tracer:ull - 2003 barcelona, october 25th, 2003 tracer.ull.es tic2002-04498-c05-05

28
Advanced Optimization Techniques for Complex Problems Técnicas de Optimización Avanzadas para Problemas Complejos TRACER:ULL - 2003 Barcelona, October 25th, 2003 http://www.tracer.ull.es TIC2002-04498-C05-05 University of La Laguna

Upload: yoshe

Post on 18-Mar-2016

55 views

Category:

Documents


0 download

DESCRIPTION

Advanced Optimization Techniques for Complex Problems Técnicas de Optimización Avanzadas para Problemas Complejos. TRACER:ULL - 2003 Barcelona, October 25th, 2003 http://www.tracer.ull.es TIC2002-04498-C05-05 University of La Laguna. Outline. Objectives Researchers Problems - PowerPoint PPT Presentation

TRANSCRIPT

  • Advanced Optimization Techniques for Complex Problems Tcnicas de Optimizacin Avanzadas para Problemas ComplejosTRACER:ULL - 2003

    Barcelona, October 25th, 2003

    http://www.tracer.ull.es

    TIC2002-04498-C05-05

    University of La Laguna

  • OutlineObjectivesResearchersProblemsBranch and Bound and Divide and Conquer SkeletonsKnapsack ProblemMatrix ProductConstrained two-dimensional cutting stock problemCALL and LLAC: tools for Complexity AnalysisSymbolic regression ProblemAn analytical model for Pipeline and Master-Slave algorithms over heterogeneous clustersResource allocation problemPrediction of the RNA Secondary Structure problemResults

  • TRACER::ULL ObjectivesThe TRACER::ULL main objective is to achieve an efficient resolution of the following complex problems by developing new optimization procedures:Constrained two-dimensional cutting stock problemSymbolic regression problemPrediction of the RNA secondary structure problemWe propose the design, implementation and evaluation of solving tools using exact techniques:Divide and Conquer Branch and Bound Dynamic ProgrammingIt is an objective to provide sequential, parallel and distributed implementations for academia problems:Resource allocation problemKnapsack problemMatrix ProductA second research track is related with the building of a methodology and the associated tool for the complexity and performance analysis of both sequential and parallel algorithms.Another goal is the implementation ofAn Internet execution systemsA Problem repositoryPerformance AnalysisWeb site: http://www.tracer.ull.es

  • ResearchersULL StaffCoromoto Len HernndezIsabel Dorta GonzlezDaniel Gonzlez MoralesCasiano Rodrguez LenJess Alberto Gonzlez MartnezForeingRumen AndonovStudentsJuan Ramn Gonzlez GonzlezGara Miranda ValladaresMara Dolores Medina Barroso

    GrantsBranch and BoundDynamic ProgrammingPerformance Analysis ToolsandSymbolic regression problemDivide and Conquertwo dimensional cutting stock problemPrediction of the RNA secondary structure problem

  • Shared Memory Branch and Bound Skeletons// shared variables {bqueue, bstemp, soltemp, data} // private variables {auxSol, high, low} // the initial subproblem is already inserted in the global shared queue while(!bqueue.empty()) { nn = bqueue.getNumberOfNodes(); nt = (nn > maxthread)?maxthread:nn; data = new SubProblem[nt]; for (int j = 0; j < nt; j++) data[j] = bqueue.remove(); set.num.threads(nt); parallel forall (i = 0; i < nt; i++) { high = data[i].upper_bound(pbm,auxSol); if ( high > bstemp ) { low = data[i].lower_bound(pbm,auxSol); if ( low > bstemp ) { // critical region // only one thread can change the value at any time bstemp = low; soltemp = auxSol; } if ( high != low ) { // critical region // just one thread can insert subproblems in the queue at any time data[i].branch(pbm,bqueue); } } } }bestSol = bstemp;sol = soltemp;

  • 0-1 Knapsack ProblemThe 0/1 Knapsack Problem can be stated as follows:

    "We have been provided with a knapsack of capacity C and with a set of N objects; p[k] and w[k] are the profit and weight associated to object k. Without exceeding the capacity of the knapsack, the objects must be inserted into the knapsack providing the maximum profit". Martello, S., Toth, P. : Knapsack Problems Algorithms and Computer Implementatios. John Wiley & Sons Ltd. (1990)

  • Comparison between MPI and OpenMP skeletonsOrigin 3000- CIEMAT

  • Distributed Branch and Bound skeletonInitialization PhaseResolution PhaseConditional CommunicationMessage ReceptionAvoiding starvationComputeBest bound PropagationWork queryingEnding resolution phaseSolution Building

  • Distributed Branch and Bound skeleton

  • Distributed Branch and Bound skeleton

    Grfico3

    1.85462562041.81391680921.83427121481.8107320389

    2.66282361762.58518007033.03875658592.5422060483

    3.25238061693.10915146134.08869679533.1438695605

    3.68530961644.6123264595

    4.2342436156

    4.6189117154

    ULL 500 Mhz

    ULL 800 Mhz

    ULL 800-500 Mhz

    ULL 1400 Mhz

    number of processors

    speedup

    Knapsack N = 50.000 ULL

    Sequential

    89

    50000

    run18989

    Problema:50000100000

    Computednodes:2614392750821.8784182653.43164063

    1639893780Sequential623971637800747.323883052608.43508913

    1639893780623971637500006996.1347665002850.596435571546.03527857

    run21400429.94886781234.27526861

    Problema:

    Computednodes:2614392

    1639893780Sequential623971637

    163989378062397163750000821.878418

    beowulf5

    run1

    Problema:

    Computednodes:2614392

    1639893780Sequential623971637

    163989378062397163750000747.8862305

    run2

    Problema:

    Computednodes:2614392

    1639893780Sequential623971637

    163989378062397163750000746.7615356

    beowulf8

    run1

    Problema:

    Computednodes:2614392

    1639893780Sequential623971637

    1639893780623971637500002850.368164

    run2

    Problema:

    Computednodes:2614392

    1639893780Sequential623971637

    1639893780623971637500002850.824707

    beowulf16

    run1

    Problema:

    Computednodes:2614392

    1639893780Sequential623971637

    163989378062397163750000437.1088867

    run2

    Problema:

    Computednodes:2614392

    1639893780Sequential623971637

    163989378062397163750000422.7888489

    89

    100000

    beowulf0

    run1

    Problema:

    Computednodes:627786

    6562332564Sequential2499610869

    65623325642499610873100000651.1342163

    run2

    Problema:

    Computednodes:627786

    6562332564Sequential2499610869

    65623325642499610873100000655.7290649

    beowulf5

    run1

    Problema:

    Computednodes:627786

    6562332564Sequential2499610869

    65623325642499610873100000608.3323364

    run2

    Problema:

    Computednodes:627786

    6562332564Sequential2499610869

    65623325642499610873100000608.5378418

    beowulf8

    run1

    Problema:

    Computednodes:627786

    6562332564Sequential2499610869

    656233256424996108731000001546.311768

    run2

    Problema:

    Computednodes:627786

    6562332564Sequential2499610869

    656233256424996108731000001545.758789

    beowulf16

    run1

    Problema:

    Computednodes:627786

    6562332564Sequential2499610869

    65623325642499610873100000230.1702271

    run2

    Problema:

    Computednodes:627786

    6562332564Sequential2499610869

    65623325642499610873100000238.3803101

    107

    50000

    beowulf0

    run1

    Problema:

    Computednodes:155149

    1641207946Sequential628145372

    16412079466281453755000082.7580719

    run2

    Problema:

    Computednodes:155149

    1641207946Sequential628145372

    16412079466281453755000082.42076111

    beowulf5

    run1

    Problema:

    Computednodes:155149

    1641207946Sequential628145372

    16412079466281453755000081.75357819

    run2

    Problema:

    Computednodes:155149

    1641207946Sequential628145372

    16412079466281453755000081.58392334

    beowulf8

    run1

    Problema:

    Computednodes:155149

    1641207946Sequential628145372

    164120794662814537550000239.3715057

    run2

    Problema:

    Computednodes:155149

    1641207946Sequential628145372

    164120794662814537550000238.8070831

    beowulf16

    run1

    Problema:

    Computednodes:155149

    1641207946Sequential628145372

    16412079466281453755000033.86289597

    run2

    Problema:

