A Petri Net Approach for. Performance Oriented Parallel Program Design. A. Ferscha. Institut fur Statistik und Informatik, Universitat Wien

Size: px
Start display at page:

Download "A Petri Net Approach for. Performance Oriented Parallel Program Design. A. Ferscha. Institut fur Statistik und Informatik, Universitat Wien"

Transcription

1 A eri Ne Approach for erformance Oriened arallel rogram Design A. Ferscha Insiu fur Saisik und Informaik, Universia Wien Lenaugasse 2/8, A-1080 Vienna, AUSTRIA Tel.: , Fax: A4424DAB@AWIUNI11.BITNET 1

2 roposed running head: erformance Oriened arallel rogram Design A. Ferscha Insiu fur Saisik und Informaik, Universia Wien Lenaugasse 2/8, A-1080 Vienna, AUSTRIA, Tel.: , Fax.: A4424DAB@AWIUNI11.BITNET Absrac erformance orienaion in he developmen process of parallel sofware is moivaed by oulining he misconcepion of curren approaches where performance acivies come in a he very end of he developmen, mainly in erms of measuremen or monioring afer he implemenaion phase. A ha ime major par of he developmen work is already done, and performance pifalls are very hard o repair - if his is possible a all. A developmen process for parallel programs ha launches performance engineering in he early design phase is proposed, based on a eri ne specicaion mehodology for he performance criical pars of a parallel sysem. The eri ne formalism is used o dene rogram Resource Mapping-ne (RM-ne) models, ha serve as an inegraed performance model of parallel processing sysems, combining performance characerisics of parallel programs (-ne), parallel hardware (R-ne) and he assignmen of programs o hardware (Mapping) ino a single performance model, while simulaneously represening he specicaion of a parallel applicaion. redicable parallel algorihm skeleons are worked ou as RM-ne models o simulaneously serve as generic program emplaes o be insaniaed according o specic needs, hus represening he saring poin of he developmen process. The sysemaic inegraion of a se of ools supporing performance oriened parallel programming resuls in he CASE (Compuer Aided arallel Sofware Engineering) environmen, which is being buil around he RM-ne mehodology. Specicaion and performance predicion of parallel applicaions a he algorihm srucure level are demonsraed by example. 2

3 1 Inroducion Alhough he skills in developing parallel algorihms and programs are growing wih he use of parallel and massively parallel hardware [16], here are no general developmen rules or guidelines in order o be conform wih he primary moivaion for using his echnology, i.e. gaining maximum performance from he hardware. I is clear a leas oday, ha peak performance (he heoreical upper limi of machine insrucions execued per ime uni) and performance aainable by a `pracical' applicaion ofen diverge by orders of magniude. Summarizing curren eors o escape from he `parallel sofware'-crises shows, ha alarmingly few aenion is given o he mos criical issue in his concern, namely performance: Compiler Approach Vecorizing compilers have been successfully used for resrucuring innermos loops of programs o be execued on vecor processors [2] arallelizing compilers [46] ry o exrac parallelism a he insrucion and/or loop level, hus making explici parallel programming and he parallelizaion of exising code unnecessary. The compiler approach seems promising in he deecion and exploiaion of ne grain parallelism, bu is no realisic for large grain parallelism [29]. Hence we can expec saisfacory eciency only for a small subclass of parallel archiecures (SIMD). arallel Languages A broad specrum of programming languages [40] for coding parallel applicaions has been developed inuenced by communicaion and synchronizaion mechanisms [3], as well as by programming paradigms as such [5]. A leas imperaive parallel languages are relaively easy o implemen, and reec o some exen he underlying hardware model, so ha he programmer can embody his applicaion domain knowledge o opimally pariion and assign his problem. However, he is ofen concerned wih he burden of correcly managing communicaion and synchronizaion, which is overwhelming his menal capaciy. The funcional, logic and objec oriened approaches ry o relief his burden by providing higher expressiveness of he language, bu considerably complicae implemenaion [44]. Theoreical approaches of parallel programming [9] are sill far from ecien implemenaions. Suppor for parallel programming a a higher level of absracion o ge rid of he underlying hardware is given by language consrucs for manipulaing a shared daa srucure like he Linda uple space [8], bu again ecien implemenaions are lacking. The language approach ends o follow he same developmen as sequenial languages (from hird o forh and fh generaion) by absracing more and more from machine models and hus enlarging expressiveness. No aemps seem o be underaken o suppor any kind of performance analysis a he language level. 3

4 Environmens The needs for oolses assising program developen for parallel machines have rs been addressed in environmenal prooypes like he FAUST [25] or he CODE [7] projec. Recen advances like he SE projec [31] address arge machine independen design and implemenaion of parallel sofware sysems based on large grain daa ow using a graphical and hierarchical consrucion of realisically sized programs. Working projecs deal e.g. wih source code generaion from exual/graphical design, code visualizaion in erms of annoaed graphs like in aragraph [4], concurrency analysis of Ada asking programs in Toal, visualizaion and ineracive parallelizaion of Forran code in a C programs in Aspar, daaow and saic dependency analysis ec. [26]. Wih respec o performance he focus is on visualizaion ools [39], all of which use some combinaion of color, graphics, animaion and even sound o presen performance daa of execuing programs on he y, or raced daa (gahered by sofware or hardware moniors) pos morem. Examples for resource uilizaion visualizaion are he Sysem erformance Displayer [41], or he aragraph environmen [27], which akes execuion prole daa colleced from message passing parallel compuers o animae he behaviour of a parallel program in various modes of moion via several simulaneous views. The TraceView [32] ool has nealy he same aims for machines supporing smaller grained parallelism. The IS-2 insrumenaion sysem [34] suppors a hierarchical view of programs and provides performance daa for each level (program-, machine-, process-, procedure-, insrucion level). From his (seemingly represenaive) examples on developmen environmens we can conclude, ha he parallel program developmen suppor given is dominaed by radiional life cylce models for sequenial sofware. erformance engineering aciviies come ino he developmen process as soon as a running implemenaion of some parallel applicaion is a hand. Of course, a ha ime i is already oo lae for performance invesigaions, because pifalls deeced in he nal implemenaion would cause remendeous reimplemenaion eors, someimes even a redesign of he enire applicaion. This work is o be seen as an aemp in favour of a suppor o he process of developing parallel sofware. We rs propose performance oriened parallel program design and sysemaically describe a parallel sofware developmen cycle (Secion 2). In a second sep a eri ne based design mehodology inegraing funcional specicaion and performance evaluaion echniques (Secion 3) is inroduced. The desing process is concepually demonsraed for he specicaion of a general purpose, reusable applicaion emplae a he algorihm srucure level (Secion 4), and pracically performed in he CASE developmen environmen (Secion 5). 4

5 TOOLSUORT Algorihm Idea High Level Specificaion Implemenaion Skeleon Coding Execuable rogram Assignmen Running Version of Applicaion Hybrid Ediing, Diagramming rove Funcional Validiy erformance redicion Ediing Code Generaion Compiling, Debugging, Tesing Reusabiliy, Library Managemen erformance redicion By Hand lacemen Daa ariioning & Mapping rogram Decomposiion & Mapping Auomaed Mapping Tuning erformance Measuremen Efficien arallel Applicaion Visualizaion User Inerface Figure1: hases of arallel rogram Developmen. 2 erformance Oriened arallel rogram Design Compared o he radiional developmen process of sequenial sofware where performance issues are insucienly considered one is now (naurally) convinced ha performance evaluaion is a criical facor in he upcoming parallel sofware developmen mehodology. The challenges are o suppor he process of developing parallel sofware by eecive (auomaed) performance engineering during he whole sofware developmen cycle. In conras o approaches in he environmens presened above, in a performance oriened parallel program design performance engineering has o approach in he early design phases, accompanying his process unil he compleion of he applicaion. erformance engineering aciviies mus hence range from performance predicion in early developmen sages, performance analysis (boh based on analyical and simulaion modeling) in he deailed design and coding phase, o nally monioring and measuremens in he esing and correcion phase. The following seps are o be aken owards an ecien applicaion in he performance sense (gure 1 shows developmen phases as well as useful and demanded ool suppor.): For expressing he rough algorihm idea of a parallel program he designer for rs should no have o use radiional programming languages and immediaely sar wih an implemenaion, as his is usually ime consuming and errorprone. This work proposes a high level specicaion mehod (based on he eri ne formalism) able o suppor auomaic vericaion of funcional validiy and performance prediciion, while simulaneously exploiing he graphical expressiveness of he framework (op of gure 2). Tool suppor in his developmen phase is given by graphical and hybrid 5

