------------------------------------------------------------------------------------- Comparaison performance de differentes machines en compilation / execution (calcul) ----------------------- Mesures effectuees en Janvier 2007 , R. Ansari / C. Magneville ------------------------------------------------------------------------------------- (a) eros3 : Bipro-bicoeur Xeon@2.4 GHz Linux (xeon-lx-2.4GHz) , gcc 3.2 (b) ccali : Bipro-bicoeur Xeon@2.8 GHz Linux (xeon-lx-2.8GHz) , icc 8.0 ou 9.0 Flags de compilation avec [-O -g] (bb) grid49 : Bipro-bicoeur nouveau Xeon 5140 @ 2.33 GHz (bb4) grid50 : Bipro-quadri-coeur nouveau Xeon E5345 @ 2.33 GHz --> 8 coeurs (c) sgsda: AMD Bipro AMD opteron 248 @ 2.2 GHz (amd-lx- Flags de compilation avec [-O -g] (cc) grid-saclay: AMD opteron 275 Bipro-bicoeur @ 2.2 GHz (amd275-lx) Flags de compilation avec [-O -g] (d) asc: bipro alpha (@ ~1 GHz) server DS20 OSF (osf1) , cxx 6.5 (osf-asc) (nouveau asc 420 MFLOPS, moins puissante que l'ancien asc 800 MFLOPS) (e) xp1000-dapnia: alpha xp1000 @ ~ 600 MHz ? OSF1 , cxx ? (osf-xp1000) (e') cool: alpha xp1000 @ ~ 667 MHz ? OSF1 5.1 , cxx 6.3 (f) superosf-dapnia: multi-proc alphaServer ES80 6 procs EV7 @ 1 GHz (super-osf) (g) ccsvx01: XServe G5 bipro @~1.8-2GHz (Darwin/OSX) (G5-osx-2GHz) , gcc 3.3 (h) PowerBook-Reza : Apple G4 @ 1.25 GHz (G4-osx-1.25GHz) , gcc 3.3 (i) MacBook-Reza: Apple/ Core double-coeur Intel @ 1.83 GHz (core-osx-1.83GHz) gcc 4 (j) MacPro-Grosdidier : Apple / Xeon 2 double-coeur @ 3 GHz gcc 4.0.1 , compil SOPHYA -O2 -g (p) IBM-AIX regatta , xlC , IBM eServer pSeries 655 , 8 proc power4 @ 1.1 GHz (q) IBM-AIX meso , AIX 5.3, xlC V8 , IBM Power5 , 8 proc bi-coeur P575 @ 1.9 GHz (s) SGI-IRIX64 magique, CC NOTES : - Sur les machines Xeon, il y a une interaction entre process / threads par rapport a l'occupation des CPU's. On perd un facteur 3 en performance multi-threads/multi-taches. La machine MacPro avec OSX se debrouille quand meme mieux. - Effet du systeme ou carte mere ??? Flag de compilation - Flag de compilation par defaut [-O -g] en general - Sur eros3 (xeon-linux gcc 3.3) [-O -g] OU [-O3 -g] - Sur Darwin [-g] ou [-O2 -g] (ou [-tune G5] sur XServe G5) Sur les mac (en particulier G4/G5), grande difference entre -g et -Ox -g mais peu de difference entre -O -O2 -O3 - Sur machine aix-meso [-O -g] ou [-O3 -g] X/ Performances brutes cpupower et donnees SPEC ((http//www.spec.org) ---------------------------------------------------------------------- (1) MFLOPS -> cpupower 2 (x/y : -O -g / -O3) SPECint2000 (3) / SPECfp2000 (2) (http//www.spec.org) X.1/ Performances en calcul double csh> cpupower 0 3000000 5 3 10^6 operations doubles - sur memoire 3x3 10^6 doubles (~50 MO) ===> ~ 24 MO / MFLOPS csh> cpupower 2 1.6 10^9 operations doubles - sur 3x20000 doubles (~0.5 MO) Compilation avec -O (optimisation) (1) cpupower 0 : debit memoire en MO/s (2) cpupower 0 , MFLOPS (5) cpupower 2 , MFLOPS Compilation avec -g (debug / sans optimisation) (3) cpupower 0 , MFLOPS (6) cpupower 2 , MFLOPS Compilation avec -O3 ou -fast ...