From 2f04605c2607edb3cf245d6af58ceb13aff27453 Mon Sep 17 00:00:00 2001 From: jean-claude Date: Fri, 27 May 2016 11:38:05 +0200 Subject: [PATCH 1/1] corrections --- mpi-energy2-extension/Heter_paper.tex | 16 ++++----- mpi-energy2-extension/review/review.pdf | Bin 138154 -> 138674 bytes mpi-energy2-extension/review/review.tex | 45 ++++++++++-------------- 3 files changed, 27 insertions(+), 34 deletions(-) diff --git a/mpi-energy2-extension/Heter_paper.tex b/mpi-energy2-extension/Heter_paper.tex index 5ef2729..a72c553 100644 --- a/mpi-energy2-extension/Heter_paper.tex +++ b/mpi-energy2-extension/Heter_paper.tex @@ -107,9 +107,9 @@ -\title{Optimizing Energy Consumption with DVFS for Message \\ - Passing Applications \textcolor{blue}{with iterations} on \\ - Grid Architectures} +\title{Optimizing the Energy Consumption \\ +of Message Passing Applications with Iterations \\ +Executed over Grids} @@ -143,10 +143,10 @@ scaling (DVFS) is one of them. It can be used to reduce the power consumption of In this paper, a new online frequency selecting algorithm for grids, composed of heterogeneous clusters, is presented. It selects the frequencies and tries to give the best trade-off between energy saving and performance degradation, for each node - computing the message passing application \textcolor{blue}{with iterations}. + computing the message passing application with iterations. The algorithm has a small overhead and works without training or profiling. It uses a new energy model - for message passing applications \textcolor{blue}{with iterations} running on a grid. + for message passing applications with iterations running on a grid. The proposed algorithm is evaluated on a real grid, the Grid'5000 platform, while running the NAS parallel benchmarks. The experiments on 16 nodes, distributed on three clusters, show that it reduces on average the energy consumption by \np[\%]{30} while the performance is on average only degraded @@ -192,7 +192,7 @@ This heterogeneous platform executes more than 7 GFlops per watt while consuming 50.32 kilowatts. Besides platform improvements, there are many software and hardware techniques -to lower the energy consumption of these platforms, such as DVFS, scheduling \textcolor{blue}{and other techniques}. +to lower the energy consumption of these platforms, such as DVFS, scheduling and other techniques. DVFS is a widely used process to reduce the energy consumption of a processor by lowering its frequency \cite{Rizvandi_Some.Observations.on.Optimal.Frequency}. However, it also reduces @@ -1229,8 +1229,8 @@ The experimental results, the energy saving, performance degradation and trade-o presented in Figures~\ref{fig:edp-eng}, \ref{fig:edp-perf} and \ref{fig:edp-dist} respectively. As shown in these figures, the proposed frequencies selection algorithm, Maxdist, outperforms the EDP algorithm in terms of energy consumption reduction and performance for all of the benchmarks executed over the two scenarios. -The proposed algorithm gives better results than the EDP method because it -maximizes the energy saving and the performance at the same time. +The proposed algorithm gives better results than the EDP method because the former selects the set of frequencies that +gives the best tradeoff between energy saving and performance. Moreover, the proposed scaling algorithm gives the same weight for these two metrics. Whereas, the EDP algorithm gives sometimes negative trade-off values for some benchmarks in the two sites scenarios. These negative trade-off values mean that the performance degradation percentage is higher than the energy saving percentage. diff --git a/mpi-energy2-extension/review/review.pdf b/mpi-energy2-extension/review/review.pdf index fa094326cab6529b6734a3e5c8d1bb13edf9987f..61518cfdb0b16f40fd0f3af6241f9666c95da34d 100644 GIT binary patch delta 31721 zcmV(@K-Ryiw+OPk2#_QKG%}M>87O~sSWA!NHVnS+ugI+mFi9<2e(j~$7JY2d9*PdI zD7HOlY)#^hAE{(dHvhhSNVYTZUM3<%Nfh<*BR$dXw2O8>zl;3u_mA(s{Ye(PI67uo zobDc3eC+p!I4Q)RraJBqD$0LE`T&|V7aw`zd6>}KI4MQb?_+SyHm9lSPpl(H$xc@z ztMCr^YrdFj8fVK7?P;*hc<$IBA-F6Pu6(zE1by6+deMDFtIS6^~eS{ywfmQmO6k4$hB$N4o`pUYsZt`2PK}! z$T+{nXmxE40WWK>2Uwq=krb!q6;Cofi*D#>vq;o>)EIIVi>GPi*{gR?3e9H?VaEF) z#p^zWi`qs%H7!DT#NBV-aU4GF3n{$eOF!VvuPjwVcbP(<-+1e8ZZsWSR>_i*1tXzp zU)JgNiV~~5ojVUGUcG-r*TNC_mj?g5bNeC|@KCu_;x8QIW6-u1*FY#bxQNkivIH8` zV0JF4xOSE=LXQuQV#+iW(X8O~!hz|HJ>7V^g6x~RORz^^Tb>zuziviT&;v9fj)*Op zir>umjl1(C^ZfAkFefW2k*;$ zA^{zsE0As0oTgS=7G~`ENlNqAeee`CZq!e$;c?&f&~)h z;N%M&L0>mDD|^^v7lZDN=v&J5bHAAj5(kf=BVg+%URdd)`DsZLkJq_%~4 zxYK>`n7Esqn^TAjK@rFeL`-5Srpkl`n-t!37jCT(o<09EY!Jzz!f~)y6%Hre#nW~J zc=lFY_7Eb3P-0Av<0Uh;!=srUt!ZHi2e$PT+J+_o1%iLT)Zt$bO3%QU#m_uk66iP| zwc^xY{C>FDa}KF@BI=Uh?poFWE4&(k6ChU>nAaTFdgbGe`|hhj9C&3T~&jvIo+V;Au0hiM*|dV^2T{=ewn{P>ZpCB zD1p@UZ`Ll&6=-J&K5S7Gtf9A`L!;ocOS1;I`GS9kpl%cGXvH{O7gB_9ngM)g}~1ILHGy91ET(qPc7ln8<8}6EwuKeYfT-eLPN3#r2D} zKfyx9z5BO<%;IANAMbH_EQ&n*8)R+rsQQ?Q>9@@fDh1@loRhJa+RG;NiHAOK43{`hP8Z*`B<&|>(eiAI1=UBLoOuH z-2fVmetbZ*JMN<0zi$5VaPznKy4b~0SVTpD@!M`xxBre}DK9 zB+Xh<8y!ZbfQd*(qs+~_hnqitZepNCyZADfHld1&UD@7z{5{%L!2EF+g}N}iClYO8 zG83kWhIX^NzxnS?bfsawi{r3JQ$-qT6~;-i%cB_nhFAZ-8=v+%4&3k^>l&x)F{ug@ z6MI$fy8S_AS@6C-&V%E2ehdHH1k;HKY2Z3H9G{t4_T4zQ=c(>xWdF#2f053DPdFOJ zlzaG{_HoeK;q%D6v2VT{P^;ZR^f<&xm?Rl%Q+LPx0eU=HlH@_z*l}dq4@|LL#rNJi3Z0T~7jkl^dD!WuK<7X6^eje!{t)_R0hfbRbTg@eH&#BxuEiS_H@* zNLcL%=}&0iy4lp6$G|y`^`Furo&CvT<2>}A8rO2C9{Jz(6E=#Vg~@SGCaCJ)qBtR2 zrA36c8r)+)IPer*#FzQXzCF*AMei6lb;9Xj0LYO^=dLbAn&?x1-8xv5dMX&~gxixb zwNRTLZTGa7vq09<2`z~&G#vidXMue>dTj7<(R|%FIa_V1Np?~S&ILA{=p^`+7C_Q; zA~Pw{4pqaZCylLb8nlVhK|L|8b+#M*SchsL*K$LNv7zC*em!?u&S)EoK%K`y)stU$ z%o2}?g2%!AF+(zVW+^{(;s5i)FT#WdAGA;!WeFZDn_S|5ADw79~y)4988X7-;zdyvm zJ$F_q+XSb>mgEWGHM3oU%-SWj}}_Wivxg9mc<7 z)=Yzcb4rl?U@EV;z1k8X3M?*+KalX#hdUM%EilA~a-D=dVkT!bo%zYPWH zP^ol{f^-7j=t>4QI??`I04_ZFv;iX2wjMh$jqDG9v|++F6zS7q0qC&B1QziCpW>ih z+D?Ff5vz7(_9f*gNQ~pk_bVnz0fd%4pPOe&%Gi@|ryO_Yz~K6qE70?Z{yxBowoKvt z0)P!X#UA}gi0@@Sd)`bA=Vo13jQ0|Qsmcv&&H~zy*W#Q`^p>nVZb>pJkO|a@(IVD=k&gR%mgEdU-{C5lZ;@l^&#(;ti@#te z3l%p6nmF!(dvDsTnoDvZWeUl9QWcTVxhHPCj54I18#}=!TFm?aCGG^r@Jx?9(_&|y z-ZAwIn;bZKU^du1%B3zLP+ly^UCY=En{{`E)KkRNU+R7qwpa%3oTqRhB4R(VXyQ?S z_;q4RIMj^T4tnmZD`&LQ76>6|3^8Pvh|=nHR?><$0T`A_WvU736JzQp+lb6I%vS>Q z^$TS&F#s%Nn66$iH`L|4LH?65=)*=U_7!-&_AWG6MdzS7!z1ev>uo(W&)&BP_KD0% z8W^sPwAl)Yzgu{@jbPcz=MtME#&5WP48km{c+h`3QpE@+z=j?8JRtJRiE=tpaes}p zFQ&jKeq{l0*vn1H1ARKgOmgm}Ez zU=$0oP8{?3#egk!HM_D0JPCKlrk?~0{T8JW?sbGl%tF=&0Nxfu1~U(*1#NNG0J$Y= zn03}r@;IE%{6X+)9PE`WVQ(>#}i+NF&h>MKN!IOeSPGHu>i(0T?F?% zKE*hu{)FJ4U>1(S+6Q-qR-QzbdJVj5oQ)JKEA;XPgl*ag{J~srBFFJ07t~mT8f}k< zr(@+Wv9W#)RDnjLSWsM4W3hIxNAv?7R2p$#`H+(FiwJN$LK^E&6kjfZ{j-Yk4}vpqn(jla zuK5df4XaBI1(5>-@KJ#|au+djTfDqxqDt3?cXg43K~cWkBO@&H<=`wNm|IT4%Pj-N zPIdR$H|^wG!d*fsGUOqM>Ylkv_{omkK_hIGLwT{J3nxEoVUbvW9NFI*m&d;gOegQB-=n0qXGB{Q-x zCF=SD9bpFKBqt{lFZT5{FpB(T$@ohLkl?c9Z%QcI&yxqrh{VS?v);QOKHybjc@-}$ z=18E3(U65R5YUq^=k*$3-^emNs>jJ!M*@`>3j6BcjQWX7ll|`FRrqe=Z8* z+g@q4Pog?F^_$N>ML`3T-v`k?t*hXfNPF0<+UHpca2Ncr`THii=P=HLINsNJo)Cv= zvX8SmD5Dtu6`ww}db`V_I&`WRXfnOtZ+AtNg&+U;#JFbM?vg6h+c;T)UeW8xf0`5HpId-n!S8s4W%z8SmLXCj0#<)D`kAC&!d};U`h=Q1C8z!a`3vX;t-84 zs=`A{8>DDyF`3+eyH0vIf630s7XGJJ5La~tM)A%(rVA;zrlWo&%Eq7t=?3qYgwx3X zQ>Uyszv2K)0!}8Yz(H@^JT%(633-o0(lVGbu#03`s1!>q;r&x>nQ6ovnZpTLV+Eld z4{JP*vqg6+Kx3ED8e*gu3`!0G)_9eRdI1BPYvpo`Al`vONDEZEe@w0E3`xDt@NP*G zenPt&?HZ4GEKH(}<(`{753g2Om`@W=fveDr*FVC13z%g6m!aT$gl9MLNF>ITy%|hz z{-YbRYKkD8Or1v%kd@QsM{aZGRs`AhusV$^v79)6qbI~1Inv!6Lu)j*xVSAEM%$|n znc)dW1Nc9dot8Sboi%IZEb=^o7?08J3$`c}C9->F#-|0@HB@nLF9w``N z3D29WPLANq&oATJQ{&!lxy*D7$*iU51UN>6iVSM#(yu} zT>>Id^e-jOAmXMx8V+D#4a1OO$R~MZCSGY|JkHbo-L)(_dY`^O0}3C7e~&PL&@cW($mg6HgcQ_5i36^k6f?J9_Vz17t020g| z$|`2FdPDSimw|Eo4NsyliQApV{@Q7hAE;PaIfW%Vf0{+C#O%m_^a|5Ry%PA7?`xXh zGT~G5fw%i>9#ervDagx1d0=X(kU_#`Tn=Mb2dLwc6Jz@q$7G%Pvebd-lp-uLsw?16C8o^kp)ttA#bG9h6D(^H?-k{@GqT4~HJn(V!CeX!I)Pwe}~r#JNs!Gt$XicS&@L{rmz{5NC6*B;hcIGlL z%oR%|Iho%F?