6 (combined graphical and exual) ediing ools consisen wih he specicaion mehod. Afer arriving a an implemenaion skeleon (program emplae) in he specicaion phase, one is now ineresed wheher he prediced performance promises an ecien implemenaion. To his end he specicaion is rened unil a level of absracion is reached, where deailed funcionaliy is specied in erms of program code (middle of gure 2). Code generaors can ranslae from he specicaion ino high level languages (boom of gure 2), he use of several programming language paradigms, faciliies and formalism for language deniions by he user and library mainenance ools become essenial. The reuse of specicaions as well as program fragmens has o be suppored by rerieval and adapion faciliies. erformance predicion in his phase relies on addiional user informaion, hence ineracive ools for he specicaion of he expeced dynamic program behaviour are necessary. On he oher hand he designer may wish o improve he prediced performance by experimenally modifying implemenaions, again by using performance predicion ools. If funcional validiy of he program is assured and performance gures are accepable, he designer can now urn o assign componens of his program o devices of he arge archiecure. Tool suppor can be by eiher oally auomaic (bu generally inecien) process-o-processor or daa-oprocessor mappers, by providing a se of sandard mappings, or by faciliaing o nd an opimum mapping by experimenaion. In he laer case an objec oriened, graphical, direc manipulaive user inerface wih associaed assignemen-code generaors and insrumenaion ools are required. Finally, if a running parallel program has been achieved, performance sudies using monioring echniques can be fruiful in deecing sysem bolenecks and aid in `ne-uning' he applicaion. Furher oolsuppor in his phase are required for visualizing he running sysem on he basis of compuaion and communicaion evens. The resul of his phase is eiher an ecien parallel applicaion, or he nding ha he implemenaion has o be modied or rebuil. The inegraion of he ools menioned o form an environmen is illusraed in gure 8. The major focuss of his work alhough will be on he presenaion of he specicaion echnique and is applicaion in he rs developmen phase. 6

7 3 RM-ne Models as arallel rogram Specicaions 3.1 Formal Specicaion and eri Nes Among he mos frequenly used formalisms for specifying he funcional behaviour of dynamic sysems are auomaa, process algebras and eri nes. rocess algebras (e.g. CS [28] or CCS [35]) like eri nes allow a precise descripion of sysems due o heir formal synax and behavioural semanics, bu also algebraic reasoning, deducion of properies and equaional ransformaion preserving behaviour. Boh approaches have proven highly suiable for parallel and disribued sysem consrucion, as simple composiion operaors (like sequenial, parallel, ieraion and nondeerminisic choice) can be given o incremenally consruc complex sysems ou of simpler ones. Addiionally, eri nes have a simple graphical represenaion hus supporing visualizaion of concurren acions. Specicaion of a parallel program (in he sense of his work) mus also capure performance requiremens eecively, no only he funcional requiremens. Beyond (radiional) requiremens o specicaion mehods (like consisency, formal framework ec.) we claim suppor for funcional and emporal specicaion, suiabiliy for he specicaion of various aspecs of parallel sysems like sofware (conrol ow, daa ow, communicaion, synchronizaion, nondeerminism ec.) and hardware (resources like processing elemens, memory modules, communicaion media ec.), simple bu expressive graphical means, he capabiliy of invesigaions on various levels of absracion, (i.e. provide conceps of modulariy and hierarchical decomposabiliy), for (auomaed) analysis concerning performance, funcional validiy, correcness and expeced behaviour, and suppor for generaing execuable and analysable applicaion prooypes as well as he generaion of high level language source code. I is well known and pracically approved ha eri nes suppor srucured, hierarchical speci- caion of phenomena ypically arising in parallel programming (e.g. asynchronous or synchronous concurren execuion of cooperaing processes), by a he same ime providing a graphical formalism. Furhermore he generaion of he corresponding performance models ou of high level specicaions is a (more or less simple) ranslaion sep, and several mehods and oolsuppor exiss for evaluaing eri ne performance models. On he oher hand - despie superior modeling power - he eri ne formalism is no acively used by sofware engineers. The fundamenal problems o applying formal mehods in general, alhough 7

8 heir analysis capabiliies are commonly recognized, lie in he diculy of heir use [14]. Therefore, for applicaion o pracical work we have o provide means o specify a a more absrac level wih a se of `easy o use' descripive mehods. In he case of eri nes his could be by reducing he generaliy of he framework o an accepable radeo beween expressive power and pracical usabiliy in sofware engineering. In his work we apply a resriced class of hierarchical eri nes (bu sucien wih respec o parallel programming) for specicaion a he highes level, namely he (parallel) algorihm srucure level. A his level, mainly communicaion paerns among independen (sequenial) asks are specied, reecing he kind of parallelism acually exploied for he dedicaed hardware. Obviously a ha level he mos criical performance decisions are being made, and performance predicion is becoming essenial. The eri ne specicaion is paramerized by performance characerisics of he dedicaed hardware and esimaed resource requiremens - analogously o classical performance modeling echniques (e.g. [17]), where a machine model (characerisics of he servicing sysem) is joined wih a workload model (characerisics of service demands) o form a sysem model for which performance indices are evaluaed. (See he illusraion in gure 2.) As soon as here is only a rough esimae of he resource requiremens, only a vague predicion of he performance behaviour will be possible. Neverheless i should be possible o judge wheher some specicaion promises saisfacory performance or no. If so, he applicaion developmen should proceed wih a renemen of he specicaion. (The possibiliy and imporance of reuse of approved algorihm skeleons (parallel program emplaes) has been poined ou in [20].) A he program code level of specicaion, a change from graphical o mulilingual exual represenaion is performed, in order o allow he engineer o use languages he is used o for expressing he funcionaliy and he dynamic behaviour of he program under developmen. Depending on he predicabiliy of he languages used in his renemen sep, he esimaion of he resource requiremen parameers can be improved by auomaic or semiauomaic paramerizaion of he model. The specicaion a he algorihm srucure level and he program code level are consisen in he sense, ha auomaic code generaion can be applied o ranslae he specicaion ino a synacically and semanically correc parallel program in exual form (along wih he specied mapping informaion), o be fed ino a radiional compiler generaing objec code and execuion schedules. erformance predicion based on a (semi-)auomaic generaion of resource requiremen parameers 8

9 Developmen phase erformance predicion Specificaion: algorihm srucure level rough resource requiremen parameers service characerisics of dedicaed hardware Transformaion erformance model Evaluaion program o hardware mapping informaion rough predicion Specificaion: program code level SEQUENTIALLY DO process A process B process C process D improved resource requiremen parameers service characerisics of dedicaed hardware Transformaion erformance model Evaluaion program o hardware mapping informaion improved predicion Code generaion Coding ARALLEL DO SEQUENTIALLY DO on ROCESSOR1 process A process B process C process D SEQUENTIALLY DO on ROCESSOR2 process X process Y process Z auomaically generaed resource requiremen parameers service characerisics of dedicaed hardware Transformaion erformance model program o hardware mapping informaion Evaluaion "precise" predicion Figure2: eri Ne Specicaion and redicion of arallel Applicaions. 9

10 is again improved. 3.2 erformance Modeling and eri Nes eri ne research has signicanly conribued o he performance evaluaion eld in recen years [43], [36], [1], [15], [45], [13], analysis and evaluaion ools have evolved [11]. The simulaneous use of he formalism for analysing qualiaive and quaniave aspecs of sysems is pracically approved and can be seen for example in [6]. The performance of parallel sysems (a parallel program execuing on parallel hardware) is no only deermined by he performance of he hardware iself (e.g. processor-, bus- or link-, memory access-speed ec.), bu also by he srucure of he parallel program (he underlying algorihm, he communicaion paern, synchronizaion of asks ec.) and he assignmen of program pars (asks ha execue concurrenly and cooperaively) o resources. Neiher he approach of resource oriened performance evaluaion of parallel processing sysems, where only he sysem resources are modeled o some exen of deail [38], nor he program or process oriened approach, where exclusively sofware aspecs are subjec o he analysis [45], [23], [10], seem adequae o characerize he performance of parallel sysems. The acual performance of such sysems is always deermined by he inerdependencies beween hardware performance and he requiremens of parallel programs, i.e. he proper uilizaion of hardware performance by he program. Wih RM-nes [19] a modeling echnique has been given considering hardware, sofware and mapping as he performance inuencing facors along wih a compuaionally ecien and accurae mehod for he predicion of performance of parallel compuaions running on parallel hardware. The performance measures of ineres are he (expeced) execuion ime of he program and he degree of resource uilizaion a he hardware level. In he sequel we briey inroduce RM-ne models as a specicaion mehod for parallel programs, focussing mainly he srucure and he resource requiremens of an applicaion and he aspec of he specicaion as far as i helps o consruc a corresponding performance model. The aspec of specifying deailed program funcionaliy is of secondary imporance a he algorihm srucure level. We will denoe funcionaliy a ha level of specicaion by he symbol, which sands for a funcional specicaion in some convenional programming language o be added o he specicaion laer on in he developmen process. 10

11 E rocess 1 fork rocess 2 Loop 1 E1 C 1.1 C 2.1 E2 Loop 1 sub 1 ((2<1>+ <2>);p) E1 Loop 1 A1 n BSR 1 ESR 1 SR 12 BSR 2 ESR 2 n E2 Loop 2 A2 C 1.1 ((2<1>+2<2> +4<3>);p) 1 sub 2 2 ((<3>;p) A1 C 1.2 C 2.2 A2 BSR 1 sub 3 ((<2>+ 3<3>);p) join BSR 1 A (a) (b) Figure3: -ne model of a parallel program 3.3 -nes A eri ne oriened process model is used o specify he srucure, funcionaliy and resource requiremens of parallel programs. A process is graphically represened by a ransiion, where inpuplaces and oupuplaces o he ransiion are used o model he curren sae of he process. A process is ready o ge acive, if is corresponding process ransiion is enabled; he process ges acive as he corresponding ransiion sars ring, and remains acive for he ring duraion. The process behaves according o a predened funcionaliy and erminaes by releasing okens o oupuplaces, herewih making subsequen processes (ransiions) ready o ge acive (enabled). The eri ne specicaion of processes (componenens of parallel programs) is called a -ne or process graph. rocesses can be arranged o be execued in sequence (gure 4 (b)), in parallel (gure 4 (c)), alernaively (gure 4 (d)) or ieraively (gure 4 (e)). Concurren processes are allowed o communicae on a synchronous or asynchronous message passing basis. In gure 3 (a) he -ne of a simple program consiued by wo cyclic processes working in parallel and communicaing wih each oher is given. I is buil 11

