Parallel Computations COSC 6374
Popular in Course
Popular in Chemistry
This 13 page Class Notes was uploaded by Lowell Harris on Saturday September 19, 2015. The Class Notes belongs to COSC 6374 at University of Houston taught by Edgar Gabriel in Fall. Since its upload, it has received 58 views. For similar materials see /class/208169/cosc-6374-university-of-houston in Chemistry at University of Houston.
Reviews for Parallel Computations
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 09/19/15
COSC 6374 Parallel Computation Introduction and Organizational Issues Edgar Gabriel Fall 2010 ml Edgar Gabriel i P STL Organizational issues o Classes Monday Wednesday 100pm 230 pm SEC 203 o Evaluation 3 homeworks 25 each 1 quiz at the end of the semester 25 o Tentative date last day of course o TA Sarat Poluri scpoluricsuhedu o In case of questions email gabrielga csuhedu Tel 713 743 3358 PGH 524 Tue 11am12pm or by appointment o All slides available on the website httplwwwcsuhedugabrielcourseslcosc6374 f10 Video streams of some lectures will be offered on the same webpage F IIIC tat 141 amquot Why Parallel Computing To solve larger problems many applications need significantly more memory than a regular PC can provide handle To solve problems faster despite of many advances in computer hardware technology many applications are running slower and slower o eg databases having to handle more and more data o eg large simulations working on even more accurate solutions Pamllelcompumuon 7 swam i PSTL Parallel Programming Exploit concurrency Internet Client and server are independent interacting applications Searching an element distribute the search database onto multiple processors Adding two arrays of integers l Processor1 l I l Processor2 I I Parallel Oompum on v Illl sugarcane CSUH PSTL Parallel Programming II o Scalar product s Zaibi o Parallel algorithm N2 1 N l s ZaibiZaibi Zltalmziiirbmlitigt ZWZMZUWZOWZUD i0 k0 i0 rank1 requires communication between the processes 11 cs UH quottlg if H J0 l Designing energyefficient airplanes Parallel Computation 1 Edgar Gabriel g C S 0 Flow around a space vehicle during reentry Parallel Computation Edgar Gabriel I Simulation of combustion I Processes ill quot 0 YHZOZIPPITI 1o Parallel Computation V m Edgar Gabriel 0 PSTL Determining phylogenetic trees mum my mu mid urn I Panllucnngma nn CSK UH z agarcan lily compute intensive the vings itals 3200 processor c lister a single scene c Vitains V o per second 24 frames per frame 49 x 3112 points with 32 or 64 bjt color I encoding Number of computeradded special fects in movies Jurassic Park 75 39 Lord of the Rings ill 540 each of the followingepisodes of Lord of the Rings doubled the number of special effects last episode of StarWars 20002500 I Iii TOP500 Sublist for November 2004 km and Rpm wlues a m GFlops For more details about other fields please click on the button quotExp i d5quot he lana 39 g non of the F el mmmn ion or VHE FIELD HPC in financial business large databases simulations for risk assessment of portfolios 13 Entries found I Processors Parallel Computation Edgar Gabriel lIii Google39 each access web search means that several hundreds MB of data have to be touched Google s database consists of hundreds of terabytes of data A web search is trivially parallel Google s clusterphilosophy many cheap unreliable nodes Reliability achieved through replication in software and hardware End of 2003 several PC cluster with each 15000 nodes Parallel Computation Edgar Gabriel Figure 1 Google queryeserving architecture Luis Andre Barrosso Jeffrey Dean Urs Hiilzle Web search for a planet The Google Cluster Architecture IEEE Computer Society Parallel Computation 39939 Edgar Gabriel CS 7UH Parallel Computation I Edgar Gabriel CS UH Pam1 r i Srrv3l 5 v General Purpose Graphics Processing Units Nvidia GeForce GTX275 240 cores 896 MB memory 220 Nvidia Tesla S1070 960 4 240 cores 4 TeraFLOPs peak performance P II IC t t 1 a CSUH B STL I Parallel Computer I Parallel Computation m Edgar Gabriel 0 Pall yineim Sm ram Cal a THr39 lae Li EzCll a39 w 7 U Paralle mputer l PC cluster in 2000 4U height per node Myrinet Network Fast Ethernet Network 1U 175 445 cm P II IC t t 1 a CSUH 0 allel Computer I PC Cluster in 2005 m 1U height Infiniband Network Gigabit Ethernet Network Parallel Computation m Edgar Gabriel 0 p a Pi TL I Ir v39quot we T Hr539oui LIaniiw I 7 IBM Roadrunner First computer to surpass the 1 Petaflop 250 FLOPS barrier Installed at Los Alamos National Laboratories Hybrid Architecture 1 7 o 13824 AMD Opteron cores o 116640 IBM PowerXCell 8i cores Costs 120 million Parallel Computation Edgar Gabriel 1O Top 500 List wwwtop500org tmnlpumrVaal Rank sue WMquot Com RH Rosa PWEI Jaguarrcvav HE camuwauunmanmam Otemnsw uvel e 77 7 7 1 Wang W Mm 416 175mm 221 nu Bastien mam NebmaerDawmw Na39wuna sunevcumputmu rcaeuua auamm 2 cemvem smnzheMNScs X565uNVmwaTes a mew 127mm 29mm cm czn nGPuzmu oawmnu Ruedmm new mm M Wad Qi z za 1 x mm away 2 3 Unneds a es anOmevunDC1BGHz 122w mun 137578 23655 Vuuawsm nmandZUUB mm mm msmme my mm my mamast Cumpmanuna W a cwenesUnwavswm fgggmb mm mw 98928 831m mm Tennessee Ovaymc Unwed sums FuvschungszemvumJuehch JUGENErEmeGerwF 5 F11 SummaryInna 256512 BZSSD 1UU27EI ZZBEEIEI Gevmanv mm Me auesrsGmeE mummum mquot m NASAAmes 39eseavch a Camms ggsgggmfgmve mm 772m 97325 mam Unneusmes 2mm sew T m Nur 4 mummyquot NanunawsupevCumnmer 7 cmmmmm gigggfg z mm mm Hume Cm n mbandZUDB NUDT B ueGeneLresewev DOENNSALLNL a WWW woenesmwumum maaz mm 55638 23295 ntvemur meeenelF Avgurme Nahuna Labuvamrv a WWW mummm mm mm mm new SamNauunmanmams RedSkvrSunB adg mm Renewame Enemv 317 mmaw 53 1D Lemmy mm man mm Unwed S a es Rangersuna adexmm Texas Advarmed anpmmu Parallel Born 11 myUnwanan gmgnnggzuggm mm mm 57538 znnnnn dgarGabn39I UN NS E ES Suanvusvstems I 500 LiSt a Prolecled Pertormance Development u l p 1quot 10 PFIUpS a m I Sum 1 quotquot339quot 1 Trend Line 1DUTHDP339 mn Tmnd 8 Llne g VDTFIups Snm Tmnd E Llne o E I TFlups IUD GFlnps 1U GFlnpr I GHops m HH y HH Ln 1 a v 391 m l n g 3 33 3 D D 8 5 N a Equot w Pa EdgarGabn el 11 lIll lmmdumun m Parallel Co m p uti n g 520an Edinon PSTL Accompanying literature I Ananth Grama Anshul Gupta George Karypis Vipin Kumar Introduction to Parallel Computing Pearson Education 2003 John May Parallel IO for High Performance Computing Morgan Kaufmann 2000 Parallel Computation Edgar Gabriel IIll P Accompanying literature II o Accompanying literature Michael J Quinn Parallel Programming in C with MP and OpenMP McGrawHill 2004 Timothy G Mattson Beverly A Sanders Berna L Massingill Patterns for Parallel Programming Software Pattern Series Addison Wessley 2005 Jack Dongarra Ian Foster Geoffrey Fox William Gropp Ken Kennedy Linda Torczon Andy White Sourcebook of Parallel Computing Morgan Kaufmann Publishers 2003 L Ridgeway Scott Terry Clark Babak Bagheri Scientific Parallel Computing Princeton University Press 2005 Parallel Computation Edgar Gabriel 12 Outline of the course 1 Introduction to Parallel Computation Parallel Computer Architecture Parallel Programming wN Parallel O Algorithms Parallel Performance Analysis Performance analysis tools Performance modeling A Pamllelcompumuon Edgar Gabriel PSTL
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'