$R>84?z9fn1;z-sn48waEnqnlBWX%t-f~VL;`i+J?5IbqUInO1+GZe z47?5_kH2mq$3r8hj;1q{PohBJUm$4Fz&Lq&uso8|gF4X&toY&W0ao2^00TSAe;!h6 zPX@x!YW)f#!5Cat()pBic)?;0Fb`h| z#qoEjgBnu93SVcf!Bmqv@wE~_Zl-|T@(4tk-;0u~x*A&z-8fl0^8ePZ!2k_$mcG4V z6gV~ZZgmYg6_odEoX{FM+Bkoxe{{!CmaM)Ay^SpjODx+jWdX8@Z#xn@6mvGtlEHhe z+-&t?roYZ}O_gRow0u-vhG~-jqF}AK3dA7e7?AdapE_g>s+_$V+G4pAkHne~l#r48 zkb!|_x4hIKM~}%}%4%?rD_FX|(@&mBiIMtVJ_|9hTtHB=gbFtV7J>(hBE@tscy_>U z_AB`bAme*f-0WxJV%F|4{P_RJ!{)!}GnLMh=O8JQVGt^R7|D{OHuUbV;E1c*gw=*z zQduUGnN%uEcPfWT4l<~1)d2D!7J-F)}3`Rd24 z3X(Xg;wlLq9$}eAWtj(uJdNTs3m%%_LwL9S=ixn2mOf=xW>H*La1hZHNV9qKu=%t} z02T+ya#mJ_P7R+Yh(NIN*JjEX#ibsOAoe&58tgh>%3NmLa@ zN)l#il;l-#h!gmheEYrXYq{NJX&hd+X&Kt}S!~lo_!)l?l2lC-*FfaL`suLsE(dszXux6cfkj2sLv<&)edVLds4_u*w=d1`5lv{Wp3foF9L zX{G@=Rz&3cvKWqb>QJ5RQKr${Au0)b&0nKVz4|*&@>=w0f|0?Ww*@?HP~xasHL~XE zNYyicfp>>Q*-1Xwj*3hH6#CIZfO17hak;dr7RYHl1q!?A^{8suMC6~&uYz3~M`e=0 z7fy<;7lC)*QhnIvRUEzpiSptQnyKr=@VqS(hzQ0Y0-g$1{Kz!0Gz@fK!3cIj;ASMN z2n=i3S_mYu^(N2Q3i?CLZ z`+Y2WFP(=7jg14t_$)c zcv~Y4?;rKRM5;HVfB*;LgdwBmuaS}$XN?NI@=L zZoQ8d6py&#OeaCLpw4uk&nXv?jpi?ZgdH#D96XcEM}i{sG}LHsW?8utG|f@JvqIil z49!!PY6<9zHWara{5^@Qtp*`tAjpvRjeA_YsIkaVO#R!hr8OGGQ;|N?% zgY@KEtgu0-(UsCYrUXccuqs1zbo}<#e$qtj=|7POoa?P1sMPU$2xp5JrcAhs;D6S~rOUu}qaYs?=9H z!mp}r8FA!{#v5fX5xj*wgI+FI!-xQax{#{jz zCAQE}4Id2yUilTchF++DKP!1fdJxd>k6yLrAXa<;k@itq+B}%7wp71pFDh6mWCLEq z)m-P)5unrD*&V{iL3d7{keH?4F;lqv>T}f+&aj=uH2eD^P;7g^%BpTv-s{P1I#AS4Z)LHKBSb_kyBOObR!c@QhuCwBi z&Ni*o#m*hWEKI^#9KUDXEw4wOx}9Lc94se=D=y|~0Rp3!_2|wyx<(yX#k#5+Gz^p^ zF-Wvp3iFyAZL?9@3JE|@cBCeg1K6~Brlq`|Ca#@vqc4JgT03Rpz=4S63Ri?vTQGoF z3jZei;%H6!hdDobMGo%)6NsMIqBVLg2WX4e*wcT}+veu(Ht3fr(Z!iRC1$G`;s=Sd zZ%-0KuySJ$ladVzu~xYEGs+!&S{@)e#$;IjtVOV#oz$J|2+9HRa^s0Ki!Y zXUVj{c}@&G!y|?Z6rMEq=U5tgH7dvK+MP8aC$0s53P>|(yayL7>>h>d8L%gZw)8TJ zV=EZ}>f&zcNp&bYT;4N=th}(ocnr8F7$4?upWXd zSJL2Ke02zyI~ScXy0{#LhmP<7Sz)3 zdr9SGl)QP^{0F{ru^E$UDJTLtIFk_}E0ez|2Y=|O1!7eX%Ax`kK_A&51=^zBMRQ1U zU~I->Doc(eXLs}KJ1@$Tqixz;EOIy;&J5@A(Qyz5Nzm=#gM&X`tl|JCFIGR^uD-j@ z${>lOGA@(g_B{;KNNW`oDvjba3vPG8>+s3d+uL7&Q%^ZFokg)OK@jn5kZ1M$cJ*h;?=*umm2K^$eJ4(^Ec0K#;Xs|?06+w0l0;>m zr=)9^Mu{qeB2M6|S&#NAEy7RuW4m>Cm4vo7_hxWybG25f3Qe=aK^AV?=EG@k2FufZ zXLimuxN%_0n>deudI7%)&kuH}+N(7%?0+sq*CVPLZ4LD&3Ud(k7zN#$m_Zv585)=B`Y|#~e^K@_r&)G01 zI8^!b3pgK*^}0_jCKsZ~cNNNe;K@{cO5!nuc=CPBJn!31R$-M0 z5`c_GmO8sW0&Cq41BDBUZJDpHC~5AN;zEv&r;bUBlKdWF^vB`LhB-!7)B=t|MV@eA zf+jkYRnzyU1FL|0%pfCjB#KU$z&6EG4Ma_p)tz@mOM2h5aC8=H+)l9&XMdsbgQn&0 z^s-?pu@rOav-jNKn*Dj1s{zg^++rnxNy6sp(IQ6&tiGeHwp(-Rh2IT}dDwG|aiVy6 zpa#z*^n5GQ>;-&wr&=A12~hDXyGC~TtTMoJAh z<0A%8mWN67Smum*`)K;Wc7LBfD6=wXA(wxqO!V3@<@#XmS(C8@TsM$awCL;LsQqB? z|C@IT>4S^j8Es8PX^usZ|A4x-Cs7DIZ+PmFZ`CLR~Zmkw`u4 zbil0`brOvj6YqU&af`A2vu}5NnTNkoj%_k7G+*s()ZUg|#(+IxZ5hAdO{^ z-UCD1@YZdMryNh~MPdDi-hZ;B4AuIhBs!OZ7;k&tD&pmM zOno$^n8pvy7&+}pmQnrgqSk__WsN2tNzq7bIy@HmmxHZdRZ35|jfNLQI&hf5@KS&7 zp+J9!K6$ucZmio^=eF9OI*0z=`%``9d=2ryi#|_NHbZ{}@d)a>1h(`7d^)d9lS;An zkCL@goPQ$j$sh=Nuk4?^>PL};@9ZG|yQCANz^1i!(v4O8fBO2~>frNl3N)Z-&nVIM&1X_5u~0n>{b!lkaHu!mjb5cM;bY zH@5cw9vBH|*Izsxz9=EG&d*G|Oh>(-J&As2 zl{J!qh;uIs1^gAsG3dRnztr7fyNjaG@A~adW`3!$EBx9J*^G75P;=k6$J59Ygg;iA zS#($Gf!bAJcyud&{GOUmTUfWKN8)(wnUx5@*;D^xDB`X*6{-i$vJ-I-$0^BLkFac< zI2nqu?mHsj>^KpAvENGNe}YrMW>1YS51JV&Zs2*3)5QLfCZ5MzZA@{{!FH#@%)hNb z11pk@b&JQADQO5c+ZPRNM>Nt2G)#j6W&S6o!>qvC-66_Er>0yJls27gk@;=$l_;0WW3J9w=F*ccZS-SjNb=z|$OVlLTP1zmMTCBF z7&JGZ>an60$sh!P?+cuv);yo5bmavbxDr}t)ZL!fdbLm)PdsaEx2A!aX>&5|D{7Jg zi|$}?!iK{Cs}&Qtxzs{$ka9W-yZzo}T z$YCqL@T0q33|p18*`EyJPvlOWIZz>}M0#-Q%>4)2ahzn=9gXPE;Q4VB`I7o<9JiVb z{o|lZ03mo2qX%M!J8?t2H7_>kM?H{n+!^L7KzR%);V!u+l+u1kN$*qnpmw_vM5KhV87-itWYt`EU14N>e6y5m<<-c ziaZJ1oDTxVLkis=R85Np%ob#`;^mvcq#{8X!V_@aRXVwW=f`1#paKHwiCOwiRFKNl)erB9Wtri#K;?-&O3BJ5yOIS?7sYI0da zeT$1Z!&xoZsS32!7mGQfoP(?>S~sI_3HiG;4H7B%#0klEt{_6?ll!|k0w5R+`k>5z z{SRou!r(!KJcIQDSGdF^9BTSK^6gjVdDB>yAx@)x}=bBhG?;-{FXIe<1eLO=6*jA z(d{Q^`$Z9i%EfknthtT~!c@(-1FZ?fp-3s9q9*lMy+3g{07Y|Rgn?P4wU`%q90(zn z7Mr6`=8A%s%ySH0*vH^St#@4aQ4j4WQ5@nfHP*-oZp7Mb{*=yoA}o;sgkl(fs6?YK zrE}$Qy8^On(1`qX6FgDtVAI6W_>XaMbZ zEDk)XXtyg>79K&$Wtj!{*R?u->m8=!R3tDCdQ6hke_Pmk-HD>0DmaZ5ML3{hWjTdP zAoQ2)5YC6gv*Yf}sl_WsNrVRmEi#@Hiwpzx!o?Qlt9kQ1B~+5L$H!pnxBvP#HOdo+ zGuZfz-4~^@adazsNjH|hgM!A=N4KECYwLc^*OqyZY#3t|Vrz~bmB(~{P<|c?NRLz> zDawubpIwX%&8CfEh)!a8tcX1RK&#_b#j&qdD9>Hdg4l5DiaE7GgU@2yFBF1BoW1fq zS}+2OMJxR8#67|XYP2)tfqDUX+=n5LQ?J~^?M(@^TXSuH_KDfWZ}N#7=OoS{#8q!_ z;959?SapWmYReBLBI59WJA9&&r7#x9c4S#L!{GjSq7k|-NXqC);Bv;sh0y=4cJw46 zMt9GCO|`G4gx_8DiW=j5#AO;b;GQ%NHQw<-{Uy=(;e$Fykemuv=+vXVE4%tcMi7mT z$7ZUGk;l-#=?h-(p=gS3Z#THPUTlec`IuuXCVE_fAxuz6(e$o=Q&Eou6=~A#jVgHh=?C5%l>_FBMtfLr7jP}?-+)ft*xJ*sWZuBA4cj!Wus)n z!!M;o;)(g)6U#NP)yQ>Cp0TKHZ4qMKdKD2auU*_f{+CCBRfm3*#{o#ptgiVf?S+ahf_x^b>}iLZ*h(`D<=6pNt-WtIjWpI$SroaY#nV$o$>tr z0tBg}N8@A55Bnq#1VIo4aVZM+K@@y{@z1-9XWz*z;1SEBj0bl+00m1^6(mZqNXX!> z2wsI(mw(;;2$boZQl^qcX$FieO^P%Z-`-ukyMN$-ivm9Jk|`5W7HsQ_SARu80h2!l z5tCUO+*{Hbn59fB32+tMUi{}GauO1wZtpn$;$-GJ;GAVz3oEk}jH@h2A`ah_xB00zL@^x91I-cbgeIWR zYva~dtzB)UdP}qI&aG|wp*wAdOCH+IR)qR9($Ti9Yg~P5y!1RomB=W}n}PtKIi3a* z8rE>YiY%=21{XdM%}~~+XO|nLba+!$4u5I&4ZxZDU=PsXs!Ud)LW%Ij%O4Ps;jwG? zU0(Nu9uE0H5Zq3S?C6RMH{H@Aw;>_Td$T1b$WWoSUg13=%N zx~+q8L;D5AHxB1iOan4nTHbiB$A8ikXH3Z^^4FDFQ>C*xH1wGEgtOQ~B7&Fz;4jx+ z$IM~TH`s`XrJTbToFI3sxm;2QuQn=!7t-O8-rDnXhQvfhFZztAAa5iq@2g z?tny}q?p7hMy)F5_X8^7fx-_&qnXoJ8Pi((=yOLTN(2eZAPyi!%+eT7%96~*u@>%- z!+R&N-14v-wg6V1D zi<>_P+fFaCqba)sc^q)xOn=LG`AEq9qBl*guO-O2=qw~Rp3dC(NO?sD4{}v*RoSd* zQ>ci=DgQ)uDxzx4{S%r@bsWqHtrQ%B#(NSt$3WwRuS+LxdvqEQv9!6t;!#dhGJ(+K zlL?xz7czd6aM${TuD@QB90cc)q}U4Yd9 zSu`WVsz-Wr`i8GD!efk&`uQpksYjC(D=ZWfatwdZudlD4|LmV8i#4bU_TZ^=z;9M| zkhV7asFWWQQhrQGwGi@!qmlu{+a!s}$H6ryyfwk~Laoc|l8nb9;OG#Osfbxk4w%;G z4;m!>kezIgD}VB53{rAXDT{nrS-x45EMX$ns8Pj7!JtwSLWEy~MMV(fvF`EiXQ|~B z!2fHTdh~2ECMB*LCH|84P_La0ZXm^^h-KW-9ab*_E(}aWvLuS<7ojDOtdQJ}pU6wI zgn=wNVj6f*TB=vm(w`^=he}6l{AxX>(5L4Z2izMgOz5BtJ0C$-W0^z^2OaYpq8R1x-F#K7VswXl!>^6`k8qTrDx&W5Vo@d z-|+cW9V4}YTZL~=5or!qEBew`mXgLZr6uHT-{ zz~-hg?a8GNQx`zQvqJP=FMs{|`SAw8wJ7~Ben#!uG(0aP!>|;Z%sr z@Wy;ein+S9{pHTqGuCfs&g;a~9bN4vqJJz>g(e?;D2!_KQBq1@-m8%8pWt~K+oFxW`&!ZqTCKNn7Vq6>GltV+yaMFeROA0zUrUI=M^nA2XWuppnuJ$ z;IDTB#WWLe@Jk~j!s;oX(qz5itU`tD~)W^+lA zi6G2!D6%p+$!wn9ZQgAn&_aJNy6O^H5Qn)}G@JX^p;vuZhr2kP52H>erCOzr7i(p>G|)&| z_r7Y6&es%e`sl*$CA>}x@*s&1yx`l`Jkk=SzuxW=k@$aYv&bj^#fg7nZI(}z)NkOl zXq@nEVW2*otzVR`8r70`Z5LhnAWr4?{m_#S-q!Jg(}RNAi3oFD_J^b;thi4B1M&Mj?OrJWN;` zZ#D5^${e~yb1{JSlJlm z80Yi2w!F$PT^9Wy>)J^nd5WQyJ$oUC&9Ke!oZ8%fH%JcA3O4t$uij6N&i#i@Gq}|9 zlOQmGN1TP6o0+=|B4{OR9yLlD+~$-K0k(*8|9jc;k_&&4yMcwLIL>^b8CyX&bZf;o{|6XSApV`r_Co@Ua^dFYHWu(joiIE)%kwo_Fdyly zg@S`XW>yRaAU<%M;&{r5S$MDt-436|yW7)Qe9s+l5YXGNxZMO4y9vI{nD%vTql;)< z97*WhS51HQ&%|JhFj%q3rHoxmXnnSsq4Wbmpl)PaSoBUNI6vyDWQw^Uq8dseiV*o75&@)!KADUU+)7pI4O1`Mxd9n#b(9_W*Bb;R>Jn<)xKPRnR0EJA>3WlpK!8 zhq2)KI3XZp-fz-Pn3a9xneZ1N{D=n zYK)vyX#Le5(C6`h#({}FgH%@=o&zvuCn=doz)asCaVuY@VHon%RJ}4+HM2d|>|~JH zht{KTuyrcU*X&IYs09*wXwKA_!9kfw{pX!lbFfB4Csh00b&)c51C7SxQbde(cg=m; zRw;krboF-_?SYZxndsoI3k=C04oTvi0izZSE&cJNS{`u*Kww!ev+Ev5)g84wpQ?g^ znS=pfyH$U=0*kY!`N98|bY-?Q31Hh;4QXt>*6FvD;JM3dkSn*M;`>&bnp~T$gvlKX z64ZBi72rb{YT^Vp(VnvW@;jj4Z_xl|<5SfNO6o`*Dltojp=Bn!LXO>{%)0(u!ec_}__JAEEx(yu$Zhy#7zU!T4|&Je*Jp`?E# zN)QNCdP(3y(L$w{CNraXmFcK`;$9=-cL9;Pz7a*5Pc{df-PEm#Z0{nFqN3^O($ShQ z! zg`HD$pk2_VW7|$T?AW$#+wRzWJM469+qP|XY}>Yz`DWJo|J%8%v(~vgH&wOwvyb$J z!+>2~z*}y3U%B1@Ay-N^Z=Af#o3k0Lw*UbzVhtOg`GM@rH_8FZ5EN*4XMi31kmP66 zjt8XQv5@rrYllgd0HYx;7Q{5dq`*P$kBXC}A7tsSL?UZ7#w`9)fx zSPk?7!UoJoB_1Lxb*TQ{6t$m_a_NXMZ4!<%Rtnr=7|rkyKJ_|{Dyp;xXEj(Ezpuna zt6w`pVo;KWy*AC{uQ+-YCtOEWetET7zk9Nqh7^l@%S` z@%Cf@Z@_H0FD4xJdcI?_r*7N_>%Yd}Ke2|1>Hn=*+|p+{69RcnLs;+2G0aH#|Z5r3;yiglw*{uwctL6%3mBt_Go-m0q ztXnWvKt)2GwF4lT#o}-v4fN6$+vaQf>#4k>=2&}f#s6+AVNfhJAEu&LOnpzy5-vSf zkf|gmj@m!LwW}SDqARpdnDX|1u|J+8nk z91;qP6rW6n7>x^-t|?GG3~bE(Jd!