12 by a se of process ransiions in a proper arrangemen deermining he dynamic behaviour of he program. Wih every process ransiion in he -ne a mulise of services oered by some physical (hardware) resource is associaed, which we call he resource requiremens of he process. To suppor hierarchical specicaion, process composiions can be folded o form a single, compound process, graphically represened by a single ransiion (box), by aggregaion of he resource requiremens of all he consiuing processes. The opposie is also possible: a single process can be rened by specifying is inernal srucure in erms of complex process composiions. Figure 3 (b) shows ha process C 1:1 is consiued by hree subprocesses sub 1, sub 2 and sub 3, each of hem requiring a cerain amoun of he physical resource services 1 ; 2 and 3. The ype of resource p (processor) is also specied. When aggregaing sub 1, sub 2 and sub 3 o comp 1, he resource requiremens are cumulaed. More formally we have: Deniion 3.1 (-ne) A -ne (or process graph) is a six-uple = ( ; T ; R ; M sar ; R; ) where: (i) (, T ; R ) is he underlying ne wih = fp 1 ; p 2 ; : : : ; p n g and T = f 1 ; 2 ; : : : ; n T g. The elemens i 2 T are called processes. (ii) 9p E 2 wih p E 62 O 8 2 T. p E is he enry place of (iii) 9p A 2 wih p A 62 I 8 2 T. p A is he erminaion place of (iv) 8 2 T : is eiher a primiive or a compound process. (v) The direcion of each r 2 R denes he direcion of he ow of conrol. (vi) M sar = fm 1 ; m 2 ; : : : ; m n g is he iniial marking wih m E = 1 and m i = 0 8p i 2 n p E. (vii) R = f% 1 ; % 2 ; : : : ; % n T g is he se of resource requiremens of f 1 ; 2 : : : ; n T g where % i, he requiremen of process i, is a se of uples (;!) and is a mulise of primiive processes requiring a resource of ype!. (viii) = f 1 ; 2 ; : : : ; n T g is he se of funcionaliies of f 1 ; 2 : : : ; n T g. 12

13 E E E send receive f 1 1 f 2 2 f n T nt A (a) Synchronous Communicaion E A E E E fork A (d) Alernaive Composiion 2 E send receive n T -1 n T E A A (a ) Asynchronous Communicaion... nt join n A n A T+1 (b) Sequenial Composiion A (c) arallel Composiion A (e) Ieraive Composiion Figure4: Graphs of process composiions. We disinguish wo ypes of process ransiions: rimiive processes are deerminisic in behaviour, i.e. hey have deerminisic resource requiremens in ha hey always require he same amoun of services from he physical resources. They are no furher divisible and hence represen he building blocks of a parallel program. The se of all primiive processes wihin a parallel program is denoed by = f 1 ; 2 ; : : : ; n g. The graph of a primiive process is a single ransiion (represened by a bar) wih an enry place and one erminaion place. Complex srucures of parallel programs are represened by process composiions, which are being buil as an arrangemen of primiive or compound processes. A sucien [28] se (o specify any kind of block srucured parallel applicaion) of process composiions is given in erms of process graph composiions in gure 4. For he aggregaion of resource requiremens when folding a process composiion o a single, compound process (represened by a double bordered box) we have he 13

14 following rules: Sequenial A sequenial composiion seq = ( ; T ; R ; M sar ; R; ) of processes T = f 1, 2, : : : n T g wih resource requiremens R = f% 1 ; % 2 ; : : : % n T g and funcionaliies = f 1 ; 2 ; : : : ; n T g is aggregaed o a compound process = ( = fp E ; p A g, T = fg, R = f(p E ; ); (; p A )g, M sar = f1; 0g; R = f%g; = f g), where % = S n! i=1( i ;! i ) denoes he se of all uples of mulises of primiive processes and resource ypes respecively if n! dieren ypes of resources are required by he processes 1 ; 2 ; : : : n T. = n T i=1 i is he (absrac) produc of funcionaliies i. Le % i = Sn! i k=1( k ;! k ) be he resource requiremen of he process i 2 T, assuming ha n! i is he number of dieren ypes of resources required by i. i = f! 1 ;! 2 ; : : : ;! n!i g is he se of all ypes of resources required by i, and = S nt i=1 i he se of all ypes of resources waned by he whole composiion. The resource requiremen (of he compound process) is % = [ jj!j 2 ( X ij(k;!j)2%i k ;! j ); where is a symbol for he sum of mulises. This means ha when aggregaing a se of sequenial processes o a single compound process as a maer of absracion, he resource requiremens of he consiuen processes have o be cumulaed wih respec o dieren ypes of physical resources. arallel The process graph par of a parallel process composiion comprises wo addiional processes: a fork process f and a join process j, such ha he parallel processes ( 1 ; 2 ; : : : ; n T ) are allowed o ge acive concurrenly if he fork process has erminaed. The join process j ges acive as soon as he las 2 f 1 ; 2 ; : : : ; n T g has erminaed. Folding a parallel process composiion o a single compound process is similar o he sequenial case wih respec o cumulaion of resource requiremens, since i is necessary o serve all he requiremens of he (poenially parallel) subprocesses. (We assume ha neiher he fork-, nor he join-process require resources for heir execuion.) Communicaion arallel processes, i.e. processes having a fork-ransiion in common, are allowed o communicae wih each oher. Inerprocess communicaion wih synchronous message passing is expressed by maching send (!) and receive (?) primiives and akes place if boh 14

15 processes issuing hese commands are ready for he message exchange a he same ime. We recognize he operaions! and? o be processes and specify he communicaion of processes by a synchronizaion ransiion (see gure 4 (a)). (Noe ha he inerruped bar for he synchronizaion ransiion is only a drawing convenion and represens a single ransiion in he usual sense.) On he oher hand, asynchronous message passing where he send primiive is nonblocking, a faciliy like a monior or buer is used o deposi he message for collecion a a laer ime. In his case he operaions! and? can again be seen as processes. A deposi place is used o specify asynchronous communicaion (see gure 4 (a 0 )). Alernaive The graph of an alernaive process composiion al has only wo places = fp E ; p A g, allowing a mos one of he alernaive processes ( 1 ; 2 ; : : : ; n T ) o ge acive a he same ime (free choice conic). Boolean guards f i are assigned o he arcs (p E ; i ) for conic resoluion. If none of he guards f 2 ff 1 ; f 2 ; : : : ; f n T g is rue, none of he corresponding processes can ge acive. If more han one of hem are rue simulaneously, one of he corresponding processes is seleced a random o ge acive (nondeerminisic choice). The aggregaion of he resource requiremens in al hence is probabilisic: Le q i = ff (p E;i) = g be he probabiliy ha guard f i is rue ( nt i=1 q i = 1), hen al is aggregaed o a compound process wih resource requiremen: % = [ jj!j2 ( X ij(k;!j)2%i q i k ;! j ): Ieraive The graph of an ieraive process composiion ier conains a 'loop place' p L wih enryand exi-processes E and A (which are again assumed o require no resource services). The arc (p L ; A ) is a couner arc [15] denoing a specic number of loop ieraions (consecuive enablings of he 'loop body' ). (In he example -ne in gure 3 (a) ( Loop1 ; A1 ) is a couner arc labeled by n, denoing ha A1 is enabled if here are n + 1 okens in Loop1. ( Loop1 ; C 1:1 ) is he corresponding couner-alernae arc enabling C 1:1 when he coun in Loop1 is beween 1 and n inclusively. Firing of C 1:1 is allowed whenever a new oken eners Loop1, bu does no remove okens from here, and places okens o subsequen oupuplaces.) Le n be an esimae of he loop ieraions, hen ier aggregaes o a single, compound process 15

16 wih % = n![ k=1 (n k ;! k ): Concise specicaion of replicaed -nes For process composiions wih replicaed srucures -nes can be specied in he avour of r/t-nes [24] o ge a more concise and easier o undersand specicaion of composiions of huge ses of idenical processes. This is of pracical imporance especially for sequenial and parallel processes composiions. We call seq wih T = f 1 ; 2 ; : : : n T g a replicaed sequenial process composiion if all i 2 T are insances of one and he same process (see gure 5 (a)). In his case seq is denoed by he abbreviaion as in gure 5 (c), wih an inerpreaion according he semanics of a r/t-ne as in gure 5 (b). The variable predicae E x iniially holds for all individual processes < i >. 1inT The ransiion selecor x = min i < i > of selecs processes by increasing ideniers o be execued (red) by (removed form E x and deposied ino A x). (In general: an individual < i > is allowed o leave E x only if he individual < j >, j = i? 1 has arrived in A x.) Muliple replicaions are possible and denoed by subsequenial annoaion of replicaors. The same is possible for replicaed parallel process composiions wih noaions as in gure 5 (d)? (f). Noe ha has no ransiion selecor, i.e. can re concurrenly in all modes of x. The range of replicaion is enclosed in brackes (gure 5 (f)). The following relaion among individual okens in muliple (hierarchical) replicaions and plain okens guaranees consisency wih plain -nes: Le u k i k o k be he range of individuals in he k-h replicaion hen X (ukik ok) < i 1 ; i 2 ; : : : ; i k > = < i 1 ; i 2 ; : : : ; i k?1 > holds wih (u 1 i 1 o 1 ) < i 1 > =, i.e. presence of all individuals of single replicaion is equivalen o he presence of he plain oken. Firing Rules -nes have he same ring behaviour as plain lace/transiion nes [42] wih he excepion of he guarded ransiion ring in alernaive process composiions. In his case he conic among enabled ransiions is solved depending on daa in he specicaion - usually performed (ineracively) by an applicaion designer ineresed in he behaviour of he program when simulaing he -ne. For he purpose of invesigaing srucural or performance properies of he -ne i is no 16