    Computednodes:155149

    1641207946Sequential628145372

    16412079466281453755000033.95829773

    107

    100000

    beowulf0

    run1

    Problema:

    Computednodes:701008

    6569265691Sequential2510207120

    65692656912510207122100000629.5508423

    run2

    Problema:

    Computednodes:701008

    6569265691Sequential2510207120

    65692656912510207122100000626.1575928

    beowulf5

    run1

    Problema:

    Computednodes:701008

    6569265691Sequential2510207120

    65692656912510207122100000585.0124512

    run2

    Problema:

    Computednodes:701008

    6569265691Sequential2510207120

    65692656912510207122100000575.9658813

    beowulf8

    run1

    Problema:

    Computednodes:701008

    6569265691Sequential2510207120

    656926569125102071221000001455.44397

    run2

    Problema:

    Computednodes:701008

    6569265691Sequential2510207120

    656926569125102071221000001429.06665

    beowulf16

    run1

    Problema:

    Computednodes:701008

    6569265691Sequential2510207120

    65692656912510207122100000211.1840057

    run2

    Problema:

    Computednodes:701008

    6569265691Sequential2510207120

    65692656912510207122100000215.8686676

    127

    50000

    beowulf0

    run1

    Problema:

    Computednodes:1157429

    1639547063Sequential623661898

    163954706362366190050000435.9469299

    run2

    Problema:

    Computednodes:1157429

    1639547063Sequential623661898

    163954706362366190050000437.7729187

    beowulf5

    run1

    Problema:

    Computednodes:1157429

    1639547063Sequential623661898

    163954706362366190050000425.9438477

    run2

    Problema:

    Computednodes:1157429

    1639547063Sequential623661898

    163954706362366190050000422.4737549

    beowulf8

    run1

    Problema:

    Computednodes:1157429

    1639547063Sequential623661898

    1639547063623661900500001428.572632

    run2

    Problema:

    Computednodes:1157429

    1639547063Sequential623661898

    1639547063623661900500001413.713501

    beowulf16

    run1

    Problema:

    Computednodes:1157429

    1639547063Sequential623661898

    163954706362366190050000204.9870148

    run2

    Problema:

    Computednodes:1157429

    1639547063Sequential623661898

    163954706362366190050000215.970047

    127

    100000

    beowulf0

    run1

    Problema:

    Computednodes:1439587

    6567126329Sequential2496892260

    656712632924968922601000001550.543457

    run2

    Problema:

    Computednodes:1439587

    6567126329Sequential2496892260

    656712632924968922601000001543.898315

    beowulf5

    run1

    Problema:

    Computednodes:1439587

    6567126329Sequential2496892260

    656712632924968922601000001448.219482

    run2

    Problema:

    Computednodes:1439587

    6567126329Sequential2496892260

    656712632924968922601000001448.633545

    beowulf8

    run1

    Problema:

    Computednodes:1439587

    6567126329Sequential2496892260

    656712632924968922601000003613.065186

    run2

    Problema:

    Computednodes:1439587

    6567126329Sequential2496892260

    656712632924968922601000003658.391113

    beowulf16

    run1

    Problema:

    Computednodes:1439587

    6567126329Sequential2496892260

    65671263292496892260100000531.3937988

    run2

    Problema:

    Computednodes:1439587

    6567126329Sequential2496892260

    65671263292496892260100000558.8529663

    607

    50000

    beowulf0

    run1

    Problema:

    Computednodes:628256

    1641653455Sequential625216905

    164165345562521690750000241.8257904

    run2

    Problema:

    Computednodes:628256

    1641653455Sequential625216905

    164165345562521690750000241.2855225

    beowulf5

    run1

    Problema:

    Computednodes:628256

    1641653455Sequential625216905

    164165345562521690750000236.5175629

    run2

    Problema:

    Computednodes:628256

    1641653455Sequential625216905

    164165345562521690750000236.1403961

    beowulf8

    run1

    Problema:

    Computednodes:628256

    1641653455Sequential625216905

    164165345562521690750000801.116272

    run2

    Problema:

    Computednodes:628256

    1641653455Sequential625216905

    164165345562521690750000805.5685425

    beowulf16

    run1

    Problema:

    Computednodes:628256

    1641653455Sequential625216905

    164165345562521690750000117.050354

    run2

    Problema:

    Computednodes:628256

    1641653455Sequential625216905

    164165345562521690750000118.3657761

    607

    100000

    beowulf0

    run1

    Problema:

    Computednodes:1417954

    6569912055Sequential2505118086

    656991205525051180881000001384.396118

    run2

    Problema:

    Computednodes:1417954

    6569912055Sequential2505118086

    656991205525051180881000001387.828979

    beowulf5

    run1

    Problema:

    Computednodes:1417954

    6569912055Sequential2505118086

    656991205525051180881000001269.646729

    run2

    Problema:

    Computednodes:1417954

    6569912055Sequential2505118086

    656991205525051180881000001270.676392

    beowulf8

    run1

    Problema:

    Computednodes:1417954

    6569912055Sequential2505118086

    656991205525051180881000003316.035645

    run2

    Problema:

    Computednodes:1417954

    6569912055Sequential2505118086

    656991205525051180881000003296.381348

    beowulf16

    run1

    Problema:

    Computednodes:1417954

    6569912055Sequential2505118086

    65699120552505118088100000491.7569275

    run2

    Problema:

    Computednodes:1417954

    6569912055Sequential2505118086

    65699120552505118088100000504.0476074

    709

    50000

    beowulf0

    run1

    Problema:

    Computednodes:860072

    1643145388Sequential623786837

    164314538862378683950000337.1434326

    run2

    Problema:

    Computednodes:860072

    1643145388Sequential623786837

    164314538862378683950000337.1317749

    beowulf5

    run1

    Problema:

    Computednodes:860072

    1643145388Sequential623786837

    164314538862378683950000332.9222412

    run2

    Problema:

    Computednodes:860072

    1643145388Sequential623786837

    164314538862378683950000329.8166809

    beowulf8

    run1

    Problema:

    Computednodes:860072

    1643145388Sequential623786837

    1643145388623786839500001129.134155

    run2

    Problema:

    Computednodes:860072

    1643145388Sequential623786837

    1643145388623786839500001136.878052

    beowulf16

    run1

    Problema:

    Computednodes:860072

    1643145388Sequential623786837

    164314538862378683950000165.2099915

    run2

    Problema:

    Computednodes:860072

    1643145388Sequential623786837

    164314538862378683950000164.4545288

    709

    100000

    beowulf0

    run1

    Problema:

    Computednodes:2106991

    6553347953Sequential2501446153

    655334795325014461541000003220.858154

    run2

    Problema:

    Computednodes:2106991

    6553347953Sequential2501446153

    655334795325014461541000003218.05835

    beowulf5

    run1

    Problema:

    Computednodes:2106991

    6553347953Sequential2501446153

    655334795325014461541000003260.966064

    run2

    Problema:

    Computednodes:2106991

    6553347953Sequential2501446153

    655334795325014461541000003263.137939

    beowulf8

    run1

    Problema:

    Computednodes:2106991

    6553347953Sequential2501446153

    655334795325014461541000007526.094727

    run2

    Problema:

    Computednodes:2106991

    6553347953Sequential2501446153

    655334795325014461541000007687.610352

    beowulf16

    run1

    Problema:

    Computednodes:2106991

    6553347953Sequential2501446153

    655334795325014461541000001126.287842

    run2

    Problema:

    Computednodes:2106991

    6553347953Sequential2501446153

    655334795325014461541000001348.139526

    809

    50000

    beowulf0

    run1

    Problema:

    Computednodes:564492

    1640204504Sequential624972454

    164020450462497245550000188.7664185

    run2

    Problema:

    Computednodes:564492

    1640204504Sequential624972454

    164020450462497245550000188.4552155

    beowulf5

    run1

    Problema:

    Computednodes:564492

    1640204504Sequential624972454

    164020450462497245550000185.1395874

    run2

    Problema:

    Computednodes:564492

    1640204504Sequential624972454

    164020450462497245550000185.0117645

    beowulf8

    run1

    Problema:

    Computednodes:564492

    1640204504Sequential624972454

    164020450462497245550000634.4938965

    run2

    Problema:

