coprocessors, specifically graphics processing units (GPUs). With experiments performed at Google Cloud, the Purdue Tier-2 computing center, and combinations of the two, we demonstrate the acceleration of these ML algorithms individually on coprocessors and the corresponding throughput improvement for the entire workflow. This approach can be easily generalized to different types of coprocessors and deployed on local CPUs without decreasing the throughput performance. We emphasize that the SONIC approach enables high coprocessor usage and enables the portability to run workflows on different types of coprocessors.
AU - CMS,Collaboration
AU - Hayrapetyan,A
AU - Tumasyan,A
AU - Adam,W
AU - Andrejkovic,JW
AU - Bergauer,T
AU - Chatterjee,S
AU - Damanakis,K
AU - Dragicevic,M
AU - Hussain,PS
AU - Jeitler,M
AU - Krammer,N
AU - Li,A
AU - Liko,D
AU - Mikulec,I
AU - Schieck,J
AU - Schöfbeck,R
AU - Schwarz,D
AU - Sonawane,M
AU - Templ,S
AU - Waltenberger,W
AU - Wulz,C-E
AU - Darwish,MR
AU - Janssen,T
AU - Mechelen,PV
AU - Bols,ES
AU - D'Hondt,J
AU - Dansana,S
AU - De,Moor A
AU - Delcourt,M
AU - Faham,HE
AU - Lowette,S
AU - Makarenko,I
AU - Müller,D
AU - Sahasransu,AR
AU - Tavernier,S
AU - Tytgat,M
AU - Onsem,GPV
AU - Putte,SV
AU - Vannerom,D
AU - Clerbaux,B
AU - Das,AK
AU - De,Lentdecker G
AU - Favart,L
AU - Gianneios,P
AU - Hohov,D
AU - Jaramillo,J
AU - Khalilzadeh,A
AU - Khan,FA
AU - Lee,K
AU - Mahdavikhorrami,M
AU - Malara,A
AU - Paredes,S
AU - Thomas,L
AU - Bemden,MV
AU - Velde,CV
AU - Vanlaer,P
AU - De,Coen M
AU - Dobur,D
AU - Hong,Y
AU - Knolle,J
AU - Lambrecht,L
AU - Mestdach,G
AU - Amarilo,KM
AU - Rendón,C
AU - Samalan,A
AU - Skovpen,K
AU - Bossche,NVD
AU - Linden,JVD
AU - Wezenbeek,L
AU - Benecke,A
AU - Bethani,A
AU - Bruno,G
AU - Caputo,C
AU - Delaere,C
AU - Donertas,IS
AU - Giammanco,A
AU - Jaffel,K
AU - Jain,S
AU - Lemaitre,V
AU - Lidrych,J
AU - Mastrapasqua,P
AU - Mondal,K
AU - Tran,TT
AU - Wertz,S
AU - Alves,GA
AU - Coelho,E
AU - Hensel,C
AU - De,Oliveira TM
AU - Moraes,A
AU - Teles,PR
AU - Soeiro,M
AU - Júnior,WLA
AU - Pereira,MAG
AU - Filho,MBF
AU - Malbouisson,HB
AU - Carvalho,W
AU - Chinellato,J
AU - Da,Costa EM
AU - Da,Silveira GG
AU - De,Jesus Damiao D
AU - De,Souza SF
AU - De,Souza RG
AU - Martins,J
AU - Herrera,CM
AU - Mundim,L
AU - Nogima,H
AU - Pinheiro,JP
AU - Santoro,A
AU - Sznajder,A
AU - Thiel,M
AU - Pereira,AV
AU - Bernardes,CA
AU - Calligaris,L
AU - Tomei,TRFP
AU - Gregores,EM
AU - Mercadante,PG
AU - Novaes,SF
AU - Orzari,B
AU - Padula,SS
AU - Aleksandrov,A
AU - Antchev,G
AU - Hadjiiska,R
AU - Iaydjiev,P
AU - Misheva,M
AU - Shopova,M
AU - Sultanov,G
AU - Dimitrov,A
AU - Litov,L
AU - Pavlov,B
AU - Petkov,P
AU - Petrov,A
AU - Shumka,E
AU - Keshri,S
AU - Thakur,S
AU - Cheng,T
AU - Javaid,T
AU - Yuan,L
AU - Hu,Z
AU - Liu,J
AU - Yi,K
AU - Chen,GM
AU - Chen,HS
AU - Chen,M
AU - Iemmi,F
AU - Jiang,CH
AU - Kapoor,A
AU - Liao,H
AU - Liu,Z-A
AU - Sharma,R
AU - Song,JN
AU - Tao,J
AU - Wang,C
AU - Wang,J
AU - Wang,Z
AU - Zhang,H
AU - Agapitos,A
AU - Ban,Y
AU - Levin,A
AU - Li,C
AU - Li,Q
AU - Mao,Y
AU - Qian,SJ
AU - Sun,X
AU - Wang,D
AU - Yang,H
AU - Zhang,L
AU - Zhou,C
AU - You,Z
AU - Lu,N
AU - Bauer,G
AU - Gao,X
AU - Leggat,D
AU - Okawa,H
AU - Lin,Z
AU - Lu,C
AU - Xiao,M
AU - Avila,C
AU - Trujillo,DAB
AU - Cabrera,A
AU - Florez,C
AU - Fraga,J
AU - Vega,JAR
AU - Guisao,JM
AU - Ramirez,F
AU - Rodrig
DO - 10.1007/s41781-024-00124-1
PY - 2024///
TI - Portable Acceleration of CMS Computing Workflows with Coprocessors as a Service.
T2 - Comput Softw Big Sci
UR - http://dx.doi.org/10.1007/s41781-024-00124-1
UR - https://www.ncbi.nlm.nih.gov/pubmed/39248308
VL - 8
ER -