17 E E x j 1xN E x j 1xN < i > (i=1..n) E fork < i > 1iN 1iN <x> <x> E E [i=1..n] x = min <i> i <x> <x> (i=1..n) A join [i=1..n] A A A x j 1xN A A x j 1xN (a) (b) (c) (d) (e) (f) Figure5: Replicaed Sequenial/arallel rocess Composiions. longer necessary o consider guarded processes, one is raher ineresed in wha he `real alernaives' are. Hence we can resric our furher consideraions o -nes meeing he following assumpion of deniive choice: whenever he enry-place of an alernaive process composiion is reached, a leas one guard is rue. According o his assumpions we can eliminae guards from -nes by simulaneously assigning appropiae random swiches for conic resoluion. 3.4 roperies of -nes Every -ne is safe in every marking M Sar ; M i reachable from he iniial marking M sar = fm E ; m 2 ; : : : ; m A g = f1; 0; : : : ; 0g, and saises he free choice condiion by deniion. For he invesigaion of behavioural properies of -nes we rs exend -nes as follows: Le = ( ; T ; R, M Sar, R; ) be a -ne. Is exension o = ( ; T ; R ; M sar ), wih: = ; T = T [ e ; R = R [ (p A ; e ) [ ( e ; p E ) is called he exension of. (The exension of he -ne in g. 3 is consruced by adding a ransiion e and arcs from A and o E hus `looping back' he oken ow from A o E.) The exension of a valid -ne is a srongly conneced Free Choice Ne (FC--ne) (I mees he free choice condiion and j p I j 1; j p O j 1; 8 p 2 ). If here are no alernaive or ieraive process composiions in, hen is a srongly conneced Marked Graph (MG--ne) (8 p 2 ; j p I j = j p O j = 1) 17

18 rocess 1 rocess 2 rocess 1 rocess 2 n C 1.1 BSR 1 Loop 1 ESR 1 C 1.2 SR 12 C 2.1 BSR 2 Loop 2 ESR 2 C 2.2 n n C 1.1 BSR 1 Loop 1 ESR 1 C 1.2 SR 12 C 2.1 BSR 2 Loop 2 ESR 2 C 2.2 n o rocessor 1 Link 1-2 (a) o rocessor 2 o rocessor 1 (b) Figure6: RM-nes for (a) wo processor (b) one processor mapping. To deermine wheher a parallel program is free of saic deadlocks we consider a -ne wih M sar = fm E ; m 2 ; : : : ; m A g = f1; 0; : : : ; 0g o erminae if M sop = fm E ; m 2 ; : : : ; m A g = f0; 0; : : : ; 1g is reachable from every marking M Sar ; M i. Now le be he exension of. erminaes if p E as well as p A are covered by all (minimal) place invarians of (for a proof of an equivalen heorem see [18].) Thus, if an invarian does no cover p E as well as p A hen wo undesirable siuaions appear depending on wheher he invarian is iniially marked or no: if i is marked, hen here is a cycle in he process causing livelock, if no, hen (some) subprocesses can never ge acive (dead ransiions). Solving for he minimum place invarians of he exension of he -ne in gure 3 (a) (solve I 0 ~i = ~0) we obain ha he wo of hem boh cover p E as well as p A ; hus will erminae. Moreover in his example he number of invarians (2) obained simulaneously expresses he degree of exploiable parallelism; can (poenially) execue on wo dieren processing elemens. We can conclude ha -nes due o heir deniion allow he applicaion of powerful mehods (and ools) developed for srucural analysis wihin he framework of /T nes. -nes a ha sage represen qualiaively analyzable specicaions of parallel applicaions a he algorihm srucure level. 18

19 3.5 RM-nes Afer qualiaive analysis of he -ne model one is ineresed in he expeced performance of he parallel applicaion under developmen. This requires inclusion of hardware performance characerisics and program o hardware mapping informaion ino he model. arallel processing sysems ypically employ pools of resources like memory, processing elemens and communicaion devices, allowing heir concurren usage. The pool of resources, heir conneciviy and ineraciviy as well as heir poenial performance are in he sequel modeled by R-nes (resource nes). We assume ha for every resource in a parallel processing environmen one can idenify is ype and a se of services (primiive processes) oered o applicaions. Deniion 3.2 Le = f 1 ; 2 ; : : : ; n g be he se of resources and processes ha can be execued by 2. A R-ne is a resource graph R = ( R ; T R ; R R ; M ini ; T ) where: he se of primiive (i) R = fh 1 ; h 2 ; : : : ; h n g is a nie se of 'home' places for he resources 2. (ii) T R = f 1 ; 2 ; : : : ; n T g is a nie se of ransiions and R R ( R T R )[(T R R ) is a ow relaion. (iii) M ini : R 7! iniially assigns resource okens i 2 o 'home' places h i 2 R. (iv) T = f 1 ; 2 ; : : : ; n g is a se of funcions, each i : i 7! Z assigning deerminisic iming informaion o primiive processes execuable by i. Every resource in he sysem is modeled by a oken in he R-ne having is proper home place. resence of he (resource-) oken in ha place indicaes he availabiliy of he resource (idle o serve). Arcs in R R describe he direcion of resource ows and help, ogeher wih ransiions in T R, o model ineracions and inerdependencies beween resources. Wih every resource is associaed a se of primiive processes, along wih iming informaion () for each 2. () is he ime i would ake he resource o serve he primiive process. The assignmen of parallel (sofware) processes o resources is now expressed by a se of arcs combining -nes and R-nes o a single eri ne: Deniion 3.3 A mapping is a se of arcs M ( R T ) [ (T R ) where 19

20 (i) R T is he se of arcs leading from home places o processes such ha if (h i ; j ) 2 R T and he ype of i is!, hen 2 i 8 2 for all uples (;!) 2 % j. (ii) T R is he se of arcs leading from processes o home places wih ( j ; h i ) 2 T R ) 9(h i ; j ) 2 R T as in (i). Assigning home places o process ransiions is allowed only if all he primiive processes required by he ransiion are oered as services by he resource (he resource also has o have he desired ype). We nally call he riple RM = f; R; Mg, he combinaion of a -ne and a R-ne by a mapping o a single, inegraed ne model, RM-ne model. The RM-ne model is a he same ime he specicaion of he parallel applicaion a he algorihm srucure level (see gure 2 and a deailed example in he nex Secion). Figure 6 gives sample mappings of he -ne in gure 3 (a) o a se of resources. Assume ha he compound processes C 1:1, C 1:2, C 2:1 and C 2:2 all require recources of ype p (processor), while rocessor 1 and rocessor 2 being resources of ha ype, hen he process ransiions are allowed o be mapped o hem if hey can serve all he primiive processes required by C 1:1, C 1:2, C 2:1 and C 2:2. Figure 6 (a) shows he assignmen of he parallel program o wo processors conneced o each oher by a communicaion link, fully exploiing he inheren parallelism. During one ieraion sep C 1:1 and C 2:1 can be execued concurrenly by rocessor 1 and rocessor 2. The communicaion process SR 12 synchronizes he wo processes when being execued by a link ype resource. (The link resource is made available only if boh processes are ready for communicaion, i.e. here is a ow oken boh in BSR 1 as well as in BSR 2. This is expressed by bidirecional arcs beween BSR 1 (BSR 2 ) and he ransiion preceding place Link, enabling his ransiion only if BSR 1 and BSR 2 are marked.) Finally C 1:2 and C 2:2 are execued concurrenly. In he second case (gure 6 (b)) he program is mapped - wihou any change in he specicaion of he algorihm srucure - o a single processor (capable of serving also communicaion processes, e.g. emulaed by shared variables). All he processes C 1:1, C 1:2, C 2:1 and C 2:2, as well as SR 12 are execued sequenially. rocesses are scheduled according o heir ow precedence in he -ne: When saring a new loop ieraion only C 1:1 and C 1:2 are ready o ge acive. This is due o C 1:1 and C 1:2 being enabled by okens in he loop places (m Loop1 = 1, m Loop2 = 1), and he residence of he resource oken of rocessor 1 in is home-place (m rocessor1 = 1). The conic among C 1:1 and C 1:2 is resolved as in ordinary eri nes by nondeerminisic selecion of one ransiion (process) o re (execue). 20