    Computednodes:564492

    1640204504Sequential624972454

    164020450462497245550000630.4830322

    beowulf16

    run1

    Problema:

    Computednodes:564492

    1640204504Sequential624972454

    16402045046249724555000091.59538269

    run2

    Problema:

    Computednodes:564492

    1640204504Sequential624972454

    16402045046249724555000091.65031433

    809

    100000

    beowulf0

    run1

    Problema:

    Computednodes:1975641

    6564250986Sequential2504513367

    656425098625045133721000002053.576904

    run2

    Problema:

    Computednodes:1975641

    6564250986Sequential2504513367

    656425098625045133721000002054.891846

    beowulf5

    run1

    Problema:

    Computednodes:1975641

    6564250986Sequential2504513367

    656425098625045133721000001939.881958

    run2

    Problema:

    Computednodes:1975641

    6564250986Sequential2504513367

    656425098625045133721000001918.79248

    beowulf8

    run1

    Problema:

    Computednodes:1975641

    6564250986Sequential2504513367

    656425098625045133721000005011.946777

    run2

    Problema:

    Computednodes:1975641

    6564250986Sequential2504513367

    656425098625045133721000004837.681152

    beowulf16

    run1

    Problema:

    Computednodes:1975641

    6564250986Sequential2504513367

    65642509862504513372100000771.5793457

    run2

    Problema:

    Computednodes:1975641

    6564250986Sequential2504513367

    65642509862504513372100000743.9428101

    947

    50000

    beowulf0

    run1

    Problema:

    Computednodes:1221939

    1645790850Sequential624198382

    164579085062419838450000425.0844116

    run2

    Problema:

    Computednodes:1221939

    1645790850Sequential624198382

    164579085062419838450000425.4591064

    beowulf5

    run1

    Problema:

    Computednodes:1221939

    1645790850Sequential624198382

    164579085062419838450000408.0883179

    run2

    Problema:

    Computednodes:1221939

    1645790850Sequential624198382

    164579085062419838450000407.8194275

    beowulf8

    run1

    Problema:

    Computednodes:1221939

    1645790850Sequential624198382

    1645790850624198384500001501.571655

    run2

    Problema:

    Computednodes:1221939

    1645790850Sequential624198382

    1645790850624198384500001514.792847

    beowulf16

    run1

    Problema:

    Computednodes:1221939

    1645790850Sequential624198382

    164579085062419838450000224.7654114

    run2

    Problema:

    Computednodes:1221939

    1645790850Sequential624198382

    164579085062419838450000220.1774597

    947

    100000

    beowulf0

    run1

    Problema:

    Computednodes:2038624

    6571517811Sequential2503042370

    657151781125030423731000001818.657837

    run2

    Problema:

    Computednodes:2038624

    6571517811Sequential2503042370

    657151781125030423731000001818.823853

    beowulf5

    run1

    Problema:

    Computednodes:2038624

    6571517811Sequential2503042370

    657151781125030423731000001666.956909

    run2

    Problema:

    Computednodes:2038624

    6571517811Sequential2503042370

    657151781125030423731000001662.710083

    beowulf8

    run1

    Problema:

    Computednodes:2038624

    6571517811Sequential2503042370

    657151781125030423731000004337.479004

    run2

    Problema:

    Computednodes:2038624

    6571517811Sequential2503042370

    657151781125030423731000004304.633301

    beowulf16

    run1

    Problema:

    Computednodes:2038624

    6571517811Sequential2503042370

    65715178112503042373100000643.7877197

    run2

    Problema:

    Computednodes:2038624

    6571517811Sequential2503042370

    65715178112503042373100000660.6271973

    1009

    50000

    beowulf0

    run1

    Problema:

    Computednodes:876400

    1640077661Sequential623868781

    164007766162386878150000335.5855103

    run2

    Problema:

    Computednodes:876400

    1640077661Sequential623868781

    164007766162386878150000332.5362549

    beowulf5

    run1

    Problema:

    Computednodes:876400

    1640077661Sequential623868781

    164007766162386878150000323.5028687

    run2

    Problema:

    Computednodes:876400

    1640077661Sequential623868781

    164007766162386878150000323.5307922

    beowulf8

    run1

    Problema:

    Computednodes:876400

    1640077661Sequential623868781

    1640077661623868781500001118.355103

    run2

    Problema:

    Computednodes:876400

    1640077661Sequential623868781

    1640077661623868781500001115.620239

    beowulf16

    run1

    Problema:

    Computednodes:876400

    1640077661Sequential623868781

    164007766162386878150000166.1370087

    run2

    Problema:

    Computednodes:876400

    1640077661Sequential623868781

    164007766162386878150000160.9194336

    1009

    100000

    beowulf0

    run1

    Problema:

    Computednodes:1174148

    6569682646Sequential2503472173

    656968264625034721771000001326.023804

    run2

    Problema:

    Computednodes:1174148

    6569682646Sequential2503472173

    656968264625034721771000001323.400146

    beowulf5

    run1

    Problema:

    Computednodes:1174148

    6569682646Sequential2503472173

    656968264625034721771000001263.724365

    run2

    Problema:

    Computednodes:1174148

    6569682646Sequential2503472173

    656968264625034721771000001263.974854

    beowulf8

    run1

    Problema:

    Computednodes:1174148

    6569682646Sequential2503472173

    656968264625034721771000003235.939941

    run2

    Problema:

