ࡱ> +-*bjbjUU :??]]]]]qqq8 qb48L]]]]].q20b](b #:Resumo: Arquiteturas mais recentes de GPUs permitem que kernels sejam executados de forma concorrente no mesmo hardware. Este recurso incrementa bastante o poder destes dispositivos, uma vez que viabiliza a concorrncia de execues de tarefas diferentes no mesmo hardware. Entretanto, as decises de hardware sobre a ordem de execuo depende forte- mente da ordem com que os kernels so submetidos, fazendo com que o aproveitamento do hardware no seja otimizado. Este trabalho propem uma nova abordagem de otimizao para ordenar os kernels submetidos focando na maximizao dos recursos utilizados, melhorando a mdia do turnaround time. O modelo de simulao dos kernels com os recursos do hardware ser modelado como um Problema da Mochila onde usado o algo- ritmo Guloso e Programao Dinmica. Os resultados so medidos usando kernels com diferentes tamanhos e recursos requeridos e demonstram um ganho significativo na mdia do turnaround time e system throughput comparado a uma execuo padro de kernels concorrentes. Palavras-chave: gpu; ordenao; kernel; concorrente; maximizar; utilizao; recursos; escalonador; otimizao. Abstract: New generation of Architectures GPUs allow concurrent execution kernels on the same hardware. This feature can increase the power of these devices, since it makes the concurrent execution of different tasks on the same hardware possible. However, the hardware decisions on the execution order depend heavily on the order at which the kernels are submitted, causing the utilization of the hardware to be not optimized. This work proposes a novel optimization approach to reorder the kernels submittion focusing on maximizing the resources utilization, improving the average turnaround time. The model the simulation of the kernels with the hardware resources represented as a knapsack problem and an algorithm greedy and dynamic programming approach to solve them. The results are measured using kernels with different sizes and resource requirements, resulting on significant gains in the turnaround time average and system throughput compared with a standard execution of concurrent kernels. Keywords: gpu; order; kernel; concurrent; maximize; utilization; resources; scheduler; optimization.      k l    ( R w { ٻٻٻٝٻَvdS h5\^JmHnHsHtH"hGh5\mHnHsHtHhmHnHsHtHhmHnHsHtHhGhmHnHsH tHhGhmHnHsHtHhGhmHnHsHtHhGhmHnHsH tHhGh^JmHnHtHhGhmHnHtHhmHnHtHhGh5\mHnHtH  z { ggdG fgptu{|űhCJaJmHnHsHtHhjhU&hGh5\^JmHnHsH tHhmHnHsH tH"hGh5\mHnHsH tHh^JmHnHsH tHhGhmHnHsH tH#8P:p. A!n"n#n$n%R Dp^  777777777wwwwwwwww777777?7777777777777777777777777777777777777777777777777iI7777777777777777777777777777777777777777777777777777777777777777772 0@P`p2( 0@P`p 0@P`p 0@P`p 0@P`p 0@P`p 0@P`p8XV~_HmHnHsHtH@`@ NormalCJ_HaJmH sH tH DA D 0Default Paragraph FontRiR 0 Table Normal4 l4a (k ( 0No List .U. 0 Hyperlink>*lolG0Padro$dha$4B*CJOJQJ^J_HaJmHnHphsHtHuPK![Content_Types].xmlj0Eжr(΢Iw},-j4 wP-t#bΙ{UTU^hd}㨫)*1P' ^W0)T9<l#$yi};~@(Hu* Dנz/0ǰ $ X3aZ,D0j~3߶b~i>3\`?/[G\!-Rk.sԻ..a濭?PK!֧6 _rels/.relsj0 }Q%v/C/}(h"O = C?hv=Ʌ%[xp{۵_Pѣ<1H0ORBdJE4b$q_6LR7`0̞O,En7Lib/SeеPK!kytheme/theme/themeManager.xml M @}w7c(EbˮCAǠҟ7՛K Y, e.|,H,lxɴIsQ}#Ր ֵ+!,^$j=GW)E+& 8PK!Ptheme/theme/theme1.xmlYOo6w toc'vuر-MniP@I}úama[إ4:lЯGRX^6؊>$ !)O^rC$y@/yH*񄴽)޵߻UDb`}"qۋJחX^)I`nEp)liV[]1M<OP6r=zgbIguSebORD۫qu gZo~ٺlAplxpT0+[}`jzAV2Fi@qv֬5\|ʜ̭NleXdsjcs7f W+Ն7`g ȘJj|h(KD- dXiJ؇(x$( :;˹! I_TS 1?E??ZBΪmU/?~xY'y5g&΋/ɋ>GMGeD3Vq%'#q$8K)fw9:ĵ x}rxwr:\TZaG*y8IjbRc|XŻǿI u3KGnD1NIBs RuK>V.EL+M2#'fi ~V vl{u8zH *:(W☕ ~JTe\O*tHGHY}KNP*ݾ˦TѼ9/#A7qZ$*c?qUnwN%Oi4 =3ڗP 1Pm \\9Mؓ2aD];Yt\[x]}Wr|]g- eW )6-rCSj id DЇAΜIqbJ#x꺃 6k#ASh&ʌt(Q%p%m&]caSl=X\P1Mh9MVdDAaVB[݈fJíP|8 քAV^f Hn- "d>znNJ ة>b&2vKyϼD:,AGm\nziÙ.uχYC6OMf3or$5NHT[XF64T,ќM0E)`#5XY`פ;%1U٥m;R>QD DcpU'&LE/pm%]8firS4d 7y\`JnίI R3U~7+׸#m qBiDi*L69mY&iHE=(K&N!V.KeLDĕ{D vEꦚdeNƟe(MN9ߜR6&3(a/DUz<{ˊYȳV)9Z[4^n5!J?Q3eBoCM m<.vpIYfZY_p[=al-Y}Nc͙ŋ4vfavl'SA8|*u{-ߟ0%M07%<ҍPK! ѐ'theme/theme/_rels/themeManager.xml.relsM 0wooӺ&݈Э5 6?$Q ,.aic21h:qm@RN;d`o7gK(M&$R(.1r'JЊT8V"AȻHu}|$b{P8g/]QAsم(#L[PK-![Content_Types].xmlPK-!֧6 +_rels/.relsPK-!kytheme/theme/themeManager.xmlPK-!Ptheme/theme/theme1.xmlPK-! ѐ' theme/theme/_rels/themeManager.xml.relsPK]     G %{g@@@UnknownG*Ax Times New Roman5Symbol3. *Cx Arial;. *Cx HelveticaACambria Math"J'J'_m!)]}  00([{  )0 0 0000000579;=?ACGY[]b$P %{!xxResumo HelioHelioOh+'0x  4 @ LX`hpResumo HelioNormal_WordconvHelio2Microsoft Office Outlook@Ik@k.@(/._m՜.+,0 hp|   Resumo Title  !#$%&'(),Root Entry Fp..1Table WordDocument:SummaryInformation(DocumentSummaryInformation8"CompObjy  F'Microsoft Office Word 97-2003 Document MSWordDocWord.Document.89q  F#Documento do Microsoft Office Word MSWordDocWord.Document.89q