21 Assume C 1:1 beeing chosen o execue rs, hen, afer replacing he resource used in is home-place, only C 1:2 is enabled and will re. Afer ha, conrol ow in he -ne forces he communicaion SR 12 o happen (m BSR1 = 1, m BSR2 = 1 and m rocessor1 = 1) ec. To conclude, a ransiion (process) assigned o some resource is enabled (ready o ge acive) if all of is inpuplaces in he -ne hold a (ow-) oken, and he required resource oken is in is home place. The ransiion (process) res (execues) by removing all he ow okens from he inpuplaces in he -ne and he resource oken from he R-ne, making he resource unavailable for oher processes. Afer a cerain ring period ow okens are placed o oupuplaces in he -ne, while he resource oken is placed back in is home place (R-ne), making he resource available again. Timing in RM-nes To suppor independence among he specicaion of he algorihm srucure of an applicaion and he characerisics of resources we inroduce he noion of ineracive iming in he evaluaion of RM-nes. A he ime a parallel applicaion is being developed, he conguraion of he arge hardware is generally no known. The number, ype and arrangemen of processing elemens for example is ofen deermined on he basis of he parallel program so as o achieve opimum performance. To his end a model of he program has o reveal he amoun and ype of services required from hardware resources, independenly of an acual, physical resource consellaion. The -ne provides all hese informaions: he amoun of resources required is impliciely specied by he number of processes and he resource ypes required by hem. The amoun and ype of services is expliciely in he model in erms of resource requiremens associaed o process ransiions. The resource requiremens are expressed in erms of mulises of primiive processes during he paramerizaion of he -ne, and aggregaed in he case of process composiions according o he rules given above. The acual execuion ime of some process is based on performance characerisics of an assigned resource; hese characerisics are expliciely in he R-ne model in he shape of iming funcions for primiive processes, he assignmen informaion is expliciely in he RM-ne model. Given now a process ransiion i wih resource requiremens % i = (;!), where = n kjk2i k k is a mulise of primiive processes ou of i (n k denoes he mulipliciy of k in j ), assigned o a resource j wih services j and service imes j (), 2 j ( i j ), hen he (deerminisic) ring 21

22 ime for ha ransiion is ineracively (a he ring insan) calculaed as X kjk2i n k j ( k ): The compound process C 1:1 in gure 6 (a) for example, wih resource requiremens as in gure 3 (b), assigned o resource rocessor 1 wih execuion imes 1 ( 1 ), 1 ( 2 ) and 1 ( 3 ) for he primiive processes 1, 2 and 3 would ake 2 1 ( 1 )+2 1 ( 2 )+4 1 ( 3 ) ime seps o execue. In he RM-ne (gure 6 (a)) his would cause (afer ring of f ork - which is no assigned o a resource since i does no require services - in zero ime) a removal of he resource oken from place rocessor 1 in he R-ne and he ow oken from place Loop1 in he -ne a ime zero, and a release of boh okens (he resource oken back o place rocessor 1, and he ow oken o BSR 1 ) a ime 2 1 ( 1 )+2 1 ( 2 )+4 1 ( 3 ). During ha ime period, neiher he specic ow oken, nor he resource oken would be available (visible) in he RM-ne (noe ha his ring policy is in conras o he usual `race policy'). Based on he above iming convenions, he expeced overall execuion ime of he parallel program in he specicaion phase is evaluaed by simulaion of he RM-ne model. Tools [12] have proven useful for his ask, alhough preprocessing is necessary o simulae ineracive iming. Wih he simulaion of he RM-ne one a he same ime observes oken disribuions of home-places in he R-ne, represening he basis for he derivaion of gures describing resource usage. A his poin i is easy seen ha RM-nes serve as he basic model for performance predicion, while a he same ime acing as he specicaion of he parallel applicaion a he algorihm srucure level. Moreover RM-ne models bear he poenial of reusing hem as general purpose applicaion emplaes, insanciaed by a assigning a specic funcionaliy for a special purpose. We will demonsrae his aspec of RM-ne based specicaion in he nex secion, by demonsraing he developmen seps from he algorihm idea downo an implemenaion skeleon (gure 2), and he former aspec (performance predicion) in erms of anoher example in he CASE environmen aferwards. 4 Specicaion of a rocess ipeline The principle common o pipelined parallel algorihms [37] is ha daa is owing hrough a cascade of processes (pipeline sages) being modied by process aciviies. rocesses acing as ransformers of daa usually have dieren funcionaliy. Applying his funcionaliy in a predened sequence 22

23 upon he inpu daa sream generaes an oupu daa sream represening he problem soluion. The behaviour of a pipeline sage is o accep daa from a predecessor sage, apply a se of operaions on ha daa and pass he resul o he successor sage. Consequenly, as he sages have dieren funcionaliy, he compuaions in a pipeline are asynchronous. Communicaion in a pipeline is always beween consecuive sages and usually a daa iem is allowed o ener a sage if he predecessor has already released i, i.e. communicaion is regular. As he arrival and deparure of daa drives he compuaions in a pipeline, he regular communicaions naurally synchronise he asynchronous compuaions. For he developmen of a pipelined parallel applicaion we sar according o he scheme in gure 1 wih an algorihm idea, which is has verbally been described above. Formally we can hink of a pipeline sage as a box wih some hidden funcionaliy as a he op lef corner of gure 7, and he pipeline iself as a linear arrangemen of sages. A -ne specicaion for a single sage is easily derived: Firs a daa iem is acceped from he previous sage, which is specied as a receive process wih resource requiremen of f(h?i; l)g (where l denoes a link ype resource). Afer daa is received, some operaion wih funcionaliy has o be performed, which is expressed by a process ransiion wih resource requiremens o a processor f(: : : ; p)g (a mulise of primiive processes describing he resource requiremens o he processor is inenionally lef open here). Finally daa is passed by o a succeeding sage, specied by a send communicaion process wih resource requiremen f(h!i; l)g. This has o be repeaed ieraively for a given se of daa, being specied by an ieraive process composiion ha will erminae afer a cerain number (n) of ieraions. A -ne for he behaviour of a single pipeline sage is developed in he upper hird of gure 7 Assuming synchronous message passing communicaion, he pipeline sages have o be ied o each oher such ha he oupu (!) of some sage i forms a single (synchronizaion) ransiion wih he inpu (?) of sage i + 1. In he middle of gure 7 a hree sage pipeline is given, where he ieraive processes represening single sages are forced operae in parallel by being arranged as a parallel process composiion. The -ne derived can now be analyzed from a correcness and erminaion poin of view by applying srucural analysis upon he corresponding eri ne. Qualiaive analysis of he -ne reveals ha a process in sage i can ge ready for execuion in ieraion j only if sage i? 1 has already compleed ieraion j and passed he resuling daa o sage i. Furhermore, all he sages can poenially execue heir processes in parallel, sage i in ieraion j, sage i?1 in ieraion 23

24 Behaviour of Single ipeline Sage Algorihmic Idea E E n A A {(<?>, l)} {(..., p)} {(<!>, l)} E f Three Sage ipeline (-ne) change mapping {(..., p)} n {(<!>, l)} n {(<?>, l)} {(<!>, l)} {(..., p)} j A n {(<?>, l)} {(..., p)} Three Sage ipeline Mapped on Three rocessors processor link processor link processor {(..., p)} processor 1 n {(<!>, l)} n {(<?>, l)} {(<!>, l)} {(..., p)} n {(<?>, l)} link 1-2 link 2-3 processor 2 {(..., p)} processor 3 erformance redicion Implemenaion Skeleon Figure7: Specicaion for a rocess ipeline. 24

25 j + 1, sage i + 1 in ieraion j? 1 ec. The maximum degree of parallelism is deermined by he maximum overlap of sage compuaions, which can be veried by applying invarian analysis. For performance predicion we have o provide addiional informaion o he -ne model. Firs by giving hardware o mapping informaion, and secondly by providing service characerisics of he hardware in use. This is achieved by exending he -ne o a RM-ne. In he case of he availabiliy of a hree processor pipeline, one possible mapping resuls from assigning a single pipeline sage o a single processor, and communicaions among sages o communicaion links connecing processors (lower hird of gure 7). The RM-ne can be evaluaed under is paramerizaion (se of all resource requiremens in he -ne) by simulaing he corresponding imed eri ne. The eciency of pipelined algorihms is deermined by he homogeniy of sages, i.e. he balance of operaions in he sages and he grain size of sage compuaions, i.e. he communicaion frequency beween sages. If a p-ype resource is able also o serve a l-ype requiremen (emulaion of message exchange), hen communicaion could also be mapped o a processor. Wihou changing he - ne, several dieren mappings could be invesigaed o balance he operaions across processors by variaions of he R-ne and he mapping arcs yielding dieren paramerizaions of he RM-ne. (racical case sudies wih a real applicaion on a real muliprocessor sysem are repored in [19] [22].) As soon as a specicaion for which he prediced performance is saisfacory (or opimum) is found, he specicaion is kep as an applicaion or program emplae, and he designer can sar wih he implemenaion by sepwise renemen of he specicaion, i.e. providing he funcionaliy of sages in erms of a high level programming language. rogram emplaes a ha high level of absracion can be insaniaed in fuure applicaions by simply assigning dieren funcionaliy o sages, hus serving as general purpose, reusable, performance opimized implemenaion skeleons which can considerably ease applicaion design and save developmen ime. 5 Applicaion Specicaion in he CASE Environmen 5.1 The CASE Environmen In conras o he echnology of compuer aided sofware engineering (CASE) which provides se of ools for sofware developmen, according o radiional mehodologies like he waerfall-model, 25