    KNP P=89

    Beowulf

    8950000747.886230589100000608.3323364

    800746.7615356800608.5378418

    747.32608.44

    2411.99401.94419.241.811.861.782476.73459.93485.261.281.321.25

    409.292572459.9275818

    401.9393311480.7312622

    418.8582153479.9944763

    419.2424622477.7159119

    410.6402893485.2621765

    3289.08285.54294.762.592.622.543422.25409.10442.851.441.491.37

    288.355957415.1868896

    294.7618408409.1001587

    285.5361328411.6361389

    291.188446442.8543091

    285.5576782432.4755249

    4240.36237.05244.903.113.153.054397.24388.17411.761.531.571.48

    244.896225403.9352722

    239.2999725411.7600098

    237.0492859390.1138306

    238.3262177388.171051

    242.2415466392.2444763

    89500002850.368164891000001546.311768

    5002850.8247075001545.758789

    2850.601546.04

    21537.021499.251589.751.851.901.7921179.971126.761219.841.311.371.27

    1540.5935061219.844482

    1541.4302981140.237183

    1514.0728761208.664917

    1589.7501221126.755371

    1499.2525631204.341675

    31070.521056.151086.732.662.702.6231039.761012.261064.981.491.531.45

    1070.9680181064.982178

    1062.6365971055.480591

    1086.7266851012.25769

    1076.1022951020.051514

    1056.148561046.05127

    4876.46848.87895.303.253.363.184995.12968.701016.251.551.601.52

    872.59869381016.254456

    895.29644781002.895691

    872.1132202980.2436523

    848.8746948968.696228

    893.43988041007.50708

    5773.50762.89793.923.693.743.595931.90919.31942.881.661.681.64

    764.1020508933.7312622

    762.8852539942.8787842

    793.9178467920.7296753

    783.2210693919.3078003

    763.3865967942.8717651

    6673.22648.17701.294.234.404.066924.26906.54962.681.671.711.61

    684.7937622906.5374146

    667.3932495920.9639282

    664.4778442916.819519

    648.1668091962.680542

    701.2907104914.322998

    7617.16602.66625.904.624.734.557904.36883.65915.731.711.751.69

    602.6555786915.7276611

    622.9191284883.6473999

    625.8980713913.6972046

    607.3922729904.8256226

    617.6193237873.262207

    895000089100000

    800-500800-500

    1128.63741.42

    2615.30593.72644.901.831.901.752573.30529.02625.511.291.401.19

    593.7191162616.3682861

    594.3902588529.024292

    644.8969116537.6398926

    638.2175293625.5136108

    605.2785645557.9310303

    4371.41340.69397.653.043.312.844448.76424.78501.501.651.751.48

    377.2372742424.7812805

    340.6868591501.4951477

    347.2186279444.7687378

    397.6471863442.9904785

    394.2655945429.7557983

    6276.04265.71285.304.094.253.966437.59412.52461.441.691.801.61

    279.4191284416.6627197

    285.2971497461.4409485

    282.1201782412.5248108

    265.7080688456.3088074

    267.6360168441.0095215

    8244.70212.90280.484.615.304.028395.61385.10404.691.871.931.83

    280.4764099395.532074

    253.293045396.7980347

    223.8310547404.6850891

    252.9934692385.1011963

    212.8969879395.9093628

    8950000437.108886789100000230.1702271

    1400422.78884891400238.3803101

    429.95234.28

    2237.44232.91240.361.811.851.792176.13168.09183.951.331.391.27

    237.2677612179.1517944

    238.2421417168.0917664

    232.906601170.8267212

    240.3645172178.6315308

    238.4428711183.949234

    3169.12167.72170.452.542.562.523160.91152.08168.481.461.541.39

    170.4525299160.1579742

    168.2289886165.6603546

    169.0936584158.1922913

    170.126297168.4751129

    167.7201233152.0816803

    4136.76132.23140.743.143.253.054148.17143.32153.451.581.631.53

    132.2340851143.9399872

    135.8719635151.9837494

    137.7497101143.3235779

    140.7428436148.1447754

    137.1906891153.44841

    Spine Fast Ethernet

    8950000331.7854004

    1800330.8816833

    331.33

    2187.98185.56190.991.761.791.73

    190.5935516

    186.0215607

    186.7228088

    190.994751

    185.563446

    4108.67108.44109.003.053.063.04

    108.7705612

    108.5903015

    109.0008621

    108.4422684

    108.5354156

    871.5969.0674.624.634.804.44

    74.62387085

    71.24820709

    69.06201172

    71.53001404

    71.51041412

    1351.5349.4854.056.436.706.13

    52.40294266

    51.61725616

    50.12607956

    49.47747421

    54.04982758

    Spine Myrinet

    8950000331.7854004

    1800330.8816833

    331.33

    2181.37178.18185.541.831.861.79

    182.0614929

    179.7945709

    185.5447388

    178.1772156

    181.2948761

    4106.00104.44107.693.133.173.08

    104.4406433

    106.7957153

    105.9263229

    107.6935883

    105.1462021

    879.0274.9383.134.194.423.99

    83.12772369

    80.09178925

    74.92771912

    79.14767456

    77.81520844

    1382.1380.8486.074.034.103.85

    81.23005676

    86.06661224

    80.83776093

    80.97046661

    81.56281281

    Resultados KNP P = 89

    Beowulf

    PrimeNBest SolutionCPrimeNBest SolutionC

    8950,0001,639,893,780623,971,63789100,0006,562,332,5642,499,610,869

    500500

    Seq2850.60Seq1546.04

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifferenceProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    21,537.021,499.251,589.751.851.901.7990.5021,179.971,126.761,219.841.311.371.2793.09

    31,070.521,056.151,086.732.662.702.6230.5831,039.761,012.261,064.981.491.531.4552.72

    4876.46848.87895.303.253.363.1846.424995.12968.701,016.251.551.601.5247.56

    5773.50762.89793.923.693.743.5931.035931.90919.31942.881.661.681.6423.57

    6673.22648.17701.294.234.404.0653.126924.26906.54962.681.671.711.6156.14

    7617.16602.66625.904.624.734.5523.247904.36883.65915.731.711.751.6932.08

    PrimeNBest SolutionCPrimeNBest SolutionC

    8950,0001,639,893,780623,971,63789100,0006,562,332,5642,499,610,869

    800800

    Seq747.32Seq608.44

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifferenceProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    2411.99401.94419.241.811.861.7817.302476.73459.93485.261.281.321.2525.33

    3289.08285.54294.762.592.622.549.233422.25409.10442.851.441.491.3733.75

    4240.36237.05244.903.113.153.057.854397.24388.17411.761.531.571.4823.59

    PrimeNBest SolutionCPrimeNBest SolutionC

    8950,0001,639,893,780623,971,63789100,0006,562,332,5642,499,610,869

    800-500800-500

    Seq1128.63Seq741.42

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifferenceProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    2615.30593.72644.901.831.901.7551.182573.30529.02625.511.291.401.1996.49

    4371.41340.69397.653.043.312.8456.964448.76424.78501.501.651.751.4876.71

    6276.04265.71285.304.094.253.9619.596437.59412.52461.441.691.801.6148.92

    8244.70212.90280.484.615.304.0267.588395.61385.10404.691.871.931.8319.58

    PrimeNBest SolutionCPrimeNBest SolutionC

    8950,0001,639,893,780623,971,63789100,0006,562,332,5642,499,610,869

    14001400

    Seq429.95Seq234.28

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifferenceProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    2237.44232.91240.361.811.851.797.462176.13168.09183.951.331.391.2715.86

    3169.12167.72170.452.542.562.522.733160.91152.08168.481.461.541.3916.39

    4136.76132.23140.743.143.253.058.514148.17143.32153.451.581.631.5310.12

    Spine Fast Ethernet

    PrimeNBest SolutionC

    8950,0001,639,893,780623,971,637

    1800

    Seq331.33

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    2187.98185.56190.991.761.791.735.43

    4108.67108.44109.003.053.063.040.56

    871.5969.0674.624.634.804.445.56

    1351.5349.4854.056.436.706.134.57

    Spine Myrinet

    PrimeNBest SolutionC

    8950,0001,639,893,780623,971,637

    1800

    Seq331.33

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    2181.37178.18185.541.831.861.797.37

    4106.00104.44107.693.133.173.083.25

    879.0274.9383.134.194.423.998.20

    1382.1380.8486.074.034.103.855.23

    &L&D&C&A&R&P

    Resultados KNP P = 89

    1111

    89

    107

    127

    607

    Number of Processors

    Speedup

    Sunfire 6800 (N = 100,000)

    KNP P=709

    11111

    89

    107

    127

    607

    log

    Number of Processors

    Speedup

    Sunfire 6800 (N = 100,000)

    Resultados KNP P =709

    11111

    &A

    Page &P

    89

    107

    127

    607

    log2(Procs)

    KNP Nodos visitados

    11111

    89

    107

    127

    607

    log2(Procs)

    Number of Processors

    Speedup

    Sunfire 6800 (N = 100,000)

    QuickSort 5.000.000

    11111

    89

    107

    127

    607

    log2(Procs)

    Resultados QuickSort 5.000.000

    11111

    &A

    Page &P

    #REF!

    #REF!

    #REF!

    #REF!