~Fw7TWZ=|mLE__>9x18EN4=lu9+$Ag!jt#8CK zKK}5`rs*rF&}d#ruK!5JyS;uBqQh*=I+=K%PrR52Pm>uvY{oIQWK)Kl0pv$QJemRo za2JrLZvL0#1BU28!3~KnHdZj0U{}#eIPT&}NzP#nwX;1F`e^$+!pm_GM%oGF^Y8mU zVL{G;pLIGoL6)1grIlT$8BLWpRqAo85sVMDB3f`wkS{uv!RXIcpC`xhc11Di8S#Sk zRBvvPwdJ7tC|sIjuNu9pA|YZ|bDLF2V0d__;cX&Uhx_9gm)g5T2VGSqA21ntmBTp& z_cK4#SbRJfr|sX(an-k7eaSERWw_)^c0PiA7*Gk(zF1}d;DLC`ljXR4qp`Z_QcFAU zF6we(jQ@bgRO77Wpef6=14dZ6s=ew8xxORQYl1UywndqjNac`V=SjKSAhE{@TwGXIt0v z!g%wOHJZh%0mrH>FsNmHD0|yz%blY{mj|=WS+;RSBhzn0D~BoST;+OvIyn@|6=naa z?1ST`&}N`V<>Upg53)sUYptXEL<`6 zFecq^A4uzgFV2xo`fq{)!}nCI4i;Z=7;67SZ@C@FhJJ!!kgS(A-)KvkS-BLuXT=HI ze;?{Qd)vTOiGx5EdUnggAE21J|46}SQP4vgkp?ihe%pI?dutbC`rm(p&zK>$!I|5e zxwtx;8{7Tg%E812E(lw>1zQ<(i%3yx9F+}_gFiiL!mosHvvPOKzM|5K7MGjXsp zCESWq0a5PeY8V$-3{j+6bE2?PNSm94bKD5v2+ZK3EsZ}nsiWe-NL5HTqu9{qNapmh zdt9cTy{5VZuhN@Ry`KEH2(P!MJB;$@=MH0=Lv@8Qc9=MkNT^z1gf#RuHHT3~5qFM` z4iAow`b0Q{`oDklqmqJf$3idQYsu9+S~@^1EKObhmZ(4Hcj7a%++nlSy~CWtSa975m+5U(BF3Urjc@VuqxJ$mhq_?(I;U|^n1Tq1oa z|2bHGc%-sJfO?K1&!}sFTk(h*eUnyvM)V@RJpZ>U%IBI@!B8m_sQ2U!Q5FpUsMLrfLA}J*sJqx(eye0V^*vJ9$Ftgvo z>k3JhJ7r&;S**bQJ#BA;A3;49vgP%b~hQ@&Y8tr{DLx zk3%W7Y4Td5ax<&#!AJ?)BdmU-M;n>468P1O!U-bC$=QO^0s^uJ_S-Yyvelo;3)sCx zc%6X#u=76|*j~oHLa6n!g4{!H3gGiKer*rrBLU$I<{jLARvZDy=txkE5FtB2D@mG_ zaFzsTiT|Nw0K0$ag@n35C^wL0VIUm8zdk-EE{vxucbO!wffE2ReL_lGa;!J@^1Arj zePXn?3#6MMPzVGXS&$QCl%5U&K*899dinM!?77|%{QNdw1QY(&-zG*)GUjmnebck% z+1m5nxOaQ~Y4qCSs|EUPy) z?%SNbWrUgletu*B_mXZ8U!2Fr4V#-pYWYaO4BQ1!1{DkZ)rGpU`s@J&%sFR*GzyKI zqX9#T1$EFf$hxJ#_JGT;CYkTFK~YD^p9uMo9wA@;z90lb_E+Dx{H4ejH#fbl0{_k# zu@8jxzOz(_ph9)OHGUxXfPe%G{qC$I8vg$M0@~dR-KHbj`L*T%Xw7Pyt^ap}4%m_;1^g@2&e^zHsydl4!K;J?8 z!kFD7e?;X1p06N!?m+LK@UI7fLbn%ifdjWUXaT9;@K4abqwwGOcl?gT*~fw#ogfS; zf;*xngTyyK>duLkKza|KIDJGCl!CaKz6Z^<7N|DEg zH$6wt?vf-#25miRPfQ!%J3ZNIFoW=5!|7RpLE@s9wjvpf<%QJz+)ICkW}fF65WKiT z;U|gP+{R*bE6GnREGNs7Y5uu`Vg9@9u^DaE>(tK_c+!8D(mkK_7#6OmHJ7!;$lGsFGiYKQ>pX%g{Ji37rXc$|YEcN@>UGJol@Qa&FBANqUGNZD z3w92HYJ7pq$T$O=FbCp3Wd|=zf+vgz`&$Q{_bZbm4^4mydHh<_6-~yY{XewH=edLF zh1Zp*y%S}=KMsFH-v2e}2=EoLZAN2fG=IuWcsO?2anIF-d0cDA&tW!cs2NK^wSIzK zh+bJ&z?~V4Vn0%eTmHSl=G>_N9v8{7DmREhc+{~Z5SzzRDQru6R|9wU(Mc6x*eR!z zx>^|jdCv|cQVs~GTkv$>QOU10KvLOa*q&bfWDPlas|(cINjyaM|2P}}B4=Ij!Vrg+ zBb7QyKBMh^|tuOH2$3H6du3Aq#2c zn((W7gKCn{wbCG5{JRVa*byp$ht7}rmF44zR|wE$l5zvv#oOyX)zV^9uWG?TNm^jO z1zM|s6CxK%bkkuK3MN@QP%jpsFR~&Un{qjszi@X_#=}itABX&l7?16wx9PP-SUbc; zM6vK-5#HgXy}J~HFD;|l?ka9>x{+kBX@lXFYIKNeRoqE&6rxHg2$jiQ2!A;JT%Gq{ ze-8WyTihpm%mcj|ClwGHP8C1Hb^vRz==Poe zfqW*+#hdV}dgLL@#J~5;9d4EUIgR$4nG~gTn$xUarl#AaTGxK*G|1u2&P18iU)C;h zYfBD1t~Fuk*Z_6lxV*@o>;sY6pjr17Eld$Irb$R)HV<15P1S!J&Gf%P&9NFFR$ z)>+m_pec4GY{gV5k9{VYsFd&~nC6QOMqSiac9pXHn}{DQ@|tQ<@3MI8+iITRK+;pn zxJVz|_5NiC$^u6lD7xX50$cFIsB^zRwF?kW0Fmu`93^sgom#sp zo}jnmN$oI%4iDR|o6mJBQ8=-uAI82ta|X~O-R`m`+G>TeXIWT(;UdjOhZ!$L&<-EB z%3)R|8R6P1Q_8n@5Md{0y(2og%e``h=ao5L&2|ANW;&PROny7oCV<^SnM60-e-L&UCsc6cp)VKMX9+ zwH*Q?7q{t*Tqy~;r4AI>#>xmLnNmz7s$yHiy3B7OY+oXhz-G$J7%VuiRQyb}#2tL_2)%j4 zPNtOc&USE6Tl&;tJ$&+4-H_q#90(0NJ<#f%$2XPAlftP)5pPRONPvmIZW|rl5=L_G zXNLR2VGdFG*`q7Uo;Fth?3J$+$+Ss^*!CYk{Z}lj30Se6<4pX3S!l!?=h;iM5u6SP zpWL}ziXXCI#lU5WaIv!d6)%z0GWd|0R^8Ppz)k;Y&;33|APaY8L%^a9=Jr80KRZ^9 zH@$a+b&u}z)Bk3-j%|D|8$F$Y{-pp4CLK*Tmu0~Z#~dw)oqwnYk9;o0tHIV6v%MsP z+xB;IJg|M#A*j;D8O`|Hey`MTV-n^_CjM#w0HYI<7#HV*W@a-9UK|l zz@n2Rz2tKUXs&Hj{E~tA)I7uUs3T|k+Ve|9iifj-Y};dd5&fsiMU&2~9Yx)~QvjM% zy<&8#@Y89B{F(a$Sz}#zci7EdLS4lxDk4(l2ry04pkjRHv~P~am>U+NjrY~$XBj@j z_)4`*p91w+^0372ZU2$*J_kARweTmMx!97>uXVl2-|*0uux;zY3YwCe-PV*Fy=1&w z^bb*h@Lt#D%WlBBMK7@tyM;9iCBs$5i=k>XYcQ>mt}9F6OVMAl*`%)=EC_`3WC(K^ zZ~*P|iu=K5!kY845SNdcSxjD3KG!baxvL~5HAzYf_C$e}b~BN}MKKk9`uP{2O<@Ge zdS3hbIsSY*5`A|X%aIr$o#m3t8#YcqTy>qIP~3SKm$GQOcIV89Qi)!$g5laIXFSc@ zM&o&H88r!QUeAB$Z!bA&b3WW`;k;js2M|e?`!kB>UC5@lCH-NdTFl2$qINf9}5)7{fgNvY+ZV_(U_BT0eCB z0%HJFy(daPoX-&sncFy*;be);HY!xQ)$s+!l%7i?a<6xu2W)il-=@XrDggqLqI<44 z=_+h)In}u@rYstvE!-Qu*tM_m|IH!npn3TR$(c5Egb0RfimT(l(hNF)v`Oz%#tsyj zvqh(e%scML>8H^=#X9WSu$r5ICT?D-^k+F7!fKyT?%T2e!tR{m*_RrYF#AYTmifTM zX^ifGeFM2^KV}z{{9TdiDCP}CdOSfontz)#3%YQBZ4^|{94f-O^sHc5kM!sP@#S3H z;Ky*%@Z~lv^vrI%HkC<$uJE4vRwp3k&B??ytXk3{7Iynr{DPTMOp^GUq+H&>-b}kq zr5$@xMb^qP78h}uVI?L?h(D*G0c^!XLHpc-Q2vo(nKf^n@SD}u?}+J6iR@6V@_+7M z3~>H)HNK}tG{g3N=uTMbKWru(+TW&XJWgybo3oEyE%bj2H?A-NtF5y$GE)QVjjS@8 z)z*Y9^pO-$WwmC^d#<|7^`r1=E2-LhxQwmNRT}Rm&M5PrX*Ip|)omJY$~Du0gefq& zzAmsFE1c@7xCx}98SHgE@3x2Ts)go9e7AOwnXyf1Vj2D$F5pv>sy6T+?ZBDSaQdEk z)eo|u;g!+PhOT~~5h8~2RXImAFGNbLKTRy8Z*6kO{12WcRQB-m;E-PHqd|W&&q#*3 zCDU+G1o4_!wX2)zanSZ-QL7_xQkr5^p0>ppE3I4xh_D55yPDLUX28xYn~6%wL^R^iD46a8^hSp?Yo7_?4#={CC~7a(gcF zv|%TzUK9IKCnD`eE|j?ouzLJ^&7G&8q_L*8eQO%+Nox_sQO>%3p0{Kdzus3y7&*;a zz@i%ppg%qBPM=={+G2LtLD3NQ((m{MwTRhnz@6QMsPfT(~BRBh3 zX!vY!8%K|3SdIo9+;i%g=P?=8-Sw z5=0rY*tQz#!tBx9>li|-9XNz+Zjf9RvWUAZvj1FY>t{C5|9PpSf$B_{jQIS5IKXdX zdlY6$YF|y&3^rzbj{T}{2PwEIKQ{g|Zpb4^P|Cj_ehEF}DQ;oddwCZ6Nn<9xxQ?0@ z$f;+)!gYNus)5a7*>^d9XB%OA5O-`m`63GN38#{r99re}6IPon^l_Y>3Tb`PXyNdHwJ#)WiS zM_i#eV1dw4a2zOPe5Ccka*S&<|MG7rk#QL|cdoitX(##J{{pXYm!tB9Tk}vOTy@%7IEZD%2lYH2%DOMglJuOfQ%K_$J?Z6U67bea z#x&k#3u5lg^8FaJpw)i7#vS0j$9T=0|JwCeG|6jdQE1|eM2o-(QLO4_ispJ*a*J-P zsm}bD*lPyvKE6tU&cbJs9x2e-^bpDe;zInfg>;63hL;#F$4))3UR%wA>pUw+1+NXxejEYXzh&vuu|*>#Lg zB8tby!1MP5jqN}5ktChYd5pjV>GP+~?NmB8{5TNK~0jG14gp zf4Je*3q+410A)o@iLde28a@^A*_kbPq!zUS{JoT6w5u6 z?axZrDbfQ$JBvQbXNRZfn~eeF~ZMi?ZP5jGhC!9&E1=q%GD*REmU4`*!lQMU{ugF0#ri-bue zlj=UBCe=!sVdkXnKauh+U)n9#T;3{U-{k#AB!3O7#JVQp9=f|Gm%WiP`Dot1M;)}8 z;k~Ihs_fcqj34~Jn*a5u?EJ(Km)w++P+_+hFKmRp-*XlL|k=ers(&E#*` z`S8V}(%Tif*MP}kFWF816awq6kL&o=hIc6jFxTj%jTdN#qZv$_K~0|^#XU2=&tC6sfP&6QXF%SYPYI~q4! zmRr1S_20Ie=$ux0sT)CH{08eJF2#H4Unm^uHY8ioXzfrOi+9nDs`n|>Y-(&qa_)Lo z>Za=6a`-rilIvxd?7V8+8>`_*v)`!-8%|RI^ke+R{12Qu1xqc&`0g&Gu-l{CoqzsJ zTF#>BY%V@XEc+0r zCLdhSrh305?8;FdXq|(3(&uxXWw^xHV#yvSn0itEwc)rU=IfO&ok7)CqK`K0BJFMg z^yF8T&c9*WKBlPBFD7tJauCv|)JstHu?$?NOhQrX*LuL&*i<6#eJIrLXyk`o@8UdZNp*Y1KW?Wzqx=I@9k~gIG^G4%aT1PwALY z*OaQybkp_r@0RrEhrqYZ67x;wa8|YLYuFB6gltcb$RDcBl%8=B2RENqpt!sMG`Me_ zY&Sg!CJ?AOrZD2UjEKOeNHSB7^mv^}^Okm?u!;~cnnIX2Ku2^jagmdDe$F7E3P+sm zwCAq{>;0zXQJik4hAOgiUbxCUw3uTty3hI!&&ba#C@Zntwd{L45+z+E(8Nz zla#OV@gh0QUd&`%;MS2+i*rEmUP)+(mknYAjAe&|NmD-il+Og%Zz)}~o+5GI@d=|o zV`UQ-FxGvZ#t;tJ(|A25WEHX7iZ`XQT}q@Kg~=w3r`0@T&$! zpy-ydiXoo#UuCXGX#I51yr1i$d-Aks8j|a)vVQMRb@p=Ag9h}~O(pkJ z2ZxQ~x!ibQe6${{CH-{}dDAcT%pO);hmDEGW?ER==IF zp_--SN4%e?;{$$=?guSJ?Uom)SN93i}3WCWgFfof4Y>%MlMrI z)fb^IZ2;ZSKW!p%pnV?k77*4X!Fr)`K-PjMh9uj zawC}XR|gc|rLxT=coN2(Ns}yB8{!Z9((GcWTKOb z1z!kwwXO#FfxCHJF<71CY(%tYmj*-fRk?nJQl(d0yn~0-5+w7YPY<<-aFU8@PxmEI z_yFmH#RG({fo0DE#WV_!X>Sn>iG5^3UCb9s7I z;m=}Zru%%U5nGB4B|o}+ZpIisth@pVhVEq3RLr|$awVqs#$+k>maen2kJT~8B}4o1 z2Ym+$7$ADF&<^9z2q5BV#kmy4slo!koWM=c|x4zCuH>v;}D{;NnFC+`@f) zjUa8uR=burpD`C5ePm_^%v~hqf2TwS9V6m|!?G^P1wd;UCSQF-6E2(xB$*TS$2xf< ztz3&qE3^g3#j$nFmlgk#U;kwoe=E}u7F^eUJ?AnBb}o!#VekCuh8?v7H}P@1DF*2C zpFy0v=;x`l zF*a)Y;hGA?d+2$33p5W9)FRNb|3@NyfNDpE>or;ip8)IKM*zSbXBy|66^KGM{j(7$t3TR4>^ z5b}_WIm5tW(@d@GgW{Ub&hfl#XD6n@{g>Qi`P_ZeQMs#gRc*~bHwaTi_y<6oap;Qr zB5`d#9mnI3slWB^Pbk}Mzl?}b3+izz1hWmoyDY6xsE}_A&~rK~gsTP7+$`HJlg$sJ zv!Q>}%0YTjUwYhWH19kObt*+AY~_l4l*3Y_x`uzOX{4#fd2EhFH>m|ZgVzlysi~dE zlx&RP_y;HBDS-m2YsFV6X9XPL-WmvrmGw$e_W8PHAFU!_snXi2wvdefq=!sKD;$yR z78x1~w)@lg(^~OgQxNV*p4Q`e)%AQ3TzMG-Rnk`HdbB4JHIJdQUMJfLx{ots9Lsd< z?An!o(H7Ogw2~$R+vTana_t3pfMiFf1yn>|RIGA=yPrt2f$fuLYV)zAaom`!z z>F3^Pw^gNtMc^@f2@DBfDnR%wV7edWkT; z#@DjT$K#jU?7xKk^jqJIj1>&hel(}99tJzhBlI^6>WA^u+ZkXc(8}b33UyUS+2dPa z_nHOWJ@?lIm7aXGJ-z-f5Q@SPYkF$Ki3WzB-qAJs>s7Qd2U}P_`jS!0tk_c%u_yGs z0{h|GqtI1?N+X>8N--&jr}-q5gPlX7cAZ={jtU$e4X$R!G!W6b%{naelXeXlHT@CLW|1oBV$M%3Uy1=(hzFk%_^)60zr@ZXP zg1fjLOz2iw#y-i(GFQO2JNzH}_hVmmlG8O1c%LGyIqb=Q1L>cfVk3h{%-2z1p?oPt z8K&>PPsB~zE-vBY)U9luNXJHXa-*yneiRS*1hD#(I#tOQ)K5h8UDpvQ@%p5;;r=Bf zUj^O~@iM^rBKy?~^j`a3t@1z!IJ;_S=y+%2fz?m3+X?$VR$cwZ^ugoln|8>BEZiPN z;A|cvO(NXuFBv(jw-HL((PXrOzL47kI?BA*HMoUS?(mE*tvdbK&<=INC~p??sCT{b z4`X@a`IC+^6g=k>P*7ff*QgTCF}yj2gIt=vpDnq$PS-Fb8<$xyms z!cm}n&N*mak%;gyhuBts=}4kz8eBTMn{AcS@6c_!jR(Iv?NXVmCYNz>jNdgTzQpIu z$@AyvVd8$EGKIEm7Lmu+dz7E8Wxu`KzCtgb5uc_y+R&-Uf1hAM$eE}u@J{}~YA4K; zkl`GbSf^KLHgaG}mEK($a7z`#Y`8I7{}cf1mC`;WeO8e0>+xoqAFqNkSD=$_0s?Vc zy@{h6t-)rb0;!!FRvJdv!C|yG&Yr|Sgd0>Y-s1(wEdGHFHYGXlNk3|%SBFC1`H0m~J9C>aOyzl(;O{tYW zwuzUZuL+)Nn{&7|>K5CI4qC%d!