26 GRAHICAL, OBJECT ORIENTED USER INTERFACE X-Windows DESIGN Hierarchy- Diagram Edior CODING Texual/Graphical Edior Compiler Debugger ERFORMANCE REDICT./MEASUREM. Model Generaor Model Edior Evaluaor Analysis/Simulaion EXECUTION VISUALIZATION View Specifier Even Tracer ASSIGNMENT rocess/daa lacer rocess/daa Mapper Library Manager Sofware Monior erformance Viewer Execuion Viewer Mapping Library Manager SECIFICATION ERFORMANCE MODELS ERFORMANCE DATA MAING BASE LIBRARIES CODE EXECUTABLES VIEW SETTINGS TRACE INFORMATION Figure8: Archiecure of he CASE Environmen he spiral- model or rapid prooyping, he Compuer Aided arallel Sofware Engineering (CASE) environmen [21] aims o assis he proposed performance oriened developmen cycle (gure 1) in all developmen phases by inegraion of auomaed performance ools and he CASE mehodology. I inegraes he RM-ne based specicaion mehod for parallel programs wih eri ne relaed srucural and performance analysis mehods. The inerdependencies and henceforh conneciviy of ools in he compound oolse is given in gure 8. All he ool funcionaliy is hidden behind a unique, graphical, window-oriened user inerface. The design ools follow he need for graphical program developmen a he algorihm srucure level by applying he RM-ne meaphor, and are used o work ou he specicaion which is used by coding ools o generae or develop program code, and by he performance ools o derive performance models. In his secion we give a avour of he developmen process from he algorihm idea o he implemenaion skeleon of he applicaion in he CASE environmen (he rs sep in gure 1 applying ools for semigraphical ediing and performace predicion). As in his phase he major performance decisions have o be made, and as he proposed mehod applies mainly in his sep (see upper hird of gure 2), we concenrae only on his subse of CASE (see [21] for a sysemaic descripion). 26

Implementing Ray Casting in Tetrahedral Meshes with Programmable Graphics Hardware (Technical Report)

Implementing Ray Casting in Tetrahedral Meshes with Programmable Graphics Hardware (Technical Report) Implemening Ray Casing in Terahedral Meshes wih Programmable Graphics Hardware (Technical Repor) Marin Kraus, Thomas Erl March 28, 2002 1 Inroducion Alhough cell-projecion, e.g., [3, 2], and resampling,

More information

Scheduling. Scheduling. EDA421/DIT171 - Parallel and Distributed Real-Time Systems, Chalmers/GU, 2011/2012 Lecture #4 Updated March 16, 2012

Scheduling. Scheduling. EDA421/DIT171 - Parallel and Distributed Real-Time Systems, Chalmers/GU, 2011/2012 Lecture #4 Updated March 16, 2012 EDA421/DIT171 - Parallel and Disribued Real-Time Sysems, Chalmers/GU, 2011/2012 Lecure #4 Updaed March 16, 2012 Aemps o mee applicaion consrains should be done in a proacive way hrough scheduling. Schedule

More information

Learning in Games via Opponent Strategy Estimation and Policy Search

Learning in Games via Opponent Strategy Estimation and Policy Search Learning in Games via Opponen Sraegy Esimaion and Policy Search Yavar Naddaf Deparmen of Compuer Science Universiy of Briish Columbia Vancouver, BC yavar@naddaf.name Nando de Freias (Supervisor) Deparmen

More information

Simple Network Management Based on PHP and SNMP

Simple Network Management Based on PHP and SNMP Simple Nework Managemen Based on PHP and SNMP Krasimir Trichkov, Elisavea Trichkova bsrac: This paper aims o presen simple mehod for nework managemen based on SNMP - managemen of Cisco rouer. The paper

More information

A Matching Algorithm for Content-Based Image Retrieval

A Matching Algorithm for Content-Based Image Retrieval A Maching Algorihm for Conen-Based Image Rerieval Sue J. Cho Deparmen of Compuer Science Seoul Naional Universiy Seoul, Korea Absrac Conen-based image rerieval sysem rerieves an image from a daabase using

More information

PART 1 REFERENCE INFORMATION CONTROL DATA 6400 SYSTEMS CENTRAL PROCESSOR MONITOR

PART 1 REFERENCE INFORMATION CONTROL DATA 6400 SYSTEMS CENTRAL PROCESSOR MONITOR . ~ PART 1 c 0 \,).,,.,, REFERENCE NFORMATON CONTROL DATA 6400 SYSTEMS CENTRAL PROCESSOR MONTOR n CONTROL DATA 6400 Compuer Sysems, sysem funcions are normally handled by he Monior locaed in a Peripheral

More information

CAMERA CALIBRATION BY REGISTRATION STEREO RECONSTRUCTION TO 3D MODEL

CAMERA CALIBRATION BY REGISTRATION STEREO RECONSTRUCTION TO 3D MODEL CAMERA CALIBRATION BY REGISTRATION STEREO RECONSTRUCTION TO 3D MODEL Klečka Jan Docoral Degree Programme (1), FEEC BUT E-mail: xkleck01@sud.feec.vubr.cz Supervised by: Horák Karel E-mail: horak@feec.vubr.cz

More information

Quick Verification of Concurrent Programs by Iteratively Relaxed Scheduling

Quick Verification of Concurrent Programs by Iteratively Relaxed Scheduling Quick Verificaion of Concurren Programs by Ieraively Relaxed Scheduling Parick Mezler, Habib Saissi, Péer Bokor, Neeraj Suri Technische Univerisä Darmsad, Germany {mezler, saissi, pbokor, suri}@deeds.informaik.u-darmsad.de

More information

STEREO PLANE MATCHING TECHNIQUE

STEREO PLANE MATCHING TECHNIQUE STEREO PLANE MATCHING TECHNIQUE Commission III KEY WORDS: Sereo Maching, Surface Modeling, Projecive Transformaion, Homography ABSTRACT: This paper presens a new ype of sereo maching algorihm called Sereo

More information

Rule-Based Multi-Query Optimization

Rule-Based Multi-Query Optimization Rule-Based Muli-Query Opimizaion Mingsheng Hong Dep. of Compuer cience Cornell Universiy mshong@cs.cornell.edu Johannes Gehrke Dep. of Compuer cience Cornell Universiy johannes@cs.cornell.edu Mirek Riedewald

More information

Analysis of Various Types of Bugs in the Object Oriented Java Script Language Coding

Analysis of Various Types of Bugs in the Object Oriented Java Script Language Coding Indian Journal of Science and Technology, Vol 8(21), DOI: 10.17485/ijs/2015/v8i21/69958, Sepember 2015 ISSN (Prin) : 0974-6846 ISSN (Online) : 0974-5645 Analysis of Various Types of Bugs in he Objec Oriened

More information

Automatic Calculation of Coverage Profiles for Coverage-based Testing

Automatic Calculation of Coverage Profiles for Coverage-based Testing Auomaic Calculaion of Coverage Profiles for Coverage-based Tesing Raimund Kirner 1 and Waler Haas 1 Vienna Universiy of Technology, Insiue of Compuer Engineering, Vienna, Ausria, raimund@vmars.uwien.ac.a

More information

Let s get physical - EDA Tools for Mobility

Let s get physical - EDA Tools for Mobility Le s ge physical - EDA Tools for Mobiliy Aging and Reliabiliy Communicaion Mobile and Green Mobiliy - Smar and Safe Frank Oppenheimer OFFIS Insiue for Informaion Technology OFFIS a a glance Applicaion-oriened

More information

A time-space consistency solution for hardware-in-the-loop simulation system

A time-space consistency solution for hardware-in-the-loop simulation system Inernaional Conference on Advanced Elecronic Science and Technology (AEST 206) A ime-space consisency soluion for hardware-in-he-loop simulaion sysem Zexin Jiang a Elecric Power Research Insiue of Guangdong

More information

source managemen, naming, proecion, and service provisions. This paper concenraes on he basic processor scheduling aspecs of resource managemen. 2 The

source managemen, naming, proecion, and service provisions. This paper concenraes on he basic processor scheduling aspecs of resource managemen. 2 The Virual Compuers A New Paradigm for Disribued Operaing Sysems Banu Ozden y Aaron J. Goldberg Avi Silberschaz z 600 Mounain Ave. AT&T Bell Laboraories Murray Hill, NJ 07974 Absrac The virual compuers (VC)

More information

Discrete Event Systems. Lecture 14: Discrete Control. Continuous System. Discrete Event System. Discrete Control Systems.

Discrete Event Systems. Lecture 14: Discrete Control. Continuous System. Discrete Event System. Discrete Control Systems. Lecure 14: Discree Conrol Discree Even Sysems [Chaper: Sequenial Conrol + These Slides] Discree Even Sysems Sae Machine-Based Formalisms Saechars Grafce Laboraory 2 Peri Nes Implemenaion No covered in

More information

Assignment 2. Due Monday Feb. 12, 10:00pm.