    log

    MatMult 512

    00

    00

    00

    00

    00

    00

    N=50.000

    N=100.000

    Nmero de procesadores

    Aceleracin

    Cluster ULL (P=89) 500Mhz

    Resultados MatMult 512

    00

    00

    00

    N=50.000

    N=100.000

    Nmero de procesadores

    Aceleracin

    Cluster ULL (P = 89) 800Mhz

    Posible buen comportamiento

    N=50.000

    N=100.000

    Nmero de procesadores

    Aceleracin

    Cluster ULL (P = 89) 800-500Mhz

    ULL 500 Mhz

    ULL 800 Mhz

    ULL 800-500 Mhz

    ULL 1400 Mhz

    SPINE 1800 Mhz Fast Ethernet

    SPINE 1800 Mhz Myrinet

    number of processors

    speedup

    Knapsack N=50.000

    N=50.000

    N=100.000

    Nmero de procesadores

    Aceleracin

    Cluster ULL (P = 89) 1400Mhz

    N=50.000

    Nmero de procesadores

    Aceleracin

    SPINE (P = 89) 1800Mhz Fast Ethernet

    N=50.000

    Nmero de procesadores

    Aceleracin

    SPINE (P = 89) 1800Mhz Myrinet

    00

    00

    00

    00

    SPINE 1800 Mhz Fast Ethernet

    SPINE 1800 Mhz Myrinet

    Nmero de procesadores

    Aceleracin

    Mochila N=50.000 SPINE

    ULL 500 Mhz

    ULL 800 Mhz

    ULL 800-500 Mhz

    ULL 1400 Mhz

    number of processors

    speedup

    Knapsack N = 50.000 ULL

    Beowulf

    70950000332.92224127091000003260.966064

    800329.81668098003263.137939

    331.373262.05

    2171.87169.11174.601.931.961.9021872.491705.272179.031.741.911.50

    169.11227421790.926514

    170.16540531785.164551

    174.59579471705.271851

    174.33406071902.052002

    171.12605292179.031494

    3144.97142.35150.342.292.332.2031470.911380.841515.752.222.362.15

    143.11094671497.821777

    146.4396211380.835571

    142.63975521515.747559

    142.34849551496.69458

    150.33609011463.456787

    4124.88121.01131.032.652.742.5341279.361186.071406.162.552.752.32

    124.05832671366.247681

    123.88193511201.218872

    121.00601961237.124634

    124.43372341186.074341

    131.02943421406.159302

    709500001129.1341557091000007526.094727

    5001136.8780525007687.610352

    1133.017606.85

    2590.61584.49596.901.921.941.9024445.034429.084460.981.711.721.71

    587.7899784429.076172

    594.007873513654.26172

    584.49133313653.98535

    596.90130624460.981934

    589.874145513654.45996

    3481.88473.50489.732.352.392.3133664.853462.763868.872.082.201.97

    489.72955323616.976318

    482.07522583868.869873

    473.5008853830.020264

    488.35266113462.755127

    475.71966553545.652344

    4407.03400.86417.012.782.832.7243210.572903.833685.992.372.622.06

    417.01123053203.208252

    408.04043582903.826904

    405.64224243685.990967

    403.623233089.941406

    400.85614013169.888916

    5368.34354.26393.913.083.202.8852889.482653.003243.192.632.872.35

    364.62585452664.402588

    368.98199462894.526367

    393.91268922992.306396

    359.92477422652.995117

    354.26342773243.192871

    6332.06325.40340.483.413.483.3362761.192446.862995.772.753.112.54

    331.73098752618.531738

    325.39926152446.857422

    329.98330692829.807617

    332.7186892995.765137

    340.48388672914.98877

    7317.16313.69319.273.573.613.5572355.962294.972454.123.233.313.10

    313.69378662454.123535

    318.51593022294.965576

    319.27484132318.804443

    322.60687262348.990723

    320.82449342218.103271

    70950000709100000

    800-500800-500

    500.284252.29

    2260.13245.17294.501.922.041.7022462.662217.192736.781.731.921.55

    245.17219542217.1875

    252.67143252736.779297

    247.22889712282.607178

    294.50186162501.605225

    261.06838992575.13208

    4193.76144.92254.032.583.451.9741741.981462.542302.852.442.911.85

    254.02952581656.553345

    184.31370542302.854248

    144.92149351593.415771

    217.44776921462.540894

    168.09135441694.527588

    6164.40145.91176.193.043.432.8461324.121210.391517.403.213.512.80

    176.19418331517.403076

    166.39987181323.878906

    145.91043091210.390259

    172.85827641224.63147

    160.62559511344.296631

    8174.93127.09226.622.863.942.2181121.78963.561375.863.794.413.09

    127.0864258963.5563354

    138.80142211074.9021

    226.62348941036.166382

    162.01670841158.427368

    220.10343931375.862793

    70950000165.20999157091000001126.287842

    1400164.454528814001348.139526

    164.831237.21

    295.1494.6096.371.731.741.712691.71622.19748.731.791.991.65

    94.67058563687.8057861

    94.5976181748.732605

    95.04615021724.2920532

    96.37287903675.550415

    95.03154755622.1938477

    375.5574.4577.842.182.212.123585.00555.95607.832.112.232.04

    77.83777618577.7873535

    74.98164368595.0422363

    74.44979095588.3903198

    75.62605286555.9546509

    74.84589386607.8273926

    463.4362.1565.732.602.652.514485.90453.43508.672.552.732.43

    62.91823578485.8868713

    62.41640472453.429718

    63.91430283508.6743469

    62.15296936497.1108093

    65.7294693484.3955078

    Spine Fast Ethernet

    709100000867.3009644

    1800867.6292114

    867.47

    2503.82464.84561.401.721.871.55

    482.4951477

    464.8403015

    493.8860779

    516.4734497

    561.4014893

    4384.48360.33401.722.262.412.16

    391.6777344

    401.7233276

    360.3330383

    369.3500671

    399.3083496

    8231.77210.77255.503.744.123.40

    232.1401215

    255.5014038

    216.0979462

    244.3478851

    210.7703857

    13160.03149.56175.165.425.804.95

    175.1571503

    149.5569

    154.8296661

    160.5673981

    Spine Myrinet

    709100000867.3009644

    1800867.6292114

    867.47

    2546.79487.32632.311.591.781.37

    490.4764099

    487.3229065

    535.6986694

    632.3133545

    588.1174316

    4359.67344.87388.452.412.522.23

    351.6657715

    388.4503784

    347.6478271

    365.7121582

    344.8717041

    8242.60221.52256.973.583.923.38

    245.1198578

    256.9691772

    221.5249786

    255.4715424

    233.9134827

    13179.59170.11187.834.835.104.62

    170.1144409

    187.8299561

    184.5726166

    183.3001099

    172.1497803

    Beowulf

    PrimeNBest SolutionCPrimeNBest SolutionC

    70950,0001,643,145,388623,786,837709100,0006,553,347,9532,501,446,153

    500500

    Seq1133.01Seq7606.85

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifferenceProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    2590.61584.49596.901.921.941.9012.4124,445.034,429.084,460.981.711.721.7131.91

    3481.88473.50489.732.352.392.3116.2333,664.853,462.763,868.872.082.201.97406.11

    4407.03400.86417.012.782.832.7216.1643,210.572,903.833,685.992.372.622.06782.16

    5368.34354.26393.913.083.202.8839.6552,889.482,653.003,243.192.632.872.35590.20

    6332.06325.40340.483.413.483.3315.0862,761.192,446.862,995.772.753.112.54548.91

    7317.16313.69319.273.573.613.555.5872,355.962,294.972,454.123.233.313.10159.16

    PrimeNBest SolutionCPrimeNBest SolutionC

    70950,0001,643,145,388623,786,837709100,0006,553,347,9532,501,446,153

    800800

    Seq331.37Seq3262.05

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifferenceProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    2171.87169.11174.601.931.961.905.4821,872.491,705.272,179.031.741.911.50473.76

    3144.97142.35150.342.292.332.207.9931,470.911,380.841,515.752.222.362.15134.91

    4124.88121.01131.032.652.742.5310.0241,279.361,186.071,406.162.552.752.32220.08

    PrimeNBest SolutionCPrimeNBest SolutionC

    70950,0001,643,145,388623,786,837709100,0006,553,347,9532,501,446,153

    800-500800-500

    Seq500.28Seq4252.29

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifferenceProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    2260.13245.17294.501.922.041.7049.3322,462.662,217.192,736.781.731.921.55519.59

    4193.76144.92254.032.583.451.97109.1141,741.981,462.542,302.852.442.911.85840.31

    6164.40145.91176.193.043.432.8430.2861,324.121,210.391,517.403.213.512.80307.01

    8174.93127.09226.622.863.942.2199.5481,121.78963.561,375.863.794.413.09412.31

    PrimeNBest SolutionCPrimeNBest SolutionC

    70950,0001,643,145,388623,786,837709100,0006,553,347,9532,501,446,153

    14001400

    Seq164.83Seq1237.21

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifferenceProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    295.1494.6096.371.731.741.711.782691.71622.19748.731.791.991.65126.54

    375.5574.4577.842.182.212.123.393585.00555.95607.832.112.232.0451.87

    463.4362.1565.732.602.652.513.584485.90453.43508.672.552.732.4355.24

    Spine Fast Ethernet

    PrimeNBest SolutionC

    709100,0006,553,347,9532,501,446,153

    1800

    Seq867.47

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    2503.82464.84561.401.721.871.5596.56

    4384.48360.33401.722.262.412.1641.39

    8231.77210.77255.503.744.123.4044.73

    13160.03149.56175.165.425.804.9525.60

    Spine Myrinet

    PrimeNBest SolutionC

    709100,0006,553,347,9532,501,446,153

    1800

    Seq867.47

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    2546.79487.32632.311.591.781.37144.99

    4359.67344.87388.452.412.522.2343.58

    8242.60221.52256.973.583.923.3835.44

    13179.59170.11187.834.835.104.6217.72

    &L&D&C&A&R&P

    0000

    89

    107

    127

    607

    Number of Processors

    Speedup

    Sunfire 6800 (N = 100,000)

    00000

    89

    107

    127

    607

    log

    Number of Processors

    Speedup

    Sunfire 6800 (N = 100,000)

    00000

    &A

    Page &P

    89

    107

    127

    607

    log2(Procs)

    00000

    89

    107

    127

    607

    log2(Procs)

    Number of Processors

    Speedup

    Sunfire 6800 (N = 100,000)

    00000

    89

    107

    127

    607

    log2(Procs)

    00000

    &A

    Page &P

    #REF!