%4Tl&74HezAM;sM{e7zSSxyJpEG?p$>J)>CASc; z{$uh)_{^NG{E;KvAaE^`1a5sie3dcpBO-%L9j47(4x|=) z9^iFsDXD5Vn*=jO)%{=Icho3wYsA~a*10TWL2srZf2MMKU%h zpS8gr{-NqQ-RjH#t}8KO-A(XCy=5I^E~#!qew#%e=h5rsbREK<&)&pqD?a_}2+aJS zKRUg6!Mf$9kr&~N=J^U!(0I19fm2#n*0fS zS+aTm_L)pHUvw`$;G}v$Aof##!I&W^JF^H(mZ|yb+?*2iErp(JjNP6B%qeJyF5l41 z;wQ_n_u9#jAK9|rbt%=QC`eG#3CsSaQSP}yN+^wmJ$68l{WoS@GDrAl#8=edzXY2& z)+1HLlv+`ec!G&%-0@vo=~o5tir^TGl^Uh(7msQ08z7y>j;Y8g4qNrt_{t}%*@Kk&Yoh71V-?`hhQYIUQmwz z>m~!}?DhjI`^e>kFz~Z^l|@iT}skHu$OE`ycmiCpr)8HA1quqYMsxzI6Z3IJpyCs`}f0EhZt;`WZYI zk{Kw|^N?l+Sa^mhapStH1az$PAKHEYr8EDaNXrKI3&jH%ep8&>PIE=L!MKt*&rFMg zBlD{?_uCpr{q4u^6vZMUl^?_*4ZP*Q{?QMy+ef2)0sgoD=r_YGWUV7}#Evl70;sNw zBUoi%P__~Ugcg;K9Q9@dk?Klt)$Ad8EgI&|epY4RFH}}Dk%B@rIvZp1!(eP!8`z(* z1ZqnX z!hrXv*X!RUo-azHgk0ORaBjwh?eOa09>R=JyjAMx+jEVl7EAp*K6_5tI@oPP`Z2}i{`a5Bs8bvJ1;(XLHza?oU_&rt z01-xJ0uk;&7yTM+pqv87nhfLb1*3`x{Yd3+{b$N3v z#eNJ(dAfc)tkN7Kc$)6lYJblB5@!%y-SsD@?QD|i+c>=SB(*_FZ+KjBncwkb^FXnDq< zUm(c1JW=pFo|u2AN{_+k!NceK`_x!_XJ7}sWz*Zl>G2b#BeH8^e|Bu?rr70_FE1eH zKU9tRk*(dE7xO#hpBFespnm4k6=KNP|E;+Hc+v>6-&`U=dK8HeX}|7uu&&;kz51`U z1ahTo!!3HmmNe#l<*Y>64#d%SLxKgS^p#SdrDypT8-^dsjyU;t>6n#3p%Ce$lJPNHvOq6T$lBN`RzB$A7ae|*n^&hJYgqrhQrqx0v? z2X3Wg^5p*Xf)JBULiQga{CLjx)FCu3_{CbbPcUiwl;9T3!SbE@7oeBQyeu$IXuw0We7w4uxe z;$>mXdM`$0nYm90Q6*EM3EyUe48_EZp|C>yhw?w9^WXG7SRL+CkzsvA$TL>ODlR!B z!*}O+wBmlj@`8Z+Sz9;Lv`u-)er6+N+3c9WV)17>r4YM+c(=7ONGx#DArL;9J58!KI7DqOdtq_P)30pvW;*@9#c~C&+0teA{qt zJ8h5bf((yVZp@mxkWu`&sxaE{>|h6MZ>~Mo5g5sr@??`OZNUOs4f~uZAn|9kGe-mK zd}gt?{vQnoh1uxUr@#B`Q*JIq9@hAFI>)@Rwz9ewFy^D~SmLuzOUJDEzhF@()RI|a zG?P74WEL=@(mYl}4Og}ACA<-Ty{B_3&*q3E!Zid;TZ7)1I0%q(3=!TNtVgl&de9eD z4Vtp>U#BW3H4VcWC(9h!0^*xUTjJJEk*|D7=~{9n;$yxWFX7oW>(b0eeo(E%v$_GC%A=dYwj@jJ%t@wkIht%5Udx4|rb_5d z@ll;|<=7}?+Wyr~MUz(woim_WyPkzEES?jNn9W>e)kDTl8uM{<7A*~ks^MP{L826r&0FG+cH*(vL7)k5`TJ14f?qu}W*BtSbtRGx2 zGhf{=M2(lrx-mpOpp9Kr6}h{%Hv!ar#NFVN{_tzymw1kSKufLm=5RCF^FaU~ME}G@ zKxg9eq9+$mTmod6HDOqcdvgvyFoZHt1>*)jvm<-u4^q4c^2d%`xGL!2WNbDE`cEYJ z<`wO}%}<>QpFp&EnV&v)F>8ucVM8w5iQjS9VPnesO(DGq7KD}49-K%V&A;6YE7t3H z76*&QJ^FZf0p}kzV&e&St}kWrB`;?wngbOw(KmACii=5^2tA{xudU?=5BE}E>QR#L zy_-C}hJ^fED-6Awgi@N$@(QQ^jx|V2h;7s@2hG00F0Wcw<5YJZuWACO(}F3C2Ld^Z zDR)V}^f3qc+_Be(@3%WIqU>Fz3MKNuza!q{k%8M=VDj)({IO`tNd6%+Qb;fj0aapX zVR3QH$eE!chBFA>i$?Ku8wD9^Aendy#qHRzW_lMKt{WMqM;5}!37Pjq$6$V2Ukm>V zQy(SQESXZNhS8q#uHeH{RlZpCfKf=(-9y1-*5xF2ZU62W7%5v5s^7m2 zMNs3OxO$m>??CcJ@>0#r)M(w5#nW=~y8_t41oYx2`W}X?w^og(ZQtfw=XVc|<8ssM z3|fKhb_1e8Ql!kvBg@Wb!ci zB=;Cf@m%)Gpemw%U!`o$LFPlfb{F_dr6tZ!)}Y9ci0WZ}AJP1Il`*XDYxUvu#*B0;c^jdLrshsH z&jRkPNA+V*$p0B}%O2AKN&C$~{eyoSVn45)#jZ2hUm_=NGgzx`*MgM={ zBo-DflK+P^68_^=5OG2o+I9IV#U}2B#Ow+rDed(8A%Lc9VkvA!R*!&gGKN+e71Bhh zk<3;-uw^x6-Th%fi3{=qFC+#E#$EchC+3QC_tyiPlmZO(v6B~l#6+16<+cksPKz`i zIc9=XegU6Ia|mwM>uCj3l9X&I?5hO1B%g{wT*2(GX+~ZRcpNq}v+uLLK1Nta<%mn& z4-RqVUw`f5!U&bo%4;DW&fdKTqHlqy*rD5y5jYbF;%yP>3r^!2KWRNRYh% delta 31197 zcmV)HK)t`Ry9lbc2#_QKGc%V_aRDfQby!P}+cprs=U4bv8d$rcM7?q;(xPcoG>4)K z1V!2dSK``g^l(XO9sm2zgOucja#?eRLvlF$=9|$*yVEY({qiRAzdt^{`To5uc5!sf zvN+v6KJR1{AB#fmvJ`&fa`#y8K8ts~yY7=n80+?jB+JA&;BPL*v*E#ilX2sJUW#Vu z9czZ?#pIvI-{8DMtd3=r^On}UG-lou49F3wj)}@*bOh|n_*hEG_HuuaGK$0>c+l%p zvc-|VBdIL7LP9u1!XbFZd~rA|-AQ>Y)07UwbHTGioE>AhJfC>HTeACnP-P@O^!r1c z6yi@)9rp(nWg>k5O`3}jJn=k#Onr^RQZ)TOh5*>+G&TK+b>t}7>1t#Z-T{BjS5r;n zZ27J|4YnE29otK9+_yI#>F1Geh2EEGu{*%8QW<;C=fuepkT_PNfh8KlN4`hoGpO~+ z8gG7u6IprN$AuUs7RIw4+2DqE8Yat92S5P1HZ9xXX?^W@()*yq6B!wQ=eHQGuFWB! zV(s++>k~AR;?%t0Nv3Df4IOP3iCT{uL#|@+G>tra_3lZb`K%$#cps#A*{5((+vumJ zMWBwj`|Ssg!>4@#7-7RN{eU;WvQ!P-WeS0Q<*mE9(R6TGB}+;PjD)6rS*P18O04pB z?mVD)^%7kRN8n!?{PWI#?Tc8zL*-J5zi^C?LEBnf1EJ{PB1XH(5@=9^Ik+U;+F8B` zJw7;!iP=y@vqESBEDTI@l2OM@#-HQlZQ1=y5<3-6#as=W?Yuvx z-eBEUl&(#SNLG>P2SkvP0lchbIKcq5g@-ii5z+y@ei{~4dUA%h*?M0xJ!^NjB_STf z=F|_?)ShKNA;PyD=5%Z^K5@1gLF1eUK^8mc0j+zFgX-@<2uf5HYjU!TWNbNI(bZ3S^r# zr>WJJg&BK(lG6NTA3Oz(8}(CbcpP}etwlyCt|gYok-s-8jPzxbGH2fbC@F9)K;D|? z0iV%4U>Q_CpI{8{dDh-7ugQdVkmnShGz*jZ10R1}w^2VE+&uy+_)PSR?E6>tPIrE^ z3LRSZ0gIcAefFu6u2Y5l?*=#D+$+IX$PWlZXDGO-cuzEJgI(jJV1a}=IQarc(AQ1P z${scuMjns)%gV&}2m?!i*25-cB7_4%m9MNgSXVX7(SmXmCg@?5S}ZwCd&y{o8LnPP zFGYVUJ_6MY3o;Z7)COyy1MI6Ciz~K;4VFHJL~2`zhdbQ|kBPg< zxjBWn5EOyjK*S`LVyaA7uu0)fcj49w;o0*q!v>KYDjWxURpD^bT|8|!fM;*TWe*`j z2qnh!I9@ViJ3N}%(V7;Pa9~?cp>1daP#}K@OdbC9p!5ukS^UhyC4r9fQ7caU)$fOk zJ?Fstsmy1LO12venn0+i(3&bmWCRC?q(Tj2h0Q8>H5$wbM2ivC+^(sm^&Pf5k^T

(Ol_kGpa9km>Z1RZ#;u=%OQW= zb4$IH`ner9zL%$fU7O0h#lvKP@zT;ml&}>+-VIFwy%{Jx!qnWJ(HL_E;S6chAe~ZA zZ~Y`@5T--Dq3eUPV)d^8E6J>z#*g4te@2N zzmprcBesr_^mHrprN+tpE=)3?Ya9$$F#@T^ZX1?`tCNBENVmi5Eyr6G4?d_ott>!BIp9Odsx(4l{H*8PdE z_pt3K@_||tGl~lnfY1=B!;SDmoe4xy#A%Y4r8V8$-lG;_q7aP1hsJe&9rYtPS)0*R zqYLV0ll$j85EDfFWs~Fq9F7G~6wzF^2ux(Q!U-B;*}hxzjXoYHsN(uX+n->e;@q%f1OxcuiQ2c zexF}4Z|echxbodDP%OIJWH$v`q|Fq4Nc*7JI%6ZgOe{N-(_dfSSp0sg{HVycXJYL@Snelr z1oL+LZu8Sd-DsF?gTOD6BqR-^&=2Bbo2dZCoKOF@x943HcxL>Xbq$j1JxS;ndG1tw z)9?0C6nax_a1wd1o8vqhCT0&f#+OCYoAI~{yeIxFhu+TJX=(cfS1cOQzDx|x zz1fUI)U_rrfKV-&7EjuaZKcuSWAY zj0U5m2rlxKL1`egKf6yECgyzYfMyTz8zB}*a zE^z5wqbG5Oj>GZ#F7TVK9w&TVB(JxV(bbxoHa#1Grmcap>V@-V}Qch%BcU-IFN@M%{7}W`S1?6xTg3Nkk~`aWtQ1 zK!GRba$A@6f4}`jR%B;0StbzTJTdVaZdMc`|EG}ZrkWvQN)A-WkHanxF@$juF@}N$ zKb5;kSdb*ZdrJ$^3>YrCft%}LZYy4pJie1Ei)04SfA!x6lQ3g2!SN<+;agYWJv|xz zqU?c@T!a(MY>d#L@LrRpBo2V(3wS|yrxp`5%rozCkb}rNoc#FhZXbB>_~QzzbFbd7 zsZIcn(h7kOOvSRrbF3uhE6ywDDxB9EZFO!(iX6rH8jAaJkCQSa_r09Q;Q}Df1#2FT zyg|0De_3*aQIKva=L@ixh<7WQf-I6Luta(9Q=ND9nGPrtC~6@bq@)@rcTDg1F<>W^ zU6AeQry|}EgYsa_t^Oa$9p=PTsv$stI3{P5##$rh zP(0uwM>}rgYSqqqNit{Cwe!x+RI>Le{vW_k+PiuQ%CzxM9sweFGGUZgHT@a z{K}*ha~q3`^d3O?;q4n{6D?gw028R&aHJK#Vs|wjrOo;oJd8J8_;HwBwL_F;UxPvl zU>uVK<0Xx+oP}T9rAX{p0yQL47{YiHhVYGb8Ap!h0eym>SJfFmDOl45orPEfrmTeJ zf4Rmw4pzpH1X3lOXJhYl={>?Lb(qZ^%Zs(PJgFKxjm?J{izvZU7_V0X5b>Kz6dt<< zT1&@*X$2tr5e|oj380^5fuIYJjOO^{uSH7`9xw^}F)TDVzYidkbg=}#W{OwjwKZ2a zq~7QQ%B)mLRxL`?JA0{{Ad^$qOQslIf5c&z9dKD()@ac*BTuCu(wt%5MJ7x}bTS2} zzve_2d6dI12K*q{AOVnhm(;orfb2E^(whK4G>1EQTJYjNuAd*UqVMInIJiaOI8YQV z#y|2R<^qnLTz-SHk$^_FhIgC=sgVxI2Kkt(=pr|G;3j+_gv|P|5F+9bRiGg?fAJ1a zNP=~Db27#7AR+ajh3inct8l|Li!~5Y-?<87nWY;ES>CFxhWnN}a&Ftt5Y4cGA_d167&YW{j|c0SrfY^YJ9t0NQC( zc<(UBJHyl;&_NWPxCh?@&{K5se{@>#qloY7F;cXgFt|rz!GMS(B=(CH#-(}`fP+mp zjP#rfU9$_+X9W#tw2K+#`3N-Knddu((dS5d!|eJv8eyaBTe+#FT!r9Bg8sa4O^||S z>(uI=%O(-1nJd(VgAgmfAQllzJytxMr>1TGAOX~1K?Vb6vk35b%?}?Xf1X+I&csn+ zmXO0ck%yF&Urw-ck~p@_ht}{C2B94~BmE6>>3JGD*bU8nDTjS)!Tj0FO{@ZFh$4tkY+qFd9WB1Scp>!@qF%SJ z3UP&@gFT@FXP0vSAt-uxom_ugF-f<_VT9!Dn01&ZSD&Eu=eX# zWF=bfa6AmDeGB5=e_+kVdbT{vA=SR{V7{1gmgb5nVifYf5lYQH8Be9_bmP#sB2n`I z_%m^@6Cs?gH7NC{(cJ$FCek3j>PS+Lqr(05h?B~+TG}Z5o z^ya)jAjS-B#V`w?S$GF5H0BN$?tS{|u=x`j_zfD9K@by@!x<9;F*z`oP;mh%f0bBS zkK;BHe)q2s0tAq;ZH;bAc?ov0Gua$KCP@Rk!OT3cCAQ6N9S$i?xAW^$7u~j7JwRTp zVpWkW7OTE0Iu7C>3G5C&Nc{TUDh_b)-RfVrt2ghmI!NNEj_V}2y$3Rlswxk*c^buO z7TmVMhwx_o>GlWEtj;yFDvRQ(e};ufCr6#t+uPM2t3-A>t(8?#8rMP7uReT=gBFB8 z1aXwrRq#lvJuFs5MVvtNRp^7ij-{4(kRL6U>hg!t=JCzbXafF zZK$p3^{@tNa_dc&rlH%9IDDjG7EXILv0$7nv&OsirpUwp>t-^aQCBnne|{c-GY+h7 zWPv;0XPUJ2E3T7J4Spb#-~RhOx(3d%N%APqOSbdnUJ-fF-;Mv$th%UzO`1nZCbO$c z<>136ujBA%+%yTvoNd^Q{h`z3geshrJu;i!GnzBsl;Je;w;2a#T5Z1}ll{&~9i@2b zxbr_)@qK#3%l%X&->f%ze^%g$t!An6#I(0svtm1*HWSa6VF%lSaU&T0;8?9Q4=2x; z;>{9n(~dSe`EkLX+}o3tVvp*aoNA`C&TdxSf?2TnB%6D?&?NKte+ro{jx*)K@JmC*OFx5|lB53NBPb}oWbe}&ity2z$qc)t&~p>Isa!L*XcMzYQ7d&tY7 z(GtdH@EM%lFm*Ri68;ClPGx2CICMu@ZeX4FX%TUL{YTmMe;hrwu$97e`I42tsa{g+ zezX^cm*Msb)E1GizJ|%`LLpfA&%r(7O(K?&x&K{PiHDWFBLAKNjn60Do)Ez4F94?Z z;t44Z%GaIOaJ-1BBD*{cl+z^~e}@TT#~xO$E7<~wRy(sBn5LkI?^3zDL8Jg+F)iyH zkxlsqgJIw+)Nm`=MlKFICte1U5SE0>((+TY3-a!B4<@r9_8OFNX6- zLI!8iEfbIO4+yI_i{%tU7mIEl9w3nQnfrmoMP2G6GQes14*Vcfs8vGZq;L_CSYU(K)I66hGdku$tT3-NBvIs=qe}nOPv4tb{JvNY#eF!cnK>=ow86H@N zdk|#$uVR55VQ*G%ZAwJ*RpE`9y~41@Dpcu%I=uPxLOR4rRNC3GeS)ABhm#yXi5OjW2%j) ze`E$|P@r}rq%ep24){9kACcHg!$Je?oh^Atn%Iev`59x^f|0&=k86hYxMEm`@pu-F zdoJ)SxONavQ{A*5=iQ!l)C?nPCSc$uxO7pIHvsN8i0a*Pc1>0_(;KP2b|i1Mb8~8k z{w{*#af-b_N{|T++@t9bjmsF3grOa1f6hh_U&Bkq(S;)KnTk3^x5uPvmp84l~h6ohxS0dT?4vn9`PtUfUHL*^g-t%QxM#Uxgy?K{Z`1Z}9QOExo zpgKlrSp=I5TDVyLXGq}%lz_i}nGwjqX|Px+kHpgVg(Yn@QY9&|KmKg=<2p+)3g|DD z%Re;WC}|m`RpEOrLek1NmmFs_e}1krqS5L}!R6y6R@Dh}07sxYG;6v86!0}y+f(a; z%-)A!$1+AA5Ey`IMSoAg)=SQYV(F-7%F_&tbP{iAj!y~SP5S2|gv1uxeayYtBI58o z#WSEw+>0U1r^BaMR7YvSDN;~GAP6F-JGe>V?e${8dwaY33&b+(9|~n|lNTZ;0XdgZ zaRDfQomfkg+_(+C_pj)qrfOP|sCTMTuJvOa`;y1ZW)EH;IBHHeV?ElUv^~zR4_;KO zr!=YTo)x=X5U9EDX>#oLD`7^<)=(`}KeFjC3(VZZ$n z+}?bB_zaTeQZgx%Fe)ohL^2u5Y(6|}zHefGphesGqAMvw6;<21-F*2PZTG;JDZ-0&Ayq=m7HYv&?csq@-Tbc%+_|w>LH=x<@Y&_B+3EU4i znpTgtW=1Op7_j%iq^Ex10UT{O{NQm7M#llz6bBOw+7llGS-RV=S$HDFW&m}q?rL;K zmQ~;kI$2%@C+$Ue?&;?nRR*SMWH#7;aWtKX$QQg0Mwmf2s5Oto#d#5orp99*J6mhf zfTkX(Wdj~_OOTmkc9?`|OFwN57e6;KU=DH&pbap&{9XS#! zI41Z{LoC8c>?ro51V?uTl2+E7k=qa7c!rlv0=jWMlUmJ@w%B9E;)@yu*L2k4rxGI+ zDQ!Q<$hIf2jK>&b_$9+}08yWsW{;T_=fPk3bfiX{evAICFbl5sDf~XOrl-DXdR%#? z$VXD$x0hR4o*|Pcp~ELR-cfCThR*Sf5(Wn&IF__%{tFqq%0GpmUwj&P8mA@SoPgv`4Z+O_@CGl}xLRX!)o zU>_$;bI0VTp>KIK7zjEH(8PP6;{zAOYw6U?4C!IAaDneBX8gq$~g;?WwL7z z%c-P5vN$CW=kO(mrcg7=YlliWajvOEc2P;ifWb6Wjie)@*H!sI>bt-H>@ZNGJorfF z;}R=Os3<>}qD4Ptt-y(Y$UHI2W5Q}4+VRL&^s;~f2`!Kdz%n}=q~87$t^)N`VXEY& ze(Kzr1MSa%aZw@|m&f4Mo~`N~ff&HEZVo-*__1XPMoq^Pz1OGF3u5>nk?R!@_#R0u z9}0`l>bWk!hsc0#OJhpuD`S9!Uc&FOZ!f3EHFYyVaU5hNB|8p(EY}S)Kwfl4kQwNY z9`<3WUU&5#FGfWO?s#h|4mva5TqO+>7q+7`T{)0lvw*jOY2#6MEXDB!q=s{hY7#Oo zTX?`logE7sz1&Go9Ijda;Nf6#d6?lfR*sIxQUpL~Q6*GPu#Nm4j2vE#HXo+$I_WO} z;qy#^S&{T(Y84@W-Wx0o9a)S;P$Md+4A23}Fv7CEAYBHHmvj0c!YAz{4IC~}sA(jK<+1JTX#HZ>HAE(jQ9m|@8Q`vglJ(hQbKS${9dqcGFH3Xk2cD_H!P#sM&3=ut zh|_(o{dm!T%-}8U_ZTgz$Rt8=H21IABiBaQy%!V0vis(_j{_Bh88LNOC5Fl%E3_X# zJ1_MZXXaT;0=#?xpmS7}KxxDlh{v$M6EgS==*uVrqMK%Xsua}-q<;PhiSPdQ_R}ph zJ`&;Q`}^JfhfnW62=#t~Yk?pZ7p>T|Y|nRCOQ6PoKTGXTRVtvb66}HgJsmVgHBJwM z-dpi2Q}C+ZeMC?3{OU1xrOJ#oJ_8_&?`>;WT?Hrs)xA^!9i$|DRTcCsfNF*949~8% zK*HpV|b~HJfp2?fLTltQgLlX;(P`1%Pa?qU40%=`w0j5qQ3qPs7_X)`pzDv zfs-?TDT*NG55Zf(&s`eSjUlJNB*$k{qrnf@_h<0f{e`3Qve4%j5L}}@wBQuMLL@tu z#7Y4s%=^EF-2(VIwVyK*FVz!Rv~m`h{3Vi3xgR{BJCTdu{2{Q8)8H433%suqQt&Vs z8qRQB`jiF|z6s3v@(jXs^Dhz@xt1A+_`psD5kaRC-_8j%-qZOjBdP_*3sMW-aIRvVl9 z9^c9lb?6iY(7b9t=z9JxuNkHiMN#4*8eC_&?szss0nRAUBKhgU%C5~LCWi)a9~p;# zYdtxUx2wN=tn>|>B3_@iGP?wwZ``Q+lC}X_%w|cx#Hah8%Ix@inzM8@)w5hYc#EZ7@|L}8^|Q2VKKbuD zNGW1^5S*gRK;|@zQ8( zcV%ZGy%&WJYMfeJ`p7>ghGj3BQ}~l-+?&;deR$aX7wxgTBa@XXC;~P$mr-#6D1V(; zOOM+&5Wf3Y2sz2XT9v4GFF^|=MFX@&8v{LT_Mo-6786-=L^+$yukQ?JsE3z}pt)G& za5x;!)rQdNnM*^THr53-8geo&$2Y=b2kdcUH zd^zFf`|S>X?Y91b8G0b+el3j#iLZM!7TjnTB|(y8q_OH?RYd;SNE*n`mugf!lP9%d z+*COEvY-s|+*Su!lc&IOOZ((ekb(2qQ-2UoBD_2(!Dfaf3SyChK0#h4tPZiQO`Qiq zcuS@Zj8xU616gp>)=<39yhkkIKQIrs9nHBKXbp|*h z48lBJ+k7$`NYDI7H8w8sImB_CUx{DYFAvhZAoFp~d~_)Rc>wva;3A)A_NuvGwJlMs z^EFLPC}GGEW9FZ$vk59W1Sa$5jG-?m6UIT7=H5=gg(3YYP1oGsEq~-j=)d_K2oRf> zzB-*-bx>pb1oI-a2?9Q#`PfvW9YGrKCSJri$9X?T=g?N?v+9nta8D|axJ2R{FbNcK4Of7i6GM%pmQO7~rL-^$l~ z|7%<^*;odNaN$RqB!B+iSiXq(%zQpm6aA&j*+8bP=gUaufa!z_ETI~hPd|0L6-2NN z4k)%6`c8e5^$lW8;Xjfe1en(#SdIvWvq?HnC5AbU#-6@L$qG$`GfXX7Mr{T7q#wa0 zAYIecZUY2NI;ERQTqD)s0<@3)JzgwNtZCo~Nra^#U`hspL4PN|jxJ069NCV@Hp)J>7oUwGMQEW#OOL5^FUg*gP= zaDaPD8h8CjgMV6DJxyYQD>_A+Eq6)owxcmn+%4f1rw+(~t8=6c#+1^~*)KiX4m z+PnyTj33Z=6e*Xg+=+bhGb4r|piqprd^79=WCakgLM4QWoTToWp2BFCB$Oift))p4 zNi!2bNn8RMg8HN^-CP$3y6xv!)FTn61_UR~>+LSjvwwL@Z-p#V%i%nIs9FxEj9m^n z|3IA$ka+#OsF=G@F%}%yd5ZNR@>>($%t7?A(w}HV*=oJKSc1aasHyjYg`=Vj;SHzi zt2%LyWeD=CJX}mQaNKxQCu5p&jXFq}jFBXTH6;Mw0JTs|~cS9OipNJ8M&F1K(Fz_DDh`#dG zj++{)*jNnp>WJ()a`3u3u@2A>VF~CjEJ3$aVLmMwe;{zM#OHEVz#jq%%aK@{qyNFyt?P9^kx(4y2&%)^Cw=YA$&HpdCqoH({U8IV{Tgfw`< zP;LUbtwu_CV187l!vrcDsH2>1WqoA!(e4}ML7q7-V5{jtdEKGdYzl!%)t_j@(N|J0?kHeVIAGzkf3aRX-ka|%R($n1bQBEh18hvOL z;N(PtDKTl}(___mFQ2?;7#bcdyoXbRYkz4pug}3!gcvMtq_SXOOCzKqyix3nX@U|5 zGJF@E1So+IApF|r4mPz}=ASIRw;;$SFnct8FWR)AY!v(+B{sA6k|GVCKM;5W=*LGP z0KD`XhXh3cgr*djlQ1J6A{*}x0(Vey?gb)lgjQ^( z4i-}#gr!$2lkvvMVopR{xV9wMzM>%+urHsJ@K zr`!$6|9G-3qRjr)E{Lh>CMEO93F14ub(*+G5vkWvpPZQGyO9N zI?@BRHPsFCJr&p#w31kjTXJ|T2%0MvVU$W>XGQblplZWu{{dJ>)U_$!>^J`bSZl6+ zle01>0Wz0QaRDlSrC3>$+_n*ZpI^b!6Sr!IAqWyA?FW~lD6Y6HmlI7&sgm`;9ddVQ z@fh*QdjEa;0$i{(kVE$xO1U$Wh8mqbML*b>OKm zbz)z^KVdnBOL21y66Idnx_u~`qP*fZ^1jr~w`+_`o|UY@g|Y#Su_Zr$SF&5Mfr zk2gW=W-@hu?2DXRZBy`gf6l71d0^^&b{^Uel>ibSRp^DGVTtmhX$PFFVet&4=~+{B5l17%}me;$3-D9^nXCR;KP*vE7SrbTUT4KtJp8I99VVKl{ zolABD-+$xTWmx!Px6f)jqpgp0)}t|A6sYxyuj4>}deMI3UTN~)>j$x3x zA1*~#WIY~41uT>wyO#na;fupt!zE_%a6P&;xHM2?!=&1tui`#xVCNp>7;2*rpH3Vv zPORL#{KZ!q#A{IGfr)tCJVm)#lk?nZ(FqLLVKBU~yuU<*LV)cE%OEo;h*B3#4;al5 zmB!wG41tXpe(@uJ4^S%5UZkQ8qCy{Fl!C=Dpj4W|zFwS)rK`P_LNC@pn7O{aIVuzH zE~x@IgI=u}e9fBQ;Lq?cw!d$gX!i>i`)Tb#f4SXHd#00kzDbtbfz}k_P#O1h5mgtz z73Wdh0VsMXYZ;g@*D&T)8hT1?K)*y`kS6YbJzY?mg2_|=p(}H4z5Oy;uLC{2zW~Ky z?z{l#g$QoM+G747CWOXBgf%jNkaZnF?TNeXJ^ux_OOQpA68!zk!Nz|Dj_5oT*zDpa z@ls=)1>5i8CR{HXe9fZ&2XI0QoTCB<^iG(y69IE?2eL{7hM0e#L%(MrxL0E); zV7EeB^0F!liM--?&6xt)*4F!ceRAhJ=Bt9(#n6dCtThMqRDk|f5>#^J&a4yX_e z>e*8Qq202BH}4*vEO*n7HJuhl?gX%ZGyTWzn3ArAe4MC7Dg)*4l`*2lb*c1iqJ``(D}tWVU<8aJnEc5a>#a_w7D& z#*+1@X P0Cqd@O&M4gz_4HAW%R@f@Cw~7JT=|($b@d8GH8O{Wq8;fx0-}bL6pt zFK@3zH?Jj2!$w~ALs#C714vFeE6@NDJt+n|SCE|ZR_K%<3tXv9PXZ0j=z6W@#29&c z{#R`#>)mBl)|_dB+m~}Kl`b(W2~%0i5XL8@6xkVZap9gQE_Hw9KI<=kUFN7jWLj;* z9d;P&vi358i^(l1OgM@CfAbs0XTPD7?V*5xhw8amQ~LbGomnR6^i}zAAun0U@N+4V zI0`>u9se7vjci#_&CwowQzjKknoLfW(guPBF!X1nUW$CIr~;ZYgxK-`!5UM>2|T3_ zPoVdja`rij;Q(63Dz&A59E8kz7ZEnBZOV^jSoOWF z#Hfx-R#w;xdnZFpi{YvUxg>ISW5X`E=ZW<0l(xTG+}!L;g0R(oSfM z4s6jY0}OV>m$B^FHJhnQM9&`zY`s0{-j|T{+;S!QXdP@6ZeA0DNP~r5S$$8F`l3*L zIy4w#*m~yOag@9}FGT7D8oeN`)|g*RZn+}nr<^^xe+3NzQbJBwDTmc`+0y9Su{%qf z!YBTz+?82>Q=CU&5N>&TV^_VFMZQ^-VUgWcyOC00$F8Xr4=2xLW)4GJpY(@ z?m9fuXUW|7?Mx~po6Cxyf_*Da0ihKnJkmDLa@pV|M>$2AmogVDK2P?HlEF%;`AWEo zSK_|`N!ghT(O96&eE3xN>JKSK9JbQCiurKi>}cOR=!c>gSxSrh(pL6O6v-cfkFdidNA?b z>wht|oh`ECA=*J6NTDW=wArKqHh0wanF(9srBf-ll7%OqIByy7k6axhwQTmZ8HB!P z67_Xc>THU~7c`kft2Y;t0yB``Sl|p}jR;?tB5r#X4FpEhWB*D8bLhl`UQ!4p8kWb^ zjbdS#p9q!}#o{gYc>pAGUtj_-6xEdZuL1n_bodXMLTuENGCL;&GC4VyQE>q%f6Z7+ zbKEu(zUx=;vb9u2qCo(BK&c$OyNS!G-Bc1Q$-&8i9zrvW`A|rXCI9>V8Vx>XM8)2! z9QFjD(QNeNLwAEB*asqbb@7keix=N06T~7i!oPVS-(A@%A?nl-T?>5Ro!@@ZeY* z2+N~HE6nTQ&BYHF!fVKa`n{L(7f@+enwe=_}+d9)0& z$Uq5VNuxZCgX=iTjApK`zPyfQSTggO1!*Lq_=>w4PLz^iQ(VU3W7Uva$A&lUqum|H zs%uGRa4jVyYC2U%TBj^%tQts=MupuSk4_{_iu?Pz+7(WKBfP@^qwl)ja}2wpuG;$I<{@pX*|e(-u+N z@U9e*1PoV<(BGR)3J?UkEO@uJh;vxjcmro zdN#3Qqim!mM12)+2@+8*=t!z$mT+@(^W}Hb?Tpd{uEH5{?r-_; z8z(52Ge4>2r<7Eme^OGPDES}HrGf-pS;}b+VbByciTJGDdKxYj#i8xYR;!-Ml{&I=u7DVJM3^fe2-SG1#`Be^oxD+&)5k7M&eXj( zI*pz~40~9@!(zfqvWb_MN0k14AZs#eR*(xC+ma0EPd|@9)+d5}RlR9V6*s*fBB6RA$ zFPg^pYdEuWl1p0;On~Yyx{13kM)Y|N!9~agF3L+uH8v zJBD3-e;mdsj=s6Z?4RHN^u??1e~OstM|9YO?_2*@+$kf%Ti1#{b+uf%8kQ3wV^7mg z*VZ3-HeXL#cn>o(nbh(0zUYg(w)K1xx%Ya@Lyfu>_PNZ)3H;z7M;>iUJk z9Vige*kSM(H&9|2ART_wT7|EH09#)1mhUIef0cRS5_9g7d~%6dJ0f2@!Z5->ChH)*2JoztfZ8G-VlZH&+~3dS$s4T ze8)`x8Lth}ec)*|k(>lfJUDs5hYusKmP&-h z{2X(uVmuQ=Wuc!wBmftrrn7EKIZDWy4!32s8+pO9G^#>t`~62I?jYgV+!+;ydaPU~CfMg9e0eu=VbjUyqCTMqf84(7fs=-pjQK>aFc0c+VIO$gJz=qS7Xag6 zYrC6h?lLSg22Y<}#FZro>X6bFBtM^p;?eaO#e~Rfb z$r3pDpd$!O!9Co#o&Tb5e*Wf-?|n}0-Dy=v+nLjAQ0bK4*3!5a0NSv~Xx|hobpEjr z;SWc|&tF|U>s<~=r=}#O=BnyTp87vRipt$qdn?=vSms{g*A&2v{+PJ`Aef!>XEIkV za-h>63pYV+a;>Y(Cl&Ly9yzU1?Qv+(Q=BCU!c;$Vb&F4?S*q5zNs z|8R5IqhV-pA9>m>Kd<_OrT;(fpB2@sW&S`u>-zUz%IdF~qEJ-+c|ziP9+B;+q}1HN zBw0d;*-#Ys-@V1S_-n)0e}X1+(6G1kv%KfL?l$IzNt?xcw6l1Rvr%%=N(}r(f8>|d z)-`E8a(SRD=AnN(q5LePyTl2fty~YL%A?(WxWq%TsE-8~VUW&_=xtFihq*hLELZ9k zXn#+H5L=gK)zT+%Mz7PF5R)wyx&?y{lkZ->;mOk3(aPVTmXX{Pe+W2i&JDxNvDnAy zo>bFoX?WIsSTeU5N0G1o=%R{>GEZ*|AupJk2FUB`DB1 ziCg_Q?l}dxO}tdL3*NS0-(LI|b4Qr`ldnW50yj06QE>q%f1OxcucS5-e(zs_BCVDq zrCGqHfxN7eD2Wm&QZ6UT!|XixfV-Ow4Lum=e%hypEPQ!Ni}1#C6{Y`(p7uJ?;kQj^+V=KT zBygjBn&;tow#cPJ)1LaF8P6B~sBt`2{pBeMuV(O)f5)Cj)A4uqL*vG#J2CZQ$1^hI zVI844;0pUhM_M0P%V#=HA5bm`ZCsCHbRG^Cj)60iaDquYIwm`i2EtzXqw#D;o;4%y z_tSW#hbtr|P~m zU5%4Ge_U!|y2|qHlRV0@axHQ^lWmIZZ!mp;0i7%`S=pr&qRoUwZE=$77qUXyz{GPYmbUJ}#Qg^?pZ_igf8}!6` z{IN_SGvnotecSe^FjlXFB|EKX|*ne{UR^&3!>x;-fo&lT@4s;D$)Y63R)m z=6SQ%3q|)zOV}!1gl2+(8yqi4RTw8HyBE}Us_^&BAGT&VVMO4rYyK2I3X)WX^{ux7$qd6L8*W7eR}hn2RO#5yGzrJ4YbSQP%u`LBb1eMe>)!W+>J0}Ezm=cTKxO(fI_BXuJ0<>fKL)`vZh#o zk*gohxcW)7z}nM0f`ME?W;P52Ks<7s!u71M5j2oMIpNb-_jo(P_ELZYL7qnScnDDF z5Ew5$?c0_s9MHJ*BNMa-m*&rTgEfN5hDMSyT)qLRr2!a@2!Gr;;D(-|fBZBwHB;OT zA_|)^iV*-s?xj%3I<=A6c9gEC8!zwr^B1)h+jq6wvKZYPn>~<1w}E+m>&&I8aG~Y$ zKy{cLLgdSEBQPT%@VtM@pkHh!vMQ{P)D&RvXy(!ktFM3hEn)^JEomsnu45i46W($| zX}~I_N*~oQ6Qghg>Gc8ne=;7RX<*{WAl0{y*T%Or^kvON1ZEt+1F!rui{qG=4w^N0 z7qd8HOHT%wLrfM01Cvp4y``^)KwUJUV4y-};g8VSEx{ThT4UW8ccCyg1J3+)F(Q1; z-7?=mRSIyr`A?i203*>dNiOb!4Z{P6B<@@RMhOgw{(80@&o~2se_>rNi~AWT+rx9^ zdTuHPW@HBV+HcxR3M||`gTD1klvPk^Qh;r~8seO+)(?MRf|rul1vl0B#Ydeq63Jn} z#19e?#23Irs%je0N!4Z^6RKsUPx->9Vj-!BBwwMBZ<&Xt)DSE*6dT_(^@!$5ezRwj-t z6mw}qu`AeL*vUpve;RpCy-{!fuiii&3hICK21&kZXUc1e^bRlA`qkE(xcPOXnEt6 zp1Z`t@t7~3J3y^k18sE8JD;76kPQ>u#rPA}hQL;G8$dg8htDG?ONi2*yKK1N0A*R8 zc_3L@J&1b%4UN_Dv%l3mC~(~O*sdu5d{+;D62RJ7d3xjG5M}sE%I3zOs1GfYIy{Ek z>Hzz}RuhTbf5IIwED-a3S| z@=%48p+0ogT0ZMLj~EnQG#4t4AY5W&MkH8kulh`-dZ|h-vWYg5%|bv!o1AKI8Wie` zv3_yec^*IET=0ifN{a9DE|uv}2p^>%j4V}qw9_ZbubjU!SI4)xl-Jp)W-XR_VO_+) zdCzR({Ngps6Um|{&tMtK;6fVBfl3po1SVyBZ;5 zBY5?SMky&V)z38?I!i&l2syg`J1fg1|0epgMY32bw|d07?H4Vrj-Z0np932qNbC}P znskZl9j;nZ<-_aQ1#rFbBOpxDlDgr6nR4Odk=8}ENzJ~Mm}r&Wr`|;7G%}4$R4ii) zRldxLqq-@{V`+97Dz~tLsUOPW`H1+l;_9YD0-ekRif|yb9|R)NlasPTzo1$;8dsLb z*7HfTo_TntRNmd;kTY@{BgZwz-oF!*4)0obM2nyFT0z83!=dS zgqAW`lhH@#oLK`oB+McQD&8H|ov;E!V+Auzf)=Z=z$UJucemTS!O?I;=rhCoZPn%{)&UHYg83ns8KEbhuM> zE3j%w?JDx>O5u`@EPNtaOSxgqE$xLE0!jh(Xq;Ga4Y_;n*4EZCu{0*Bsl+kjruzz? zQI*pTw#j5u2yU}8#Z7Z6 z=xs93MG>g3u@BSmBAF7etxrT5Kxh15Hsw|}Wzbt*6c#o(MmaNk3s*}PVs0ibuH=SQ z3`k}sW={6x#3m}BhnX6NH5P+ZgjCsMge&qEU<*Ki#MFnv%ESgE747N~PP#=b#VI92 zwk72;E810(^-v3W<-#tn0qc@M{Tm9(%^0%L2?k7`#>QIa z9E=+TA79rNA3x{x?_X^cN}bzYLRW1&rcG3^V7TvoFIXUSa1NoZ3UL{9B?|(al4A=b zNIQs-zA$kP( zEitbo2R3k1B_$DZ7tW~>%rbzW-w~^xdslM<1T%r)HUD^Z)(TWx17);VH;W0LQWFIq z!INoKWU5bTG@RI?oEsa-lZ&W;x(-6A8Cvv*l=e&95cJcJAA}h1Hs2K}`6B}n_>BaW znc07NDXioHvxsI0+5iczW|p#=vzhZ7gt+{TkrX(#jQWg%>HX{QIpWkshA1}-oAlqe-14W6WhcMY4uvpp<+EQUCg zGH4ec3~WIIvvfYKhd0;qzA!QNcM;Ok-?N_#MSfPyVco%Q9UUESU2ehpkU;9QGciBv zM1g-duljp)`wWi51-kgIjGjTF!!`Q!QQ&U}u{z4g#vlnKIYpqsK#x&a zDFdii(ZIxiI6s1MZ11rASHZZ35Dok8>O+J;M(bt-dt(Tnr|_Yg-G0))eP%Cld7R;* zDyM$fzT$OM(2k*QO^?9AesgMfBx>@zxXqK(i2Ja0rLyMd{%9k0`?4Wu$^xmGx$G2zk$(v9%DbU;JuBX zz_DGRO_Ebf^VhWDEAr8galj>U%Qeu2kYi)*Oc*dx4_zbllU@n_ExVb-r1S9e1RUoL zB>YPX+WhzHgoY%J1%by?X97FMKf$ zD5MY4xgdCJM7EY&y}Ju+;@@QHOs2hz2N3}2d>~37>bt&&_N5wm;{2@31ZHp#?jbX- zI2`dITnfIvu);!h22RgfP#BqufG($k%BLTpnbuZchz?LeOQ(0_REv6zSSskvn@~(L zXPsF!Mdh2C^8~VtEYs5@A*JKMt5W9KsH-hqZ=O`EvFN0!l`1E{oJ8yx^D~ZBP&Qnk zHoBpaQ$q^&b)qa%i8yKJ10ynZ+^yz~d;ip3X2iLbkiAk3mZ7W`&u?mfGs{Ot=n8#| zMun*cTh%S|%$%{lPhoaS{VXvJCWxhCSc~r+(>`Lau;i5FeQ$ehDnzv?89-w7K8a=# zQ|em%(RZf49yMEk;Qrm3Nu=wpddnwob~o%X-5F)RDOd)3)K+m{4765jwjwAJZ}N5@ z`>VJ^b<~F@0MKzVQffA+@) zP{I!C7IST_*1@JSMohgxC9%I?3m!ZFY)tV`VNWG4v+x=>>59AeC3ml3OQHu&q2@qK zQJpIhq)I+LzM3M+1j@RGj(9n%@NDU4QOrhc@jA6`Wo#lh?82_R zl5xjIA)U|WUJMKMSsuFTq>r6wjPoUARWO~z*&}qMcnG1W0=~2IzIhr58{{IHNFGuO z0u^3S+*^zvI+dT;GUHrV@7STuKNRh|K?&f;(m8h-||5y+TW-~BZa|O zlBkQF7|kNjyDNOPHKp{#b2e7ftklu(WbpbN7C2Mg38vfGNO{@hV)-j` z1d*2x64H?R0X36cjn0WQewOxXbuUp&jCc+kvAthBdpadk&I`%)vNrw63&dOpwK!xA zFM6~jxYNn25toDt(h~$LE9q3H9#zJRrGx6E1BtP81^L8QQXDY+^j=R(ho;1~H1jil znGhRHzfXKC8G=niC)1+y7xPss*_V%tUIy3usmKt#0Hs=Q3!O}SP;4V-9k^;z4y=^_ zX`As%cByh$KaAx3Njy_-`{P7#S{ha2M>*=j|2Rj|89@7Bwu+?GWFuUVM}QD zOntrBsC5WD{U>$Ui|FOxx!blmiSPAYh8u4DlcmCn*5YVM$x6mIz+?9Cx=?cKws5Av z+hh@XfJhM^(Xf@Hj3>XFsXaPZ#~)o|=}UY=p{4J&>VO^Je5P^r{BaeR@|aG>oWqqw z3+TyIZl`dvl3#CDooaxcVGw>DNp_Op0CImdZUCKJuek-TV13f zT8pw8L;E;if6v*U=@27v)6GLL3kPR8DF6O8aS7=^s7hLF#iZgw&HW@5=zFk~20Ao@ zQ0Kn;v~%?%?+2+RlkZuREP6kOWqMb_0`>n2e=C}PrQHbVf(Oe;?P1pqy0H)qZ&o5! z$(Z}s;Twa?6kL>QFFtg}!K~s_tJXaSi=5~@2n6@4iqSu*8-NSBBmqXn2wnVD8_ZHH zbJcJ@qO{_Doa?$yn?C8K7v9?#timF@kj)=Ch#&mqPzzAtOa&*ckFX}(vRMtKfxljs zy#qgD$_&hIYyOMqFUi{<;yoTwCchK>2v>Lk*vIJiqy#6bSV1W19~zNApfWu->fN4S zQ0YLSQf;XH;Ku4TJE~-yPll75)95cB=IFc4lMFvpB?SHKvVC@l z6ts<*yx`}tO&5s7`ffECWvBPGBrSE{GM!oCJNI1hpKoOSnf>Q#K810qa_XZ-NPfaU zJMXX_LN^)1Ze(9^`gwB{+PQKU)s6S2ik5*{c5_`QsNX|mL4f1DC-YlN;MWzOwZ&#b ztmU$2QOdv_+gI7B(PRkH01m?KdS5b5bC_P`YDd&sHDW6vHM zXXo7J?g$Lk(o?6-y&~kbFgWP0j@>-+5E2<98d5vcbsj-J6L~3Zr~I0V&8Z=B9?rgF z)TNBeICZciRPg;BJ^G~<(El2d_%=^5LQkVfn|2Jl>`d&@Eo%4jLf2D#|aC-4|;9}Np+@<#`9h|@6jIc zCPp);zvY?7q~9ouX8aYqMf!)uByc0d0%V-I+=)HFX_bA>I}Q^Tco5pD>Vn%eyn^zB zGet)07fgPCDXvZNesVSZbSFJ7m_DEXkA>W@MuG%wR7tD|Cr-{)z2g@fJ6^uFsN-jx z0y!c5J>Ku3j9Vw>t{Clzfq?{ETj$j21!qB!4gwb)XU zTY`}O3?7Mwi=aCXPs2zYD59B_2BRK_ON+J?)OTV*<+Y3iyp#s-In3E+GdO$9SI*Y# zao*45N*L3-FXU2*W#YRz9bB*VUm(@9n60BBb$3T+^BG7!uU-0B5yJqSka8l=MaVLC=u_H zM`k+)e!ycPEZ7g{QSOmvz+=^v9=4+qQz7UUr~#*mNgOBTW_}UF)mNhk@TY7eoqon8n{MeoAo7nU5;b9GJXd6(>lg{+Z z4V=p>+^-v@AN+REk__&L1U4Psw#-7LWeEil)VizeTkvA2r&xG^F=BPdPY}RLirN z7jC--F4|wR`IDxSCJH7Ery~1bw?h_shaCCYVC@b#nJy_vq|B{}L3RRKo?{S_Q>l#* zD^sIgVhJU-T?EpocJ$Vx8sk4He&)bwk4FZKXT&HInDd0P3$xR2V{KUzy?PDc$rfH-BHz(|Zub35@svnE~DO916-7bW}acgeS*ZOtf(Fgd$pN z{%Vf4`HdwiK6_Efv7GF_$-@ioT=>0y==F-S=XTlF;R#jK>nOZOtrS=ab7kX{_aPZ~ z;t~|lO7J|Io|yM4k?DehJ={%Tdf-dRl@xk;+)H+x&1dTO8w_nZ&f*yC*{&3bTZFK{ zL!1#XmVS1mdL9>O;itS`&4W&goUA-^WrI!T2$?55kUtuNr0nCG-@3=yS}Dtmmsn(Y z{xa@E=P+BrL!S*rpIy~8U7FF~M)FS1q;^OBk`IW1@3vLLH+h`p?}fc&hzE^z{3q^)H zhPF-nE9g$FVj8&YPJ%&IC9%>ixmruEHnAGchT-09T_Slumh8e+FMKBdReEnDGln_y z9)bWeEo$Q-v+mbz>C45_x*q?@+9S^m{2#C_&##$|QWb{luw0`Fjo6Fdc7X^$R2Z*N z_&y+%rC}bhg-~qB_%%!CPNPo$~$1XVI%MQCj{k-1R(985dA7!4JuR`lH>cqa5b`X8E$tonkGlp&IO#{3Z(?fTLo!Rm%z=w5_Hk@0KrZ+yyK@`*CHBiXI1x!$k##AkZ2Sv7mx$d|7ws3_ea@~9wCaPd?*@xQ$no}{dTnzBLsRtUG#k0 zqf}Coiazu3)u$;JVyXiW|6twA>rUj3s}aV7lyAN1We7qG7Iz4BMfJvHMum9FUh7PG zbp_WsZ78|cHfk=YYmp?TnYSo`^{Tp-Abo*$X;N*D;tH;z4b>4?p6#7omS|Dlxnr8~ zZi^-Nvqs)4D(rlMhRAai=8mt|a(oY|kK6KFw)xgKkC;Au((=cKY@Me^$E%egs@a?@ z@C?Jxx-K9}bq1cV%DK6Ct|ytYdam@5#_sSWkJCV>74`;-(nIN@RFb^ab+o1?IK#T; zv(l`S9!kb;Dr*{u>CTnIwNq;E*NSq+_6-6#+XerRA=72~rVzr@-!FZ_M~bVz-yZ?2HNO-r64wG~-3 z2HUN_()oT&p!~y%i@D_-YMI@wQ}TnDJd?>?+|9lu$0;af!g27d7vXe zKKnG8Q`@ph^e=2k+%$r;^WMyuebv7%q>OK9aZb?nF|&QLTwYll15;7^`~`vM(xys; zBK9|d-?lQ4tOm?1VTiLCz*i%2FyG{{eA!A@Bf){*ax)-nu%iD~&1hcSwE4Tc<#k_u zW{+?Clo|hX?^H)FkJAc*VejKOBF5WWS;&N_r=kqgpedN<0Y;!y-qlonSZnvMy~k@c z*iK6*~q7QUC4pkesA) zA)F&@R~o~&r)F02I8i~2MY10EDqDDdCTOy}xJ4XyM-Vq(tODH`*ec~dx=8=*%PDch z;XMzWkvMON;lDd)vK2^>w_dr?F^K!E*-f<{-lI1Vf(p0d)fhP^YLkMZ`qIwoudt-b zhj2WgUYDnQ>7v=+G4xEP;yh|$x9i+-y^O`%&&QPD69<%Zn$E13oG)s>n_(Q^^`nqENg(QX@gdw4qO}zXs zaE_^H9f{dI=5ICrB-YC{;hl>tl@&%hdJj+kTLq+FLnX9-T?W@TPFo2nww=en{fw*X zY~h!%I%ey=qT=%Y@Get@bP*Ix(D%az8<4j7Grd` zbv|&QZhJaB8|ua*A2c@svOx^w{8i@Br2D4x=`OJF?xcHNR+W-#UV|U)f{zbn`?jkD z5Mr1fp3j+NX2C6R8D#sb~7zt$*wLIF}8EM!Q4iXXKZtmJq!F zjs%Jw0O|MXpNKfoJ9N&|5^%t=&o&8GBl%`=W%?pfNmj1)bTTD|uArSYm3FL)T#?Dv zu3+uXEZI>%i>+f|$=EOrN>8i4NU7ZO(8VGlscFve12@a^O~d0_ta228tA7g+lA#6b zFW}hHW1&j_wr62noYn5Xci9gDHrKTd5^4=z#@u~ZvUt4IU{GEqMZ>e{j*$&4dw z_1om_>7?ti5y8ZhPi+R@cP;aUG7Uj8x7d(UPMFf+vf!zFbo?Ml>F!!g9r1zZ0fWmh zMcget*LcRjU@XVx8OX@BT2CFIbS@>9RZ9Jrw+?7w`5~l2N^|4r9B5ZamKp10qJ>?o zx+`(fkmm@ODObR}sVDlpM~dMwS*IFo zJn!u}(E(Nw)|j+Wfq!C;q|;_x0yoAk0@X*NguQ4 z^1+^}@N$?QARJ$}4x2@xcG^9owKwq!u|wK&#TY+xD4|9WQw$p?(&ln~H15P>f)Dp} zbAz=}uqEQKXZNMJ{)Gb6Le^L0b9r)`4hKLA$fpKNjT)NSlww_S;GIZOIsd_Yt~R>< zxATD@&+nDBC9XTk-SiLjKgl{HaGQ5{mbzkWh8MfL&UQ?i`PCJNyJo+U9GKz=`(LPl zG;9no_mN=$yEU=9zdZjJdI_&Ke)3*v*=jO`B(;DO<0M7J4>1a4v5`@Kc;VlHKFHFs z8uH8`OOgo&t@rs)EiFdte=3mWpEngp$+~u^lY^Z}bjtVC|6vR@%t+ z8>MFJHt0L=%aNha+9pYx0IG{c_{sR4Irs=PXQ&7Z1U5jC<;1l3*P$lfMaba|5w5@Q zuf6e&PaHLCG6_lq_h-9Pw66d_8?Iq}h+ll3ll;c9>pwf27{!b!EfmQBi$^|w%bU|0 zQ8-h@0e(lvIkBF`r11joWWFqa(GI7{D4fru3L5nms#b3!;uFlf_NT_KY6%Ib_=GRunco z!~kZdpd zsqKUAr1)Am(d=#By-qw<9KW~bf#NK8kcxgbyC5&#kd4K#+vUqDK1wXTI)^%yclWlYL9(@ODz7^}-Ni z@oNcj%sy=o2l3Y*W8<9P(u`xv*El*P(*J-sI$xzhP3a3P-%GJdN8(O*ZT>p-jqxC# zI;3}Z?ubU$CAt+oTxn=IDxdtIrHzwHRo*C?#46ymvmdb!4JGZ>rB|;}sR{W+ZTGoi z`a^q@g|sT3t>pO(8S7ik2_w4b^$f~UH=(7!8m)>o)tDr(m-^FAdperQ0+wxn$DP2_ z*>ySZh6_i$=j4V8c~EB>PkjyLbg5Ie9#6pAu}SbjjWRD1wdD`k~B{zj{46+FnP(d_mFwAydwBIwrw zH%0y=+~}m69q%I08Qx$N^E^q9!owC{(6-f9&Wa{G{#}v2Vvj{nP9{V zX>=v2FKKRlQ$b3@-yijSf!;l+xzFx~S^q+AuJot5KVpJAE?o+5Ms+#qsJ7baOM0qP zb^k%n748XcQU`-)NHcpM*bR#7pK#xS_7*=}pfjVs+2!+}IFZ?=yD~ynIjCDVNJ|m` zPF?&~;JE_BJLEv0{CS(|)_-8tHve|fHTP+Pq52d~?OE+W!72hGc0&w)qSimLkjy&D zt(s1da&MW;rY@Ln^Y!Uynj6Jhxe)0Hno>TlQ9PAd*7=(yk6x65xMeOktbZP}Cl;Kd zT2O*)jxg;7Tivy3C;gjg{ZesW@h$_Ti|isx=fNo=KM(4b?;v86hSq^$1cNBfmDdTf zOWIE6h&?p4D$uINdz|c;b2o|&EG!=TkF@rDQzMq(6#*Ta zU;0}ZGJ=GP%Ec-e(<#TD{a)d`;ndSv9;SiKAaA8{O1c8l9c#@`hfHCQITF*}8OUT} zUJcL-^QCuy>3QhvUz9-Do#HC?MbaNB-S?Wd$x)&ekN+|fXTq`Fxiji0O#}?5CvS?% zsNqz^)a)cT`F;h%akBmIfziwLb6XKw{U;Lzyzk7ASR`9oyl1-w!F^_aGQsLLa^Oy1 zos;Rx_V<+K7#I$`A9B`6EfmgkuH$gW=^PwvQFzYmHC!dypoqZaS9%jHWkDEgS}ift z>hxaOkreS+_eET*GrQo;0~qjD_I&xo+ed~g*$9XhT3#eXvMrZ)Bo65{Tr74>QE<-O z=`jJFabvDgC?ai1+uUG-i5Xu4>~Y~YQb^WSZ_Rz!(ShRnvA!Y9d6mbWe(-jcBIQ~R z?R`IG1OGrKsC*#4aI^S-PLzB4XY!bOFyj?*8QHu~ zx?OAnAJ(pM{F#GO$x6m+eLOH`r8B`=g*sy;dbv^K?H|pdNHidN@$8r&wH-b=2|*j132+p<>M*I( zO}V8K2A=?*FLR`#;xsjshmlGXJd#rhR_^`KGpXJRyDFqkHT0Js&B*=0;G!{bHVE&{ z$o?&0kH?I7t%RP3L2T$vN;%&%k{Um6eE_PV0x)O?Fy&lFccYQ*ofS~^01aVxJIxN75@Qh9kD5gSM$Oj;pK zu)`4SGMdxS6n&?PZ+{Yn9A;qik5IW0L)FD}wbm5FCRV5)P`4&{vnrNF#U5XoR7n-1 zh>s7HLt5o9ghk4lwNF)=s?4HipT?dw2dn$nMp^HIfN-A@^>NS=vWV>q5#J&>Mt=B* z7u*V|SwG4{t#zDKCdEsCP9EKqlWPvL22FACFCo8tH1ba#u>k;wOz&q|`o-C8_+EU> zHm}}$Kj}0C>#HT~3Z2JXnFbg)`$_-Z8k7ipV8v8@+RK=dvbCm9LTuOK@GyC{@9@Ok zc?aV>tI0g+v`Fvd#J;JcXf6eVjCg9wdqb}MHM>DRXsZ?i!++UR|B=SQEMPFD<#=th zcjfmT>IGOpU{tI2v5nu)PPiCx3~|&^waoJKyk%9ipEhO9;)$<+>ApuaH_@xe{(A`9W}EsSf6 zt&eb-US_st904>@D?ULDNhUU+9qwvXVB zX$QxAn`FWlw!B5bC||RFxe@^%Kq@LBSza#sz z(bm&Z0bwJ{Gshz%@Mj*_s#6z~f;{HLol33AA@SvM0w(>~;t_@uKgzaN|GsQzT-!*F zEWA}ykt!!blRo*A72n8~DosKwI6w46(bQAR-$UwXr?}iIoz*#Uw~5%uI}{`dk#m!@ zJ3&ZA$ZQ)-&QIaU1($$rrS2}HabXnF4){_8?YCiRny=Qk>KbDB_&$GO4fGrZtq)Q74E47#&SI*|> z#ndOQD?WSHK^=qmjKgO#gkSc_#ejXV#C@YGSY8>z3=mJSB|ISZc$~3gKNAgfp+L;$ z4ELOHf?#=1%m_;m9?5Ci3r{PdHMQw@o;lf5!92w&udzZ9lZC`|(!$Hjle8za5jqQ* zvX?yRM=1YLy&E!-LG}=HSrk`F(-}2nN6wJf#@e*xXpCp62l-5?Xk=*$;DlWUV>f9y zhoGrP-S$S<4?X|Eg_C;IQ zSVs@Ej(TV`Q)9zHZ>f#+qG0>0t{?|h$nCB=%_l@=??Se$^Ne`ODwdmHrpo9d?|vlT zr1$>Rnb_g$Z3&W5kq_ZtIvKO0PwE?$+1vmhL=NQE=dEx*LSJ~5PMt7J{_;F$Umy90 zcNzB;Onhc+r1e8_2mS}-1WbQuEr?MtBmzmtweet7R=wxQusDiO3O?lRy4mbk;d8ZL zEej`qh|K+*Praf%=y(Ha^ig6`6O_C4med*fGBes5R2^~e@W<)&%?#RR%gw8P%8QTU zcLBn%`n*r*O*lSD;EWs`rfC%LoIo)-0VL8=3msPbqMk)|RuT#2ceBSC@l z1EznPS->FH4Q;z;mm!S$fx|eC!yWM&zpKM7A({}ixe%#^pYTszlm;N{{|dKJ#jq9x z3t2EDfAOh3&5{Ag_{6XYLO4BrPGOfR3@<|v*Ppt=VG8b0=?WTWR^b4|^XPi@#E`o} zqjS*Fn$z`$v-L<~<8nlp$z zQY}lnW~n$iBe8pd#~djVP>gM9aia#>VDTVnBR?DQWe`GUj$LflZRTe}fy?m&M$Xg( zMyL~Y#p{?li=`)98am&%MoI&Xg}W=$-|~CE^SH_tB!jQ+EbcU5Spsjeu?r%v%RXzN z8PILR)kwdOu#S77fqDCTq9iZ7zLNjVs7O3#P($2=?Dl2GL5{D?^n{Ofe+l8M6_^?7 z3|q>rwBA}`-wCe8&?}u|N_@I8xT`ldeL=Ps&}@a=_hFx*%^J-&dGLbqD?1)>W#>G% zu>9^mf5&vYnqg);%%IRs(31Z|#n!0llgjR^fA)8DT)#9H0p>Sd#qa8Hk1u}--ZnEg zH~hOB=kHkB=O>_O-Pibd*n02osM~Z@_$lhx@p}2Q?6|WZ)T*&(Uq=*AS!n%-Tc`I6 zD&CHNXtdQ+0{y*QQ{|i&)#{%LBZMsdOA+$d?}as~nw7K@_Nznx{WlR&-AVuBnTtH{ z%0!Lz&281cNx(R%>SaQ%l=h^7Vy*37_UOpBAcf{zb<;3(@H%@t@k4 zWwVX4B=#p|A?KZ}P0m7o1M~FOA}Wl{<#``S zpBl&K1Xn#TnV-;E5LK4XvZi+I)1x83j@Z2B#5_s^H1;UERTbilK4c-F%u!rZ(%fsoaVTvj2+SpD(ycWtS!0eVKSf9B6Y1vn zQd;6m!O%FsxUM_M zHctk03IS!`l_E^p=fi9XsV@qDha(J0O+Yj)y=0N$YD-E=p}y?OVEVI>)BcXi_)S+U zoz~=<#D|7-?s1rjG8V9wm~QMmi?;)R3O)OWi2qew48z<|`BalmyGQh0?T+_Yp74Rh zNHM~cHQTaz>sX3eR~wo(iQWjBxc>2KPy?{Stw_&vKpUYHM0N_cHVRgvis7ek7N^$e zBQg@e5ZR51`4Q>apIqP?$2@2prZCi+o{0%Y2L#lPN@^K*LacJBf~d@51celgKMqS& zx8lZBM+W_lx{*o5ex zH+>18H_ow)0l(gO%_qru;|NcqZvvmlNveaYiNOy`MEB@8WD$oNck4z!+|}f-pjo{s z0K}tK#y_$>wkYM6f9gQ0*$yfsXu3ilwaK~o#phpa_tz@-MyQtFE|VmMWWx(Xd{irc z#>0<~8=2GxI#ADkiiKLYNc*(ZJ?H`?K^GOc;#ctD)rmb~MsKItSEN*^TE* z6X^7iy|REKrHSawWV{FrB;hhksSdNequp-W1l`W*e|nh~?VS-560!&W&I3VSSsm;Y@|G|B!m$2OzD zR7}lvIiJ^F4!$Ner9Yw?o1jromKrP%`UMbFKk8g^<-*yT8c9|u`Qd3Bm^kMf!IKszlY&{tOI2E&)Q1*c97Oiqpyb)SM;A{3Hmsv$n0(Nt|F|m(C`W zhbtZb)LJeU868Xw&sa{dJzr{~5uh^sRL)E$W_ z9F+mF^qwVlcSS!`4{T=LcP{_K=JOU1@M%FK5``hl!G7OjT0Y+t@XywQVgY^OvzP2_ zeeQ(_$4!X4I&jBmy;S8T4ph1If-*P)Uo9P8-u7(5u&>3+@dBWT++M($YM{BS$zZL~ z$te{T{CRoI$+?G+t*-fq`~kx|bvLH#LFe znYD-T`%vF;i>Fjy`kgWKw@&=Am-{bMrq7}*Jq?SdQbD5!#3M&G9N@NouAWBz1;Vn} z%o+CM!WoQ_nw*|qK2j-GAD>eTFBKxW{TZnFf%{YstvyVxUsiOH3*UsCw$E1{; zW?vw!;Dq9R(dRt=P!+$L`+|e*j%cZSNEqSbApc9~8Db0|hgY(X z_4D7@QF1v+$qNOy@?@ufuH_HjN^GCMl&r4UC;C^Q&^aeY7MnwB15G>t8tJzVDZoEOb~jh{y&&HQPbH$3 zm89=6?fLWl-|~$#73Z+?s>xaUE;6_+wG~ZnX9qf}l9YH;);j)L+&4}|UAhcV9%?CH_Je zGddL|r-(dUJV+TsX?BQbHO@NS^4dhvA*E~-`&kaxYoLx3qcguBS}Ur9E~J7RGQ8ZX zrb0adkN+s7J09vP{|_w}A>KfqHXQXEz(C`#1hAj4#l3qxt3+o+)C|TpXwMp&X{r*} zkJGige;-^TA==i{yKlbKChNV{<9DJ{z`oaFB6oNNS6#q|Pmd%3BZTMV;ADXV0K^p~ G;Qj}?kY~RD diff --git a/mpi-energy2-extension/review/review.tex b/mpi-energy2-extension/review/review.tex index e7525c1..4e5eef6 100644 --- a/mpi-energy2-extension/review/review.tex +++ b/mpi-energy2-extension/review/review.tex @@ -54,7 +54,7 @@ \title{Answers to the questions of the reviewers} \maketitle -\section{Questions of the first reviewer} +\section{Questions and remarks of the first reviewer} \begin{enumerate} @@ -112,16 +112,17 @@ instance). You say that on scale this would produce less energy savings, but your arguments for providing a solution for this was based that today's supercomputers are achieving massive scale. -\textbf{Answer:} In the Figure 7, the energy consumption of the benchmarks solving the class D and running on many scenarios are presented. The number of used nodes varies between 16 and 32 in the scenarios while the size of the problem is not modified. Therefore, the computations to communications times ratio is lower when 32 nodes are used instead of 16. When this ratio is small, it means there are not enough computations when compared to the communications times and the impact of scaling down the frequency of the CPU on its energy consumption is reduced. To solve this problem, the problem +\textbf{Answer:} In Figure 7, the energy consumption of the benchmarks solving the class D and running on many scenarios are presented. The number of used nodes varies between 16 and 32 in the scenarios while the size of the problem is not modified. Therefore, the computations to communications times ratio is lower when 32 nodes are used instead of 16. When this ratio is small, it means there are not enough computations when compared to the communications times and the impact of scaling down the frequency of the CPU on its energy consumption is reduced. To solve this problem, the problem should be solved on a number of nodes adequate to its size. For example, for the NAS benchmarks, the class E should have been solved on 32 nodes to have a good computations to communications times ratio. \item In Sec 6.3, why did you choose to keep 32 processes for the evaluation with multi-core clusters? How did you configure MPI for the results -\textcolor{blue}{Answer: We keep choosing 32 nodes in both scenarios -to compare them while one core per node scenario has distributed communications (one network link for each node) and multi-core scenario uses shared network link communications and thus comparing their impact on the results. -We configure MPI on one core per node scenario by choosing one core per nodes (e.g in machine file we did: node1, node2 ,node3, node4). While in multi-core scenario we choose one machine with four cores (e.g. node1 slots=4).} +\textbf{Answer:} In section 6.3, we wanted to evaluate how much energy can be saved when applying the proposed algorithm to message passing applications with iterations running over a grid composed of multi-core nodes. Therefore, the same experiments as in section 6.2 were conducted on the new multi-core platform. Instead of running one process per node as in the previous section, 3 or 4 processes were executed on each multi-core node. The total number of processes, 32 processes, was not modified in order to fairly compare the single core and the multi-core versions. + +Only the architecture file was modified between the single and the multi-core architectures. For the single core architecture, the architecture file contains the name of 32 different nodes. For the multi-core architecture, the architecture file contains less nodes and for every node 3 or 4 slots (cores) are used. The total number of slots is equal to 32. + \item shown in Fig 8a? Some MPI implementations have an option to use shared @@ -130,8 +131,8 @@ the explanation of the network card utilization, but this shared-memory optimization is possible (sometimes automatically detected by MPI if you pin processes to cores). -\textcolor{blue}{Answer: We didn't manually pin processes to cores and since the communication times -increased. We guess that the shared memory wasn't used.} +\textbf{Answer:} We did not manually pin processes to cores. Since the communication times +increased, we think that the shared memory was not used when two processes, running on the same node, exchange data. \item In P33, Sec 6.5, you mention that the proposed algorithm outperforms EDP because the former considers both metrics (time, energy) and the @@ -139,8 +140,7 @@ same time. EDP does also, but using a single metric which you have defined: energy x execution time. I think this is only a matter of phrasing. -\textcolor{blue}{Answer: we use the delay in execution time not the execution time. Then, the equation that we used is EDP= energy x (Tnew-Told). The experiments shows that our objective function -is better than the EDP objective.} + \textbf{Answer:} We agree with the reviewer, EDP also uses two metrics in the objective function: energy and delay. The sentence in the paper was modified to clarify this misunderstanding. The main difference between our algorithm and the EDP method is the used objective function. For EDP, the product of energy and delay must be minimized, while for our algorithm, the difference between the normalized performance and the normalized energy should be maximized. This new formulation of the objective function allows our algorithm to select the set of frequencies that gives the best tradeoff between the energy consumption and the performance. The objective function of EDP does not give the same frequencies as our algorithm and thus it is outperformed by our method. The results of the experiments confirm that the objective function used by our algorithm is more efficient than the one used by EDP. \item Other complementary points to consider: @@ -159,7 +159,7 @@ is better than the EDP objective.} + Same for Fig 7. -\item \textcolor{blue}{ Answer: We have considered these points in the paper.} + \textbf{Answer:} Answer: We have taken in consideration all these remarks and the paper was modified accordingly. \item From the design of experiments, did you consider using replications? There is no variability metric in your results. Have you run multiple @@ -167,24 +167,24 @@ times and got the average (execution time and energy consumption)? I feel that such variability needs to be accounted for, otherwise it is very hard to affirm anything about measurements. - \textcolor{blue}{Answer: Each experiment has been executed many times and the results presented in the - figures are the average values of many executions.} + \textbf{Answer:} Each experiment has been executed many times and the results presented in the + figures are the average values of many executions. Since we have deployed the same operating system on the booked machines and we were the only users executing processes on them during the experiments, no significant variability in the execution time of the applications was noticed. \item In summary, I think this is a very interesting work but the experimental evaluation lacks variability measurements, consider larger experiments (1K nodes for instance) to see how everything scales, and there is no overhead measurements although authors stress that in abstract/introduction. -\textcolor{blue}{Answer: We will expand the experimental over a large number of nodes in the future work while increasing the problem size and considering the variability issues. We have discussed the algorithm overhead and its complexity in section 6.5.} -\end{enumerate} +\textbf{Answer:} For the time being, we do not have the resources nor the time to evaluate the proposed algorithm over large platforms composed of more than 1K nodes. However, as said in the perspectives of the paper, the evaluation of the scalability of the algorithm will be in a conducted in a future work as soon as we have access to larger resources. We have discussed the overhead of the algorithm and its complexity in section 6.5 and given in the answer to question 2 some solutions to improve its scalability and reduce its overhead. +For the variability issue, please refer to the answer to question 1. -\section{Questions of the second reviewer} -This paper presents detailed performance and energy model for iterative message passing applications. Further a method is proposed to select the frequencies of heterogeneous cpus online. The selection method itself is not difficult. But I like the systematic modeling for energy consumption and performance. This paper is well written in general. The technical contents are presented in a logical way overall. The experiments are conducted in real platform, which shows the practicality of this work and also makes the work have more impact on the field. However, I have the following comments and concerns for this paper. The authors should clarify them in the revised version. + +\end{enumerate} --move the contributions from related work to introduction +\section{Questions and remarks of the second reviewer} \begin{enumerate} - +\item Move the contributions from related work to introduction \item why emphasize it is a grid platform? the presentation of related work follows the logic of heterogeneous CPUs. Grid is only a type of platform with heterogeneous CPUs @@ -290,14 +290,7 @@ and with or without message passing. Indeed, the proposed method can be applied The proposed method can be applied to a heterogeneous platform composed from GPUs and CPUs, since modern GPUs like CPUs allow the use of DVFS operation.} \end{enumerate} -\section{Questions of the third reviewer} -In this paper, a new online frequency selecting algorithm for grids, composed of heterogeneous clusters, is presented. It selects the frequencies and tries to give the best trade-off between -energy saving and performance degradation, for each node computing the message passing iterative application. The algorithm has a small overhead and works without training or profiling. It uses a new energy model for message -passing iterative applications running on a grid. The proposed algorithm is evaluated on a real grid, the Grid'5000 platform, while running the NAS parallel benchmarks. The experiments on 16 nodes, distributed on three clusters, show -that it reduces on average the energy consumption by 30\% while the performance is on average only degraded by 3.2\%. Finally, the algorithm is compared to an existing method. The comparison results show that it outperforms the -latter in terms of energy consumption reduction and performance. - -this paper is quite interesting and solid. But before acceptance, I suggest to have the following major revisions: +\section{Questions and remarks of the third reviewer} \begin{enumerate} \item suggest the authors to use much larger size of nodes, instead of on 16 nodes, distributed on three clusters, to see the scalability of the energy saving -- 2.39.5