Assignment 2. Due Monday Feb. 12, 10:00pm. Faculy of rs and Science Universiy of Torono CSC 358 - Inroducion o Compuer Neworks, Winer 218, LEC11 ssignmen 2 Due Monday Feb. 12, 1:pm. 1 Quesion 1 (2 Poins): Go-ack n RQ In his quesion, we review how

More information

CENG 477 Introduction to Computer Graphics. Modeling Transformations

CENG 477 Introduction to Computer Graphics. Modeling Transformations CENG 477 Inroducion o Compuer Graphics Modeling Transformaions Modeling Transformaions Model coordinaes o World coordinaes: Model coordinaes: All shapes wih heir local coordinaes and sies. world World

More information

Service Oriented Solution Modeling and Variation Propagation Analysis based on Architectural Building Blocks

Service Oriented Solution Modeling and Variation Propagation Analysis based on Architectural Building Blocks Carnegie Mellon Universiy From he SelecedWorks of Jia Zhang Ocober, 203 Service Oriened Soluion Modeling and Variaion Propagaion Analysis based on Archiecural uilding locks Liang-Jie Zhang Jia Zhang Available

More information

IDEF3 Process Description Capture Method

IDEF3 Process Description Capture Method IDEF3 Process Descripion Capure Mehod IDEF3 is par of he IDEF family of mehods developmen funded by he US Air Force o provide modelling suppor for sysems engineering and enerprise inegraion 2 IDEF3 Mehod

More information

Outline. EECS Components and Design Techniques for Digital Systems. Lec 06 Using FSMs Review: Typical Controller: state

Outline. EECS Components and Design Techniques for Digital Systems. Lec 06 Using FSMs Review: Typical Controller: state Ouline EECS 5 - Componens and Design Techniques for Digial Sysems Lec 6 Using FSMs 9-3-7 Review FSMs Mapping o FPGAs Typical uses of FSMs Synchronous Seq. Circuis safe composiion Timing FSMs in verilog

More information

Lecture 18: Mix net Voting Systems

Lecture 18: Mix net Voting Systems 6.897: Advanced Topics in Crypography Apr 9, 2004 Lecure 18: Mix ne Voing Sysems Scribed by: Yael Tauman Kalai 1 Inroducion In he previous lecure, we defined he noion of an elecronic voing sysem, and specified

More information

FIELD PROGRAMMABLE GATE ARRAY (FPGA) AS A NEW APPROACH TO IMPLEMENT THE CHAOTIC GENERATORS

FIELD PROGRAMMABLE GATE ARRAY (FPGA) AS A NEW APPROACH TO IMPLEMENT THE CHAOTIC GENERATORS FIELD PROGRAMMABLE GATE ARRAY (FPGA) AS A NEW APPROACH TO IMPLEMENT THE CHAOTIC GENERATORS Mohammed A. Aseeri and M. I. Sobhy Deparmen of Elecronics, The Universiy of Ken a Canerbury Canerbury, Ken, CT2

More information

Real Time Integral-Based Structural Health Monitoring

Real Time Integral-Based Structural Health Monitoring Real Time Inegral-Based Srucural Healh Monioring The nd Inernaional Conference on Sensing Technology ICST 7 J. G. Chase, I. Singh-Leve, C. E. Hann, X. Chen Deparmen of Mechanical Engineering, Universiy

More information

4. Minimax and planning problems

4. Minimax and planning problems CS/ECE/ISyE 524 Inroducion o Opimizaion Spring 2017 18 4. Minima and planning problems ˆ Opimizing piecewise linear funcions ˆ Minima problems ˆ Eample: Chebyshev cener ˆ Muli-period planning problems

More information

Petri Nets for Object-Oriented Modeling

Petri Nets for Object-Oriented Modeling Peri Nes for Objec-Oriened Modeling Sefan Wi Absrac Ensuring he correcness of concurren rograms is difficul since common aroaches for rogram design do no rovide aroriae mehods This aer gives a brief inroducion

More information

An Improved Square-Root Nyquist Shaping Filter

An Improved Square-Root Nyquist Shaping Filter An Improved Square-Roo Nyquis Shaping Filer fred harris San Diego Sae Universiy fred.harris@sdsu.edu Sridhar Seshagiri San Diego Sae Universiy Seshigar.@engineering.sdsu.edu Chris Dick Xilinx Corp. chris.dick@xilinx.com

More information

A GRAPHICS PROCESSING UNIT IMPLEMENTATION OF THE PARTICLE FILTER

A GRAPHICS PROCESSING UNIT IMPLEMENTATION OF THE PARTICLE FILTER A GRAPHICS PROCESSING UNIT IMPLEMENTATION OF THE PARTICLE FILTER ABSTRACT Modern graphics cards for compuers, and especially heir graphics processing unis (GPUs), are designed for fas rendering of graphics.

More information

Voltair Version 2.5 Release Notes (January, 2018)

Voltair Version 2.5 Release Notes (January, 2018) Volair Version 2.5 Release Noes (January, 2018) Inroducion 25-Seven s new Firmware Updae 2.5 for he Volair processor is par of our coninuing effors o improve Volair wih new feaures and capabiliies. For

More information

Computer representations of piecewise

Computer representations of piecewise Edior: Gabriel Taubin Inroducion o Geomeric Processing hrough Opimizaion Gabriel Taubin Brown Universiy Compuer represenaions o piecewise smooh suraces have become vial echnologies in areas ranging rom

More information

Web System for the Remote Control and Execution of an IEC Application

Web System for the Remote Control and Execution of an IEC Application Web Sysem for he Remoe Conrol and Execuion of an IEC 61499 Applicaion Oana ROHAT, Dan POPESCU Faculy of Auomaion and Compuer Science, Poliehnica Universiy, Splaiul Independenței 313, Bucureși, 060042,

More information

The Impact of Product Development on the Lifecycle of Defects

The Impact of Product Development on the Lifecycle of Defects The Impac of Produc Developmen on he Lifecycle of Rudolf Ramler Sofware Compeence Cener Hagenberg Sofware Park 21 A-4232 Hagenberg, Ausria +43 7236 3343 872 rudolf.ramler@scch.a ABSTRACT This paper invesigaes

More information

Handling uncertainty in semantic information retrieval process

Handling uncertainty in semantic information retrieval process Handling uncerainy in semanic informaion rerieval process Chkiwa Mounira 1, Jedidi Anis 1 and Faiez Gargouri 1 1 Mulimedia, InfoRmaion sysems and Advanced Compuing Laboraory Sfax Universiy, Tunisia m.chkiwa@gmail.com,

More information

Video Content Description Using Fuzzy Spatio-Temporal Relations

Video Content Description Using Fuzzy Spatio-Temporal Relations Proceedings of he 4s Hawaii Inernaional Conference on Sysem Sciences - 008 Video Conen Descripion Using Fuzzy Spaio-Temporal Relaions rchana M. Rajurkar *, R.C. Joshi and Sananu Chaudhary 3 Dep of Compuer

More information

An Adaptive Spatial Depth Filter for 3D Rendering IP

An Adaptive Spatial Depth Filter for 3D Rendering IP JOURNAL OF SEMICONDUCTOR TECHNOLOGY AND SCIENCE, VOL.3, NO. 4, DECEMBER, 23 175 An Adapive Spaial Deph Filer for 3D Rendering IP Chang-Hyo Yu and Lee-Sup Kim Absrac In his paper, we presen a new mehod

More information

Michiel Helder and Marielle C.T.A Geurts. Hoofdkantoor PTT Post / Dutch Postal Services Headquarters

Michiel Helder and Marielle C.T.A Geurts. Hoofdkantoor PTT Post / Dutch Postal Services Headquarters SHORT TERM PREDICTIONS A MONITORING SYSTEM by Michiel Helder and Marielle C.T.A Geurs Hoofdkanoor PTT Pos / Duch Posal Services Headquarers Keywords macro ime series shor erm predicions ARIMA-models faciliy

More information

COSC 3213: Computer Networks I Chapter 6 Handout # 7

COSC 3213: Computer Networks I Chapter 6 Handout # 7 COSC 3213: Compuer Neworks I Chaper 6 Handou # 7 Insrucor: Dr. Marvin Mandelbaum Deparmen of Compuer Science York Universiy F05 Secion A Medium Access Conrol (MAC) Topics: 1. Muliple Access Communicaions:

More information

STRING DESCRIPTIONS OF DATA FOR DISPLAY*

STRING DESCRIPTIONS OF DATA FOR DISPLAY* SLAC-PUB-383 January 1968 STRING DESCRIPTIONS OF DATA FOR DISPLAY* J. E. George and W. F. Miller Compuer Science Deparmen and Sanford Linear Acceleraor Cener Sanford Universiy Sanford, California Absrac

More information

EECS 487: Interactive Computer Graphics

EECS 487: Interactive Computer Graphics EECS 487: Ineracive Compuer Graphics Lecure 7: B-splines curves Raional Bézier and NURBS Cubic Splines A represenaion of cubic spline consiss of: four conrol poins (why four?) hese are compleely user specified

More information

Performance Evaluation of Implementing Calls Prioritization with Different Queuing Disciplines in Mobile Wireless Networks

Performance Evaluation of Implementing Calls Prioritization with Different Queuing Disciplines in Mobile Wireless Networks Journal of Compuer Science 2 (5): 466-472, 2006 ISSN 1549-3636 2006 Science Publicaions Performance Evaluaion of Implemening Calls Prioriizaion wih Differen Queuing Disciplines in Mobile Wireless Neworks

More information

Sam knows that his MP3 player has 40% of its battery life left and that the battery charges by an additional 12 percentage points every 15 minutes.