( optimisation poussee) (4) cpupower 0 , MFLOPS (7) cpupower 2 , MFLOPS ---------------------------------------------------------------------------------------------- MFLOPS |(1) MO/s| (2) (3) (4) | (5) (6) (7) ---------------------------------------------------------------------------------------------- (a)xeon-lx-2.4GHz | 1290 | 53 53 55 | 338 340 320 (b)xeon-lx-2.8GHzicc | 2040 | 85 80 83 | 914 409 914 (bb)n-xeon-grid49 | 3000 | 125 85 130 | 660 500 660 (bb)n-xeon4c-grid50 | 2568 | 107 103 109 | 655 500 655 (c)amd-lx | 1560 | 65 77 68 | 666 314 686 (cc)amd2-lx | | (e')osf-cool | 768 | 32 15 32 | 630 150 660 (f)superosf (g)G5-osx-1 GHz | 2100 | 88 68 88 | 1000 255 1073 (f)G4-osx-1.25GHz | 600 | 25 16 25 | 417 93 430 (i)core-osx-1.83GHz | 2500 | 107 75 107 | 855 309 884 (j)xeon-osx (p)ibm-aix-regatta | 3100 | 130 55 133 | 730 115 1750 (32 bits) (p)ibm-aix-meso | 5700 | 240 70 320 | 1500 220 3400 (32 bits) (s)sgi-magique | 336 | 14 7 15 | 340 40 460 (32 bits) ---------------------------------------------------------------------------------------------- X.2/ Comparaison performances int, float double cpupower compile avec -O (1) float , cpupowerF 0 3000000 5 / cpupowerF 2 -> MFLOPS (puissance de calcul sur float) (2) double, cpupowerD 0 3000000 5 / cpupowerF 2 (idem tableau X.1) -> MDBLOPS (puissance de calcul sur float) (3) int, cpupowerI 0 3000000 5 / cpupowerI 2 -> MINTOPS (puissance de calcul sur int=4 bytes) (4) long (ou long long (*)) cpupowerL 0 3000000 5 / cpupowerL 2 -> MLONOPS (puissance de calcul sur long=8 bytes) ---------------------------------------------------------------------------------------------- MFLOPS | (1)MFLOPS (2)MDBLOPS (3)MINTOPS (4)MLONOPS ---------------------------------------------------------------------------------------------- (a)xeon-lx-2.4GHz | (b)xeon-lx-2.8GHzicc | 166/905 90/900 166/1500 88/522 (*) (bb)nxeon-grid49 | 250/1030 125/660 250/2500 125/2280 (bb4)xeon-4c-grid50 | 207/1019 110/660 207/2460 107/2285 (c)amd-lx | 125/695 65/675 125/1570 65/1045 (cc)amd2-lx | (e')osf-cool | 60/635 32/631 62/640 31/630 (f)superosf (g)G5-osx-1 GHz | 180/1260 90/1150 165/940 81/280 (*) (f)G4-osx-1.25GHz | 45/430 25/410 45/710 24/190 (*) (i)core-osx-1.83GHz | 185/919 105/855 187/935 62/246 (*) (j)xeon-osx (p)ibm-aix-regatta | (p)ibm-aix-meso | 250/1150 250/1500 250/1200 50/200 (32 bits) | 280/1500 250/1600 250/1100 210/1000 (64 bits -q64) (s)sgi-magique | ---------------------------------------------------------------------------------------------- X.3/ Comparaison avec SPEC csh> cpupower 0 / cpupower 2 ---------------------------------------------------------------------- MFLOPS(1) SPECfp SPECint ---------------------------------------------------------------------- (b)xeon-lx-2.8GHz 166/900 1400 1400 (c)amd-lx 125/690 1600 1300 (cc)amd2-lx 675 1600 1300 (e)osf-xp1000 32/650 500 400 (f)superosf 842 1100 700 (i)core-osx-1.83GHz 110/850 1400 1500 (j)xeon-osx 2600 2900 - (p)ibm-aix-regatta 130/700 1050 700 ---------------------------------------------------------------------- A/ Compilation tout SOPHYA ---------------------------- csh> time make all (1) ou csh> time make -j 2 all (2) Temps CPU Indice de performance 100*(1000/TCPU) Temps elapsed (vrai) Temps vrai / TCPU ---------------------------------------------------------------------- CPU(s) IndPerf TElapsed , TCPU/Elapsed % ---------------------------------------------------------------------- (a)xeon-lx-2.4GHz (2) 615 s 162 410 s 150% avec -O3 -g (2) 1300 s 77 760 s 172% (b)xeon-lx-2.8GHz (2) 755 s 132 540 s 140% (c)amd-lx (2) 336 s 297 175 s 192% (d)osf-asc (1) 1920 s 52 2340 s 83% (??) (e)osf-xp1000 (1) 533 s 187 660 s 80% (f)superosf (1) 895 s 112 910 s 98% (g)G5-osx-2GHz (2) 453 s 221 250 s 182% -tune=G5 1100 s 90 -g -O 740 s 380 s 195% (h)G4-osx-1.25GHz (1) 660 s 151 710 s 93% [-g] 1500 s 94% [-O2 -g] (i)core-osx-1.83GHz (2) 209 s 478 116 s 180% -O2 (1) 367 s 272 381 96% (j)xeon-osx (p)ibm-aix ---------------------------------------------------------------------- Taille shared libs : (a) (c) 33 MO (f) = (e) = 57 MO (g) 80 MO (i) 83 MO B/ Calcul brut (Tableaux de SOPHYA) avec / sans threads -------------------------------------------------------- B.1.a/ Calcul sur vecteur 10 * V2 ~= DLO4 (V1) ~ 10 x 10 x 9. 10^6 operations double sur 2 x 9 10^6 double 900 M.Ops r_8 / ~ 1500 MO (1) time cpupower 0 # compile avec -O (/ -O -g) (2) time zthr arrdl 1 3000 1 thread (3) time zthr arrdl 2 3000 2 thread (4) time zthr arrdl 4 3000 4 thread (5) time zthr arrdl 6 3000 6 thread (6) time zthr arrdl 8 3000 8 thread ----------------------------------------------------------------------------------- (1)MFLOPS (2)CPU/Elap/% (3)CPU/Elap/% (4)CPU/Elap/% ----------------------------------------------------------------------------------- (a)xeon-lx-2.4GHz 53 (b)xeon-lx-2.8GHz 85 2.6/2.6/100% 5.3/2.9/180% 14.3/4.86/310% (5) 23/7.4/314% (bb)nxeon-grid49 125 2/2/99% 4/2/186% 8/2.45/326% (5) 12/3.6/330% (6) 16/4.7/340% (bb4)xeon-4c-grid50 110 2.2/2.2/99% 4.2/2.3/185% 8/2.5/321% (5) 13/3.1/470% (6) 21/3.8/544% (c)amd-lx 95 (e')osf-cool 32 5.7/5.8/98% 11.1/11.3/98% 22.3/22.5/98% (f)superosf (g)G5-osx-2GHz 88 2.5/2.6/99% 5.9/3.38/184% 11/6.45/173% [-O2 -g] (h)G4-osx-1.25GHz 25 6.6/7/95% 13.4/13.8/97% [-O2 -g] (i)core-osx-1.83GHz 107 2.1/2.1/98% 4.3/2.9/150% 8.3/30/31% [-O2 -g] (j)xeon-osx (p)ibm-aix-regatta 130 (q)ibm-aix-meso 150 0.7/1/70% 1.2/2./60% 3.2/2/150% [-O3] (5) 5.4/3/180% (6) 6.4/3/210% (s)sgi-magique 7 78/78/99% 167/95/175% 339/96/352% [-O -g: NON-OPT] 14 16.4/16.5/99% 33.8/22.4/150% 79/32/250% [-O -g2 OPT] ----------------------------------------------------------------------------------- B.1.b/ Calcul sur vecteur V2 = Sin(V1) + Cos(V1) ~ 50 x 9. 10^6 operations double sur 2 x 9 10^6 double, mem ~ 150 MO ~500 M.Ops r_8 / ~ 600 MO I/O (1) time cpupower 0 # compile avec -O (/ -O -g) (2) time zthr arrmf 1 3000 1 thread (3) time zthr arrmf 2 3000 2 thread (4) time zthr arrmf 4 3000 4 thread (5) time zthr arrmf 6 3000 6 thread (6) time zthr arrmf 8 3000 8 thread ----------------------------------------------------------------------------------- (1)MFLOPS (2)CPU/Elap/% (3)CPU/Elap/% (4)CPU/Elap/% ----------------------------------------------------------------------------------- (a)xeon-lx-2.4GHz 53 (b)xeon-lx-2.8GHz 85 1.7/1.7/100% 3.5/2.1/173% 9.8/3.6/275% (5) 12/3.6/330% (6) 16/4.7/340% (bb)nxeon-grid49 125 1.6/1.6/100% 3.2/1.7/183% 6.7/2.1/314% (5) 10.1/3.2/320% (6) 14.4/4.05/330% (c)amd-lx 95 (e')osf-cool 32 4.2/4.3/98% 8.2/8.4/98% 16.1/16.2/98% (f)superosf (g)G5-osx-2GHz 88 2.3/2.3/100% 5/3/165% 9.6/5.8/167% [-O2 -g] (h)G4-osx-1.25GHz 25 4.5/4.8/95% 10.9/14.6/72% [-O2 -g] (i)core-osx-1.83GHz 107 2.3/2.3/98% 4.8/3.1/158% [-O2 -g] (j)xeon-osx (p)ibm-aix-regatta 130 (q)ibm-aix-meso 150 1./2/50% 2.8/3/86% 5.4/4/130% [-O3] (5) 10/4/250% (6) 11.2/5/220%% (s)sgi-magique 7 11.5/11.7/99% 24/17/140% 51.5/18.4/280% [-O -g NON-OPT] 14 6.5/6.6/99% 13.3/12/110% 34.5/17.3/200% [-O -g3 OPT] ----------------------------------------------------------------------------------- B.1.c/ Version corrige de zthr.cc (apres 23/05/07) arr = (c1*a1) + (c2*a2) ~ 3 x 4. 10^6 operations int_4 sur 3 x 4 10^6 int_4 12 M.Ops int_4 / ~ 50 MO (1) time cpupower 0 # compile avec -O (/ -O -g) (2) time zthr arr 1 2000 1 thread (3) time zthr arr 2 2000 2 thread (4) time zthr arr 4 2000 4 thread (5) time zthr arr 6 2000 6 thread (6) time zthr arr 8 2000 8 thread ----------------------------------------------------------------------------------- (1)MFLOPS (2)CPU/Elap/% (3)CPU/Elap/% (4)CPU/Elap/% ----------------------------------------------------------------------------------- (a)xeon-lx-2.4GHz 53 0.5/1/43% 1/1.1/88% 2.8/1/262% (5) 4.5/1.8/246% (6) 6.1/2.1/310% (b)xeon-lx-2.8GHz 65 (c)amd-lx 95 0.23/1/22% 0.44/1/51% 1/1/102% [-O -g] (5) 1.6/1/106% (6) 2.2/1.2/100% (e')osf-cool 32 0.43/1.2/35% 0.6/1.33/44% 1.1/1.3/82% [-O -g] (5) 1.45/1.7/85% (6) 1.83/2.16/84% (f)superosf (g)G5-osx-2GHz 88 1.5/1.5/100% 3.2/1.7/185% 6.6/3.5/188% [-O -g] (g)G5-osx-2GHz 88 0.4/1/40% 0.9/1.0/90% 2/1.2/169% [-tune=G5 -g] (5) 3.3/2/164% (6) 4.3/2.6/165% (h)G4-osx-1.25GHz 25 3/3/95% [-O2 -g] (i)core-osx-1.83GHz [-O2 -g] (j)xeon-osx (p)ibm-aix-regatta 130 (q)ibm-aix-meso 150 0.6/1/58% 1/1/91% 1.7/1.2/132% [-O3] (5) 2.4/1.2/193% (6) 4.25/1.6/265% ----------------------------------------------------------------------------------- B.1.x/ ancienne version de zthr (avant 23/05/07) On faisait 2 multiplications par ctye suivi d'un produit matriciel ! arr = c1*a1*c2*a2 ( ~ 3 10^6 op. double) (1) time cpupower 2 # compile avec -O3 (/ -O -g) (2) time zthr arr 1 1000 1 thread (3) time zthr arr 2 1000 2 thread (4) time zthr arr 4 1000 4 thread (5) time zthr arr 6 1000 6 thread (6) time zthr arr 8 1000 8 thread ----------------------------------------------------------------------------------- (1)MFLOPS (2)CPU/Elap/% (2)IndPerf (3)CPU/Elap/% (3)IndPerf ----------------------------------------------------------------------------------- (a)xeon-lx-2.4GHz 1167 5.15/5.2/99% 11.4/5.8/196% (4)36.6/9.28/394% -O3 -g 4.9/5./99% (b)xeon-lx-2.8GHz 920 2.3/2.3/100% 6.2/3.1/198% (4)26/6.6/396% (c)amd-lx 690 3.6/3.6/99% 6.8/4/171% (4)13.5/7/193% (5)20.3/10.23/198% (cc)amd2-lx 675 2/2/99% 4.15/2.1/197% (4)8.25/4.15/198% (5)13.6/4.6/292% (6)19.8/6.5/300% (d)osf-asc 420 6.3s/6.5s/99% 16.9/8.8/192% (4)29.9/15.7/191% (e)osf-xp1000 648 5.1/5.3/96.6% 11.4/11.4/99% (4)25.2/25.5/99% (f)superosf 842 2.87/2.88/99.6% 6.25/4.1/153% (4)11.6/3.06/379% (h)G4-osx-1.25GHz 92 44s/48s/91% 86.7/99.8/92% [-g] 380 12.2/12.9/95% 24/25.3/95% [-O2 -g] (g)G5-osx-2GHz 1151 20s/20s/99% 40s/23s/170% (4) 80.8/45/180% -O -g 4.5/4.9/91% 9.3/4.7/197% (4) 18.3/9.4/197% -tune=G5 3.35/3.8/88% 7.1/3.6/196% (h)G4-osx-1.25GHz 92 44s/48s/91% 86.7/99.8/92% [-g] 380 12.2/12.9/95% 24/25.3/95% [-O2 -g] (4) 14/7.5/187% (i)core-osx-1.83GHz 855 11.5/11.5/100% 23/11.6/192% [-g] (4) 46/23/199% -O2 3.85/3.89/99% 7.7/3.9/198% [-O2 -g] (4) 15.4/7.77/198% (j)xeon-osx 2600 2.5/2.5/100% 5.1/2.6/199% (4) 11.5/3.2/362% (5) 17.4/4.77/365% (p)ibm-aix-regatta 1750/730 6.8/6.9/98% 13.1/6.75/195% (4) 26.3/11.7/225% (q)ibm-aix-meso 3600/1250 3.6/3.75/96% 7.35/3.7/197% (4) 12.46/4.2/298% (5) 219/6.7/280% (6) 24/4.5/530% (s)sgi-magique 460 60/60/99% ----------------------------------------------------------------------------------- B.2/ Multiplication de matrices mtx = mtx1 * mtx2 ~ 2 10^9 op. double / thread (1) time cpupower 2 (-O3 / -O -g) (2) time zthr mtx 1 1000 1 thread (3) time zthr mtx 2 1000 2 thread (4) time zthr mtx 4 1000 4 thread (5) time zthr mtx 6 1000 6 thread (6) time zthr mtx 8 1000 8 thread ----------------------------------------------------------------------------------- (1)MFLOPS (2)CPU/Elap/% (2)IndPerf (3)CPU/Elap/% (3)IndPerf ----------------------------------------------------------------------------------- (a)xeon-lx-2.4GHz 1167 6.5/6.5/100% 17.4/8.8/198% (4) 80.5/20.3/397% (5) 114.5/29.6/387% (6) 160/40.3/387% (b)xeon-lx-2.8GHz 920 3.4/3.4/100% 12/6.1/199% (4) 55.8/14/400% (5) 79.5/20.3/392% (6) 102/25.8/396% (bb)nxeon-grid49 660 4.3/4.3/100% 9.3/4.7/200% (4) 28.8/7.3/390% (5) 41/10.4/393% (6) 57.5/14.45/397% (bb4)xeon-4c-grid50 660 4.7/4.7/100% 10.7/5.15/199% (4) 27.8/7.1/391% (5) 57.4/10.6/540% (6) 129/16.8/776% (c)amd-lx 690 6.98/6.98/100% 14.1/8.15/173% (4) 27.7/14.23/194% (5) 41.4/21.07/196% (6) 55.4/27.9/198.7% (cc)amd2-lx 675 4.1/4.1/100% 9.55/4.8/198% (4) 20/10.27/195% (5) 32.8/11.16/294% (6) 42.75/13.8/309% (d)osf-asc 420 13.5s/13.7s/98% 32/16.5/194% (4) 67.5/34.4/196% (e)osf-xp1000 648 13/14.1/92% 27.1/27.4/99% (4) 54/54.7/99.6% (5) 80.6/81/99.6% (6) 107.8/108.3/99.5% (e')osf-cool 13/13.22/98% 26/26.1/99% (4) 51.8/51.9/99% (f)superosf 842 6.1/7.24/84% 12.35/6.29/196% (4) 24.3/6.31/385% (5) 36.5/10.9/335% (6) 50.1/18.15/276% (g)G5-osx-2GHz 1151 23/23.7/97% 46.5/27.5/170% (4) 93.4/49.4/189% -O -g 6.2/6.2/100% 14.2/7.2/197% (4) 28.3/14.36/197% -tune=G5 5.7/5.8/98% 13.3/6.8/197% (4) 26.8/13.56/197% (6) 53.8/27.25/197% (h)G4-osx-1.25GHz 333 23.5/24.5/96% [-O2] (i)core-osx-1.83GHz 855 12.6/12.7/100% 25.8/13.4/194% (4) 51.6/26/199% -O2 4.25/4.5/94% 10.6/5.36/198% (4) 20.87/10.68/198% -O2 2 jobs // 2 x 5/5.4/92% (j)xeon-osx 2600 2.8/2.8/99% 9.3/4.66/199% (4) 31.4/8.6/364% (5) 47.1/12.96/364% (6) 62.8/17.38/362% (p)ibm-aix-regatta 1750/730 9.5/9.7/98% 18.3/16.0/114% (4) 38.3/24.7/155% (p)ibm-aix-meso 3600/1250 2.3/2.3/99% 5.1/2.64/194% (compil avec -O3) (4) 11.4/4.16/272% (5) 20.2/5.85/344% (6) 29.9/6.74/442% (s)sgi-magique 400 44/44.3/99% 96.5/55/176% ----------------------------------------------------------------------------------- B.4/ Operations sur tableaux doubles- mesures avec spar csh> time spar 2 1 2000 2000 (1) cpupower 2 MFLOPS (2) MFLOPS (double) spar (3) time spar 2 5 1000 2000 CPU/Elap/% ----------------------------------------------------------------------------------- (1)MFLOPS (2)CPU / % (3)CPU/Elap/% ----------------------------------------------------------------------------------- (a)xeon-lx-2.4GHz 53 ~ 20-35 MFLOPS , 90% 20/20.2/99% [-g -O] (b)xeon-lx-2.8GHz 65 (c)amd-lx 95 ~ 20-40 MFLOPS , 99% 17.2/17.2/100% [-g -O] (d)osf-asc (e)osf-xp1000 32 ~ 15-25 MFLOPS , 90% 37.6/41.2/91% [-g -O] (f)superosf (g)G5-osx-2GHz 88 ~ 10-25 MFLOPS , 99% 45/45/100% [-g -O] ou [-g -O2] (h)G4-osx-1.25GHz 25 ~ 8-16 MFLOPS , 92% 45.5/52/90% [-g -O2] (i)core-osx-1.83GHz (j)xeon-osx (p)ibm-aix-regatta 130 (q)ibm-aix-meso 150 ~ 80-100 MFLOPS , 90% 5./23/22% [-O3] (s)sgi-magique 460 ----------------------------------------------------------------------------------- B.5/ Calcul/comparaison avec JET/tjet csh> time tjet 10 2000 2000 OU tjet 10 2000 1000 (1) TCPU EltAccess C/pointeurs (2) TCPU m1*c1+m2*c2+m3*c3 C/pointeurs (3) TCPU EltAccess SOPHYA (4) TCPU m1*c1+m2*c2+m3*c3 SOPHYA / Methodes (MultCst, AddArr ...) (5) TCPU m1*c1+m2*c2+m3*c3 SOPHYA JET ----------------------------------------------------------------------------------- (1) (2) (3) (4) (5) ----------------------------------------------------------------------------------- (b)xeon-lx-2.8GHz-icc 0.87 0.63 1.55 2.7/1.6 0.57 (c)amd-lx 0.94 0.79 1.85 3.4/2.1 0.76 (e')osf-cool 2.85 2.45 3.1 6.5/5.5 4.1 (g)G5-osx-2GHz 1.5 0.61 2.1 4.1/1.6 0.6 (-g -O2) -tune=G5 -fast 1.1 0.62 1.3 4.1/1.6 0.58 (h)G4-osx-1.25GHz 3.86 2.2 5 9.4/6.2 3 (-g -O2) (i)core-osx-1.83GHz 1.1 0.49 1.6 2.8/1.7 0.68 (q)ibm-aix-meso 0.43 0.27 0.52 1.12/0.75 0.35 (s)sgi-magique 2.45 1.9 5.65 7.45/6.3 2.8 (-O -g3) ----------------------------------------------------------------------------------- C/ Calcul fft (FFTW , FFTPack ) ------------------------------- (1) time cpupower 2 (2) time tfft 2000000 W d 0 0 (avec FFTW) (3) time tfft 2000000 P d 0 0 (avec FFTPack_Sophya) IndPerf=1000/TCPU ----------------------------------------------------------------------------------- (1)MFLOPS (2)CPU/Elap/% (2)IndPerf (3)CPU/Elap/% (3)IndPerf ----------------------------------------------------------------------------------- (a)xeon-lx-2.4GHz 1167 5.5/5.6/97% 180 7.4/7.4/100% 135 -O3 -g 6.8/7.1/96% 147 (b)xeon-lx-2.8GHz 920 3.6/3.7/98% 3.7/3.8/99% ~2x 6.5/8.8/73% ~4x 8/14/55% (15 sec elapsed) (bb)nxeon-grid49 660 3.3/3.3/99% ~2x 3.3/3.4/100% (4 sec elapsed) ~4x 4.6/4.6/99% (5 sec elapsed) ~8x 4.5/4.5/50% (9 sec elapsed) (bb4)xeon-4c-grid50 660 3.6/3.6/99% ~2x 3.6/3.6/100% (4 sec elapsed) ~4x 4.4/4.4/99% (5 sec elapsed) ~8x 7.1/7.1/99% (7 sec elapsed) ~16x 7.1/14.4/50% (15 sec elapsed) (c)amd-lx 690 3.2/3.75/86% 4.2/4.2/100% 238 ~2x 4.7/4.7/99% (cc)amd2-lx 675 2.8/2.8/99% 3.56/3.58/99% (d)osf-asc 420 13.3/13.9/95.7% 75 12.2/17.6/70% 82 (e)osf-xp1000 648 9.9/10.2/97% 101 9.3/9.46/98.5% 107 (e')cool 9.2/10.1/91% 9.1/9.3/98% (f)superosf 842 6./6.22/96.7% 166 5.1/5.18/97.4% 196 (g)G5-osx-2GHz 1151 8.5/8.7/97.5% 120 11.5/11.6/99% 87 -O -g 5.1/5.2/98% 190 5.4/5.5/99% 180 -tune=G5 5/5.1/98% 200 5.5/5.6/97% 180 (h)G4-osx-1.25GHz 92 15.2/15.9/96% 66 23.8/34.1/70% 42 [-g] 380 8.8/10.3/86% 14.7/20.9/65% [-O2 -g] (i)core-osx-1.83GHz 855 4.6/4.75/97% 217 7/7.08/99% 142 -O2 3.2/3.2/99% 312 3.6/3.6/99% 277 -O2 2 jobs // 2 x 3.9/4.3/90% 2x 4.7/5/91% 200 (j)xeon-osx 2600 2.6/2.6/98% 384 2 jobs // ~2x 3.6/4.1/87% 250 4 jobs // (p)ibm-aix-regatta 1750/730 6.25/18.9/33% 160 5.25/15.7/33% 190 (q)ibm-aix-meso 3600/1250 3.95/4.3/91% 250 3.82/4./94% 260 2 jobs // ~2x 3.88/4.2/92% 250 2.8/4.76/59% 2.1/4.45/47% [-O3] ~2x 2.5/4.7/54% ~4x 2.5/4.8/55% (s)sgi-magique 460 22/22/98% 42 24.5/25/99% 40 ----------------------------------------------------------------------------------- D/ Calcul inversion par lapack ------------------------------- lpk inverse 1000,1000 0 ---> temps de calcul inversion par lapack ------------------------------------------------------------------------------------------- CPU/Elap/% (1) ------------------------------------------------------------------------------------------- (a)xeon-lx-2.4GHz 5.6/~100% (b)xeon-lx-2.8GHz 5.34/~90% (c)amd-lx 5.5/5.5/99% (d)osf-asc (e')cool 2.8/2.9/95% (f)superosf (f)G4-osx-1.25GHz 2.3/~100% [-O2 -g] (h)G5-osx-2GHz 0.8/~100% -O -g 0.86/~100% (i)core-osx-1.83GHz 1.93/~100% -O2 (p)ibm-aix-regatta (q)ibm-aix-meso 0.55/~100% (s)sgi-magique 5.3/~90% -------------------------------------------------------------------------------------------- K/ Efficacite de gestion de lock (mutex) avec les threads et tableaux ---------------------------------------------------------------------- (32 threads - operant sur 2000 vecteurs ~ 64000 lock/unlock/wait/broadcast) (1) time zthr syncp 32 2000 4 (2) time zthr sync 32 2000 4 (1) time zthr syncp 4 15000 130 (2) time zthr sync 4 15000 130 ------------------------------------------------------------------------------------------- CPU/Elap/% (1) (2) (3) (4) ------------------------------------------------------------------------------------------- (a)xeon-lx-2.4GHz 23.5/14/168% 4.3/1.2/365% 7.9/5.5/142% 4/2.15/190% Avant ThSafeOp 17/178% avec -O3 -g (b)xeon-lx-2.8GHz (2) (bb4)xeon-4c-grid50 1/1/96% 1.6/1.2/133% 4.9/4.05/120% 6/5/122% (c)amd-lx 0.6/1/63% 0.6/1/60% 3.5/3.5/102% 2.6/2.7/98% (d)osf-asc 4.5/3.4/132% 3.35/2/170% 15.8/10.5/150% 13/8/163% 5.4/100%(NoThSafe) (e')cool 1.3/1.37/95% 1.35/1.5/89% 5.3/5.3/99% 5.2/5.2/99% (e)superosf (1) (g)G5-osx-2GHz (2) 2.6/130% (NoThSafe) -O -g 2.6/1.7/150% 6.8/3.7/187% 4/2.75/142% 4.7/3/155% (h)G4-osx-1.25GHz (1) 40.5/42.6/95% 42.2/43.5/97% [-g] 3.9/4.3/89% 3.5/4/89% 3.8/4/95% 4.3/4.6/93% (i)core-osx-1.83GHz 7.7/7.1/108% 7.8/6.7/116% 30.2/29.6/102% 30.3/29.2/104% [-g] -O2 2.7/1.8/152% 6/3.16/190% 3.4/2.4/142% 3.2/2.5/164% [-O2 -g] (j)xeon-osx Avant ThSafeOp 2.55/143% (p)ibm-aix-regatta 4.7/111% (q)ibm-aix-meso 7.5/2.8/300% 17/3.8/450% 8.2/3.05/270% 4.85/2.43/200% -------------------------------------------------------------------------------------------- L/ I/O et PPF ----------------- Ecriture/lecture de n=10^7 lignes de int+6double, Total ~ 500 MO (1) time tstdtable w xx.ppf swap 10000000 1024 0 (2) time tstdtable r xx.ppf swap 10000000 1024 0 (3) time tstdtable w xx.ppf swap 50000000 1024 0 (4) time tstdtable r xx.ppf swap 50000000 1024 0 ------------------------------------------------------------------------------------------- CPU/Elap/% (1) (2) (3) (4) ------------------------------------------------------------------------------------------- (a)xeon-lx-2.4GHz 17/26/63% 5.5/5.6/94% (b)xeon-lx-2.8GHz 7/18.5/40% 2.7/2.8/100% 7000000 (c)amd-lx 5.9/6./97% 3.4/3.4/100% 30/32/93% 24/165/13% ? (d)osf-asc (e')cool 15/30/50% 13/13/99% (f)superosf (g)G5-osx-2GHz 14/14.2/98% 6/6.14/99% (h)G4-osx-1.25GHz 26/29.7/87% 15.7/38.6/41% [-O2 -g] (i)core-osx-1.83GHz 37/37.8/98% 22/48.4/45% [-g] -O2 10.5/17.4/55% 11.2/41.2/27% [-O2 -g] (p)ibm-aix-regatta (q)ibm-aix-meso 5.5/16.8/38% 5.7/13.2/43% 32.7/85/39% 29/60/49% 6/11/55% 4/9/44% 6.3/10.6/60% 4/6/64% 2 lecture // ~2x 4/6/60% Elapsed 7 sec 1 write + 2 read // 6.3/10.4/60% ~2 3.8/6.4/59% Elapsed < 11 sec 1 write + 3 read // 19 sec 4 read // 6 sec --------------------------------------------------------------------------------------------