    #REF!

    #REF!

    #REF!

    log

    N=50.000

    N=100.000

    Nmero de procesadores

    Aceleracin

    Cluster ULL (P=709) 500Mhz

    N=50.000

    N=100.000

    Nmero de procesadores

    Aceleracin

    Cluster ULL (P = 709) 800Mhz

    N=50.000

    N=100.000

    Nmero de procesadores

    Aceleracin

    Cluster ULL (P = 709) 800-500Mhz

    ULL 500 Mhz

    ULL 800 Mhz

    ULL 800-500 Mhz

    ULL 1400 Mhz

    SPINE 1800 Mhz FE

    SPINE 1800 Mhz Myrinet

    Nmero de procesadores

    Aceleracin

    Mochila N=100.000

    N=50.000

    N=100.000

    Nmero de procesadores

    Aceleracin

    Cluster ULL (P = 709) 1400Mhz

    N=100.000

    Nmero de procesadores

    Aceleracin

    SPINE (P = 709) 1800Mhz Fast Ethernet

    N=100.000

    Nmero de procesadores

    Aceleracin

    SPINE (P = 709) 1800Mhz Myrinet

    ULL 500 Mhz

    ULL 800 Mhz

    ULL 800-500 Mhz

    ULL 1400 Mhz

    Nmero de procesadores

    Aceleracin

    Mochila N=100.000 ULL

    00

    00

    00

    00

    SPINE 1800 Mhz FE

    SPINE 1800 Mhz Myrinet

    Nmero de procesadores

    Aceleracin

    Mochila N=100.000 SPINE

    500 MhzN=50.000P=89

    Seq234567

    2,614,3921,420,559991,572788,774667,828546,031486,730

    Proc.

    01,443,947964,844745,645661,634604,854491,135

    1,283,4631,059,961891,147730,087619,358561,608

    1,365,738911,173846,609758,801571,383576,821

    1,339,2551,081,438829,093745,975573,119500,374

    1,518,2621,059,507857,844715,932593,389569,672

    1,390,1331,015,385834,068722,486592,421539,922

    11,406,406950,935806,703725,968574,789508,703

    1,571,9171,001,623748,919670,419590,034545,997

    1,400,2721,056,073697,564801,267593,183557,636

    1,635,637940,979825,154731,894554,779522,524

    1,240,688940,135768,976706,890606,617544,292

    1,450,984977,949769,463727,288583,880535,830

    21,030,280760,134644,912623,192490,362

    938,194784,315695,162531,868534,216

    1,049,269732,875681,234619,883518,512

    940,726727,988679,011592,267498,488

    948,437868,164599,963581,390459,565

    981,381774,695660,056589,720500,229

    3784,270645,849619,829470,162

    859,221626,981484,109511,687

    816,380630,456588,117463,215

    647,611626,239475,466496,208

    776,871654,360504,989497,902

    776,871636,777534,502487,835

    4593,055476,967401,939

    569,587505,228451,611

    642,886474,233445,678

    583,812463,616420,322

    573,315540,724423,305

    592,531492,154428,571

    5477,032414,635

    510,856389,355

    441,519439,868

    473,538446,815

    514,597455,232

    483,508429,181

    6473,391

    473,513

    514,042

    479,116

    487,663

    485,545

    800-500 MhzN=50.000P=89

    Seq2468

    2,614,3921,357,842793,650565,073416,313

    Proc.

    020971291358960976757810270

    21413521170002991982712161

    239663513114851045240700743

    24058411433558940904709388

    21116281361344894221690370

    2,230,5171,327,070969,821724,586

    1507830300737194989173643

    459406276446216716155999

    490480246066229429160233

    475560291655217171166450

    492555283225190593126188

    485,166279,626209,780156,503

    21402596974545692876

    1275480812833710822

    1068034947964698437

    1442933877456698993

    1200162906583660247

    1,277,841903,876692,275

    3293570191157175100

    262956226421158166

    282150215876150435

    313068186723159191

    298582199201138792

    290,065203,876156,337

    4913268719599

    936523612752

    870202656802

    795911620234

    928962569572

    888,973635,792

    5207749170763

    223038149590

    242524157374

    192830184702

    204431146721

    214,114161,830

    6690206

    665555

    615873

    634429

    633566

    647,926

    7177198

    144860

    150887

    157535

    145809

    155,258

    Nmero de procesadores

    Nmero de nodos

    Nodos medios computados (500 Mhz)

    Nmero de procesadores

    Nmero de nodos

    Nodos computados (500 Mhz)

    0

    0

    0

    0

    0

    Nmero de procesadores

    Nmero de nodos

    Nodos medios computados (800-500 Mhz)

    00000000

    00000000

    00000000

    00000000

    Nmero de procesadores

    Nmero de nodos

    Nodos computados (800-500 Mhz)

    BeowulfRecursiveIterative

    2500000079.4359741269.87485504

    80081.6916351367.76977539

    68.82

    220.7920.6720.963.313.333.28

    20.81490326

    20.74904823

    20.96067238

    20.66607285

    20.76191521

    314.5213.5616.004.745.074.30

    16.00052834

    13.56223774

    13.82237816

    14.28215599

    14.91469193

    414.2913.2615.634.825.194.40

    13.26381779

    14.31311703

    14.83181858

    13.42148113

    15.62960243

    2500000048.0230865544.76244736

    50047.3779258745.34526062

    45.05

    223.4323.1223.941.921.951.88

    23.94370461

    23.1681633

    23.56837273

    23.36433029

    23.12120438

    322.8822.5423.971.972.001.88

    22.58790779

    22.6835537

    22.54350471

    22.60605621

    23.97011757

    425.3321.9128.821.782.061.56

    21.91417313

    23.18014336

    24.45664787

    28.82420731

    28.29302025

    521.9619.9025.072.052.261.80

    21.38262367

    23.38585281

    25.07286072

    20.07789993

    19.90099907

    622.4920.2623.582.002.221.91

    23.01029205

    23.04133415

    20.25942993

    22.56987381

    23.5832634

    721.8120.8022.392.072.172.01

    20.79640961

    22.39462852

    22.22525978

    23.62381363

    23.0675354

    25000000

    800-500

    77.33

    229.5528.9930.762.622.672.51

    30.75984001

    29.48328209

    28.99179077

    29.13553238

    29.4000473

    417.7516.4119.004.364.714.07

    16.4135704

    18.46767616

    18.1777153

    18.99939156

    16.71435738

    616.2715.0018.094.755.164.28

    16.02812576

    18.08687592

    14.99854946

    15.96458912

    16.24725723

    815.2312.9317.185.085.984.50

    14.26074505

    16.44642639

    17.17788887

    12.93121529

    15.325634

    250000004.2671289444.458144188

    14004.2786917694.46160078

    4.46

    23.613.603.621.241.241.23

    3.615757942

    3.613835096

    3.605489969

    3.604753017

    3.605061054

    34.153.784.291.081.181.04

    4.291343212

    4.262543201

    4.209438801

    3.782787085

    4.192961216

    44.514.224.740.991.060.94

    4.74324894

    4.676225185

    4.295331001

    4.218391895

    4.617800236

    Spine Fast Ethernet

    250000003.7125830653.756660938

    18003.7090508943.752368927

    3.75

    26.256.116.310.600.610.60

    6.282154083

    6.111098766

    6.30720377

    6.300533772

    6.270221233

    410.278.2611.480.370.450.33

    11.42951965

    11.47532845

    10.83744907

    9.35370636

    8.258748055

    810.119.5610.550.370.390.36

    9.561342239

    10.08680725

    10.20119762

    10.1518259

    10.54567146

    138.737.5010.230.430.500.37

    7.499276161

    10.2289257

    9.037597656

    8.397551537

    8.49518013

    Beowulf

    PrimeN

    25,000,000

    500

    Seq45.05

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    223.4323.1223.941.921.951.880.82

    322.8822.5423.971.972.001.881.43

    425.3321.9128.821.782.061.566.91

    521.9619.9025.072.052.261.805.17

    622.4920.2623.582.002.221.913.32

    721.8120.8022.392.072.172.011.60

    PrimeN

    25,000,000

    800

    Seq68.82

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    220.7920.6720.963.313.333.280.29

    314.5213.5616.004.745.074.302.44

    414.2913.2615.634.825.194.402.37

    PrimeN

    25,000,000

    800-500

    Seq77.33

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    229.5528.9930.762.622.672.511.77

    417.7516.4119.004.364.714.072.59

    616.2715.0018.094.755.164.283.09

    815.2312.9317.185.085.984.504.25

    PrimeN

    25,000,000

    1400

    Seq4.46

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    23.613.603.621.241.241.230.01

    34.153.784.291.081.181.040.51

    44.514.224.740.991.060.940.52

    Spine Fast Ethernet

    PrimeN

    25,000,000

    1800

    Seq3.75

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    26.256.116.310.600.610.600.20

    410.278.2611.480.370.450.333.22

    810.119.5610.550.370.390.360.98

    138.737.5010.230.430.500.372.73

    &L&D&C&A&R&P

    1111

    89

    107

    127

    607

    Number of Processors

    Speedup

    Sunfire 6800 (N = 100,000)

    11111

    89

    107

    127

    607

    log

    Number of Processors

    Speedup

    Sunfire 6800 (N = 100,000)

    11111

    &A

    Page &P

    89

    107

    127

    607

    log2(Procs)

    11111

    89

    107

    127

    607

    log2(Procs)

    Number of Processors

    Speedup

    Sunfire 6800 (N = 100,000)

    11111

    89

    107

    127

    607

    log2(Procs)

    11111

    &A

    Page &P

    #REF!

    #REF!

    #REF!

    #REF!

    log

    0

    0

    0

    0

    0

    0

    N=5.000.000

    Nmero de procesadores

    Aceleracin

    Cluster ULL 500Mhz

    0

    0

    0

    N=5.000.000

    Nmero de procesadores

    Aceleracin

    Cluster ULL 800Mhz

    0

    0

    0

    0

    N=5.000.000

    Nmero de procesadores

    Aceleracin

    Cluster ULL 800-500Mhz

    1.92265419963.31027349562.61646384761.2357711060.5933279823

    1.96928949214.74100483934.3553491611.07523426270.3612924671

    1.77842019014.81545425674.7541870490.98884152070.3670671254

    2.0512546485.07783628670.4249818854

    2.0030310382

    2.0661756518

    ULL 500 Mhz

    ULL 800 Mhz

    ULL 800-500 Mhz

    ULL 1400 Mhz

    SPINE 1800 Mhz FE

    Nmero de procesadores

    Aceleracin

    QuickSort N=5.000.000

    N=5.000.000

    Nmero de procesadores

    Aceleracin

    Cluster ULL 1400Mhz

    N=5.000.000

    Nmero de procesadores

    Aceleracin

    SPINE 1800Mhz Fast Ethernet

    BeowulfRecursiveIterative

    51255.4240074217.39098549

    80064.7230911317.37613487

    17.38

    210.4810.4510.501.661.661.66

    10.48824596

    10.48006916

    10.45057392

    10.49565411

    10.49302197

    37.187.157.192.422.432.42

    7.187336922

    7.154805183

    7.190048218

    7.157034874

    7.186809063

    45.985.676.152.913.072.83

    6.145885944

    5.669380188

    6.052734852

    6.036333084

    6.000104904

    RecursiveIterative

    51249.8433036837.51736832

    50053.1624870337.65361786

    37.59

    221.3821.2121.451.761.771.75

    21.21115112

    21.45490074

    21.33468437

    21.4450798

    21.43061638

    314.8314.6914.932.532.562.52

    14.75371361

    14.93150711

    14.92503643

    14.85558319

    14.68775177

    411.9211.6112.533.153.243.00

    12.52764606

    11.95148373

    11.85524654

    11.63264561

    11.61279774

    510.4410.2310.583.603.673.55

    10.50299931

    10.58280563

    10.53883076

    10.35225105

    10.23198032

    69.279.159.334.054.114.03

    9.281858444

    9.292833328

    9.151416779

    9.315644264

    9.329179764

    78.418.308.494.474.534.43

    8.429540634

    8.489068031

    8.297622681

    8.241680145

    8.268388748

    512

    800-500

    23.08

    213.5113.5013.521.711.711.71

    13.50867367

    13.51665783

    13.51312542

    13.49929333

    13.50463009

    47.857.708.062.943.002.86

    7.704693794

    8.055457115

    7.925611973

    7.833773136

    7.71668005

    65.995.866.093.853.943.79

    5.859429836

    6.052252769

    5.950535774

    6.08951807

    6.015676975

    85.375.195.664.304.454.07

    5.281681061

    5.373501778

    5.664728165

    5.349915981

    5.185409069

    51217.3439483617.7393074

    140017.3517780317.76284599

    17.75

    211.9011.8911.921.491.491.49

    11.8948307

    11.89482689

    11.91731453

    11.89473915

    11.91619301

    38.168.028.302.182.212.14

    8.183294296

    8.02425766

    8.079161644

    8.212553024

    8.29794693

    46.276.196.342.832.872.80

    6.301105022

    6.239357948

    6.340665817

    6.190915108

    6.30121994

    Spine Fast Ethernet

    51214.4095830914.90150547

    180014.4299259214.86768627

    14.88

    29.349.349.351.591.591.59

    9.340826035

    9.35019207

    9.343358994

    9.340772629

    9.341483116

    45.155.105.212.892.922.86

    5.12613821

    5.130856037

    5.104866982

    5.209394932

    5.201962948

    83.283.103.404.544.794.38

    3.258775949

    3.366005898

    3.104240894

    3.27869606

    3.396981955

    133.563.243.984.184.603.74

    3.239177942

    3.984332085

    3.643779039

    3.567718983

    3.363507032

    Beowulf

    N

    512

    500

    Seq37.59

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    221.3821.2121.451.761.771.750.24

    314.8314.6914.932.532.562.520.24

    411.9211.6112.533.153.243.000.91

    510.4410.2310.583.603.673.550.35

    69.279.159.334.054.114.030.18

    78.418.308.494.474.534.430.19

    N

    512

    800

    Seq17.38

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    210.4810.4510.501.661.661.660.05

    37.187.157.192.422.432.420.04

    45.985.676.152.913.072.830.48

    N

    512

    800-500

    Seq23.08

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    213.5113.5013.521.711.711.710.02

    47.857.708.062.943.002.860.35

    65.995.866.093.853.943.790.23

    85.375.195.664.304.454.070.48

    N

    512

    1400

    Seq17.75

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    211.9011.8911.921.491.491.490.02

    38.168.028.302.182.212.140.27

    46.276.196.342.832.872.800.15

    Spine Fast Ethernet

    N

    512

    1800

    Seq14.88

    ProcsAverageMinimunMaximunSpd-AveSpd-MaxSpd-MinDifference

    29.349.349.351.591.591.590.01

    45.155.105.212.892.922.860.10

    83.283.103.404.544.794.380.29

    133.563.243.984.184.603.740.75

    &L&D&C&A&R&P

    0000

    89

    107

    127

    607

    Number of Processors

    Speedup

    Sunfire 6800 (N = 100,000)

    00000

    89

    107

    127

    607

    log

    Number of Processors

    Speedup

    Sunfire 6800 (N = 100,000)

    00000

    &A

    Page &P

    89

    107

    127

    607

    log2(Procs)

    00000

    89

    107

    127

    607

    log2(Procs)

    Number of Processors

    Speedup

    Sunfire 6800 (N = 100,000)

    00000

    89

    107

    127

    607

    log2(Procs)

    00000

    &A

    Page &P

    #REF!

    #REF!

    #REF!

    #REF!

    log

    0

    0

    0

    0

    0

    0

    N=512

    Nmero de procesadores

    Aceleracin

    Cluster ULL 500Mhz

    0

    0

    0

    N=512

    Nmero de procesadores

    Aceleracin

    Cluster ULL 800Mhz

    0

    0

    0

    0

    N=512

    Nmero de procesadores

    Aceleracin

    Cluster ULL 800-500Mhz

    00001.543321257

    00002.7974298523

    00004.3950068701

    004.0508307686

    0

    0

    ULL 500 Mhz

    ULL 800 Mhz

    ULL 800-500 Mhz

    ULL 1400 Mhz

    SPINE 1800 Mhz FE

    Nmero de procesadores

    Aceleracin

    Multiplicacin de matrices N=512

    N=512

    Nmero de procesadores

    Aceleracin

    Cluster ULL 1400Mhz

    N=512

    Nmero de procesadores

    Aceleracin

    SPINE 1800Mhz Fast Ethernet

    Cursiva -> El comportamiento parece que empeora mucho al aumentar el nmero de procesadores

    NPrimos

    50000891276077098099471009

    1000001276077098099471009

  • Distributed Branch and Bound skeleton

  • Distributed Branch and Bound skeleton

  • Distributed Branch and Bound skeleton

  • Matrix ProductDefinition:Strassen algorithm:Lets bey

  • Distributed Divide and Conquer skeleton

  • Two dimensional cutting stock Problem: User InterfaceIn this problem we are given a large stock rectangle S of dimension LxW and n types of smaller rectangles (pieces) where the i-th type has dimension lixwi. Furthermore, each problem is now to cut off from the large rectangle a set of small rectangles such that:All pieces have a fixed orientation, i.e., a piece of length l and width w is different from a piece of length w and width l (lw)All applied cuts are of guillotine type, i.e., cut that start form one edge and run parallel to the other two edges.There are at most bi rectangles of type i in the cutting plane, the demand constrain of the i-th piece.The overall profit obtained by i=1ncixi where xi denotes the number of rectangles of type i in the cutting patter, is maximized.

  • Performance: CALL & LLACMPI PVMStandard LibrariesParallel ArchitecturesWe need a well accepted Parallel Computing Model

    BSPLogP

    ......

    _964874463.doc

    Communication Network

    Memoria

    Processor

    Memoria

    Processor

    Memoria

    Processor

    ...

    _964881993.doc

    Communication Network

    Memory

    Processor

    Memory

    Processor

    Memory

    Processor

    ...

  • CALL & LLAC Architecture

  • Performance: CALL & LLAC#pragma cll mp mp[0] + mp[1]*N + mp[2]*N*N + mp[3]*N*N*N for (i = 0; i < N; i++) { for (j = 0; j < N; j++) { sum = 0; for (k = 0; k < N; k++) sum += A(i,k) * B(k,j); C(i,j) = sum; } }#pragma cll end mpSquare Matrix Product.A, B y C of dimension NN,

  • Measuring and Predicting Performancewhile(!bqueue.empty()) { auxSp = bqueue.remove(); // pop a problem from the local queue high = auxSp.upper_bound(pbm,auxSol); // upper bound if ( high > bestSol ) { low = auxSp.lower_bound(pbm,auxSol); // lower bound if ( low > bestSol ) { bestSol = low; sol = auxSol; outputPacket.send(MASTER, SOLVE_TAG, bestSol, sol); } if ( high != low ) { // calculate the number of required slaves rSlaves = bqueue.getNumberOfNodes(); op.send(MASTER, BnB_TAG, high, rSlaves); inputPacket.recv(MASTER, nfSlaves, bestSol, rank {1,..., nfSlaves}); if ( nfSlaves >= 0) { auxSp.branch(pbm,bqueue); // branch and save in the local queue for i=0, nfSlaves{ // send subproblems to the assigned slaves auxSp = bqueue.remove();

    outputPacket.send(rank, PBM_TAG, auxSp, bestSol, sol); } } // if nfSlaves == DONE the problem is bounded (cut) } }#pragma cll code numvis++; #pragma cll code numvis++;

  • How to compile?callcckpr. . . . . .

  • Number of visited Nodes Study

  • Measuring and Predicting Performanceint main(int argc, char ** argv) { number sol; readKnap(data);

    /* obj. sig., capacidad rest., beneficio */ sol = knap( 0, M, 0);

    printf("\nsol = ", sol);

    return 0;}#pragma cll code double numvis = 0.0;#pragma cll report all

  • Symbolic Regression ProblemFind the unknown complexity formula starting from the experimental data gathered by CALL.We can use Symbolic Regression: the induction of mathematical expressions on data. Rather than searching for the values of the regression constants, The object of search is a symbolic description of the system.See Scientific Discovery using Genetic Programming by Maarten Keijzer. 2001http://www.cs.vu.nl/mkeijzer/publications/thesis/.Currently we use a fitness function that measures the error of the predictions on the asymptotic side using linear regression on a small sub-sample

  • Prediction of the RNA Secondary Structure ProblemRNA molecule: string of n characters: R=r1 r2 ... rn such that ri {A, C, G, U}Nucleotides join to free energy:A UG UC GThe iteration space is n x n triangularDependences nonuniform: dependences among non-consecutive stages

  • TRACER::ULL 2003 ResultsJournals:Authors: Dorta, Len, RodrguezTitle: Comparing MPI and openMP Implementations of the 0-1 Knapsack Problem Journal: Parallel and Distributed Computing Practices. ISSN 1097-2803 (Accepted)Date: 2003Authors: Blanco V., Garca L., Gonzlez J.A., Rodrguez C., Rodrguez G.Title: A Performance Model for the Analysis of OpenMP ProgramsJournal: Parallel and Distributed Computing Practices. ISSN 1097-2803 (Accepted)Date: 2003

  • TRACER::ULL 2003 ResultsInternational Conferences:Blanco V., Gonzlez J. A., Len C. , Rodrguez C., Rodrguez G. From Complexity Analysis to Performance Analysis. Euro-Par 2003. International Conference on Parallel and Distributed Computing. Klagenfurt, Austria. 26 - 29 August 2003. Dorta I., Len C., Rodrguez C., Rojas A.Parallel Skeletons for Divide and Conquer and Branch and Bound Techniques. 11th euromicro Conference on Parallel and Network-Based Processing. ISSN 1066-6192. Genova, Italy. 5-7 February, 2003. Dorta I., Len C., Rodrguez C. A comparison between MPI and OpenMP Branch-and-Bound Skeletons. 8th International Workshop on High-Level Parallel Programming Models and Supportive Enviroments. ISBN 0-7695-1880-X. Nice, France.22 April, 2003. Dorta I., Len C., Rodrguez C., Rojas A. Parallel Skeletons. Branch-and-Bound and Divide-and-Conquer Techniques. TAM User Group Meeting 2003. Barcelona, Spain. 16 May, 2003Dorta I., Len C., Rodrguez C., Rojas A. MPI and OpenMP implementations of Branch and Bound Skeletons. ParCo2003. Dresden, Germany. 2-5 Septiembre, 2003. Dorta I., Len C., Rodrguez C. Parallel Branch and Bound Skeletons: Message Passing and Shared Memory Implementtions. 5th International Conference on Parallel Processing and Applied Mathematics. Czestochowa, Poland. 7-10 September, 2003. Garca L., Gonzlez J.A., Gonzlez J.C., Len C., Rodrguez C., Rodrguez G. Complexity Driven Performance Analysis. 10th EuroPVM/MPI 2003. Venice, Italy. Sep 29 - Oct 2, 2003.

  • TRACER::ULL 2003 ResultsNational Conferences:Dorta I., Len C., Rodrguez C. Rodrguez, G., Rojas A. Complejidad Algortmica: de la Teora a la Prctica. JENUI03 (Jornadas de Enseanza Universitaria de la Informtica). ISBN 84-283-2845-5. Cdiz. 9-11 Julio, 2003Gonzlez J.R., Len, C., Rodrguez C., UnesqueletoparaRamificacinyAcotacinDistribuido. XIV Jornadas De Paralelismo. Legans (Madrid). 15-17 septiembre 2003PFCGonzlez J. R., Esqueletos Paralelos Distribuidos. Paradigmas de Ramificacin y Acotacin y Divide y Vencers. Documento de Trabajo Interno del DEIOC: DT-03-07. Julio 2003.