Sam knows that his MP3 player has 40% of its battery life left and that the battery charges by an additional 12 percentage points every 15 minutes. 8.F Baery Charging Task Sam wans o ake his MP3 player and his video game player on a car rip. An hour before hey plan o leave, he realized ha he forgo o charge he baeries las nigh. A ha poin, he plugged

More information

4 Error Control. 4.1 Issues with Reliable Protocols

4 Error Control. 4.1 Issues with Reliable Protocols 4 Error Conrol Jus abou all communicaion sysems aemp o ensure ha he daa ges o he oher end of he link wihou errors. Since i s impossible o build an error-free physical layer (alhough some shor links can

More information

MIC2569. Features. General Description. Applications. Typical Application. CableCARD Power Switch

MIC2569. Features. General Description. Applications. Typical Application. CableCARD Power Switch CableCARD Power Swich General Descripion is designed o supply power o OpenCable sysems and CableCARD hoss. These CableCARDs are also known as Poin of Disribuion (POD) cards. suppors boh Single and Muliple

More information

Improving the Efficiency of Dynamic Service Provisioning in Transport Networks with Scheduled Services

Improving the Efficiency of Dynamic Service Provisioning in Transport Networks with Scheduled Services Improving he Efficiency of Dynamic Service Provisioning in Transpor Neworks wih Scheduled Services Ralf Hülsermann, Monika Jäger and Andreas Gladisch Technologiezenrum, T-Sysems, Goslarer Ufer 35, D-1585

More information

Chapter 4 Sequential Instructions

Chapter 4 Sequential Instructions Chaper 4 Sequenial Insrucions The sequenial insrucions of FBs-PLC shown in his chaper are also lised in secion 3.. Please refer o Chaper, "PLC Ladder diagram and he Coding rules of Mnemonic insrucion",

More information

Image segmentation. Motivation. Objective. Definitions. A classification of segmentation techniques. Assumptions for thresholding

Image segmentation. Motivation. Objective. Definitions. A classification of segmentation techniques. Assumptions for thresholding Moivaion Image segmenaion Which pixels belong o he same objec in an image/video sequence? (spaial segmenaion) Which frames belong o he same video sho? (emporal segmenaion) Which frames belong o he same

More information

Querying Moving Objects in SECONDO

Querying Moving Objects in SECONDO Querying Moving Objecs in SECONDO Vicor Teixeira de Almeida, Ralf Harmu Güing, and Thomas Behr LG Daenbanksyseme für neue Anwendungen Fachbereich Informaik, Fernuniversiä Hagen D-58084 Hagen, Germany {vicor.almeida,

More information

NEWTON S SECOND LAW OF MOTION

NEWTON S SECOND LAW OF MOTION Course and Secion Dae Names NEWTON S SECOND LAW OF MOTION The acceleraion of an objec is defined as he rae of change of elociy. If he elociy changes by an amoun in a ime, hen he aerage acceleraion during

More information

MORPHOLOGICAL SEGMENTATION OF IMAGE SEQUENCES

MORPHOLOGICAL SEGMENTATION OF IMAGE SEQUENCES MORPHOLOGICAL SEGMENTATION OF IMAGE SEQUENCES B. MARCOTEGUI and F. MEYER Ecole des Mines de Paris, Cenre de Morphologie Mahémaique, 35, rue Sain-Honoré, F 77305 Fonainebleau Cedex, France Absrac. In image

More information

AML710 CAD LECTURE 11 SPACE CURVES. Space Curves Intrinsic properties Synthetic curves

AML710 CAD LECTURE 11 SPACE CURVES. Space Curves Intrinsic properties Synthetic curves AML7 CAD LECTURE Space Curves Inrinsic properies Synheic curves A curve which may pass hrough any region of hreedimensional space, as conrased o a plane curve which mus lie on a single plane. Space curves

More information

Why not experiment with the system itself? Ways to study a system System. Application areas. Different kinds of systems

Why not experiment with the system itself? Ways to study a system System. Application areas. Different kinds of systems Simulaion Wha is simulaion? Simple synonym: imiaion We are ineresed in sudying a Insead of experimening wih he iself we experimen wih a model of he Experimen wih he Acual Ways o sudy a Sysem Experimen

More information

Visual Indoor Localization with a Floor-Plan Map

Visual Indoor Localization with a Floor-Plan Map Visual Indoor Localizaion wih a Floor-Plan Map Hang Chu Dep. of ECE Cornell Universiy Ihaca, NY 14850 hc772@cornell.edu Absrac In his repor, a indoor localizaion mehod is presened. The mehod akes firsperson

More information

Performance and Availability Assessment for the Configuration of Distributed Workflow Management Systems

Performance and Availability Assessment for the Configuration of Distributed Workflow Management Systems Absrac Performance and Availabiliy Assessmen for he Configuraion of Disribued Workflow Managemen Sysems Michael Gillmann 1, Jeanine Weissenfels 1, Gerhard Weikum 1, Achim Kraiss 2 1 Universiy of he Saarland,

More information

4.1 3D GEOMETRIC TRANSFORMATIONS

4.1 3D GEOMETRIC TRANSFORMATIONS MODULE IV MCA - 3 COMPUTER GRAPHICS ADMN 29- Dep. of Compuer Science And Applicaions, SJCET, Palai 94 4. 3D GEOMETRIC TRANSFORMATIONS Mehods for geomeric ransformaions and objec modeling in hree dimensions

More information

Coded Caching with Multiple File Requests

Coded Caching with Multiple File Requests Coded Caching wih Muliple File Requess Yi-Peng Wei Sennur Ulukus Deparmen of Elecrical and Compuer Engineering Universiy of Maryland College Park, MD 20742 ypwei@umd.edu ulukus@umd.edu Absrac We sudy a

More information

FUZZY HUMAN/MACHINE RELIABILITY USING VHDL

FUZZY HUMAN/MACHINE RELIABILITY USING VHDL FUZZY HUMN/MCHINE RELIBILITY USING VHDL Carlos. Graciós M. 1, lejandro Díaz S. 2, Efrén Gorroiea H. 3 (1) Insiuo Tecnológico de Puebla v. Tecnológico 420. Col. Maravillas, C. P. 72220, Puebla, Pue. México

More information

User Adjustable Process Scheduling Mechanism for a Multiprocessor Embedded System

User Adjustable Process Scheduling Mechanism for a Multiprocessor Embedded System Proceedings of he 6h WSEAS Inernaional Conference on Applied Compuer Science, Tenerife, Canary Islands, Spain, December 16-18, 2006 346 User Adjusable Process Scheduling Mechanism for a Muliprocessor Embedded

More information

MOTION DETECTORS GRAPH MATCHING LAB PRE-LAB QUESTIONS

MOTION DETECTORS GRAPH MATCHING LAB PRE-LAB QUESTIONS NME: TE: LOK: MOTION ETETORS GRPH MTHING L PRE-L QUESTIONS 1. Read he insrucions, and answer he following quesions. Make sure you resae he quesion so I don hae o read he quesion o undersand he answer..

More information

Visualizing Complex Notions of Time

Visualizing Complex Notions of Time Visualizing Complex Noions of Time Rober Kosara, Silvia Miksch Insiue of Sofware Technology, Vienna Universiy of Technology, Vienna, Ausria Absrac Time plays an imporan role in medicine. Condiions are

More information

COMP26120: Algorithms and Imperative Programming

COMP26120: Algorithms and Imperative Programming COMP26120 ecure C3 1/48 COMP26120: Algorihms and Imperaive Programming ecure C3: C - Recursive Daa Srucures Pee Jinks School of Compuer Science, Universiy of Mancheser Auumn 2011 COMP26120 ecure C3 2/48

More information

The Roots of Lisp paul graham

The Roots of Lisp paul graham The Roos of Lisp paul graham Draf, January 18, 2002. In 1960, John McCarhy published a remarkable paper in which he did for programming somehing like wha Euclid did for geomery. 1 He showed how, given

More information

Design and Application of Computer-aided English Online Examination System NONG DeChang 1, a

Design and Application of Computer-aided English Online Examination System NONG DeChang 1, a 3rd Inernaional Conference on Maerials Engineering, Manufacuring Technology and Conrol (ICMEMTC 2016) Design and Applicaion of Compuer-aided English Online Examinaion Sysem NONG DeChang 1, a 1,2 Guangxi

More information

Syntax Specification by Graph Grammars and Meta-Models

Syntax Specification by Graph Grammars and Meta-Models Ou Synax Speciicaion by Graph Grammars and Mea-Models Mark Minas Insiue or Soware Technology Universiä der Bundeswehr München Germany (Some) Dimensions o Visual Languages & Ediors DiaGen Edior archiecure

More information

M(t)/M/1 Queueing System with Sinusoidal Arrival Rate

M(t)/M/1 Queueing System with Sinusoidal Arrival Rate 20 TUTA/IOE/PCU Journal of he Insiue of Engineering, 205, (): 20-27 TUTA/IOE/PCU Prined in Nepal M()/M/ Queueing Sysem wih Sinusoidal Arrival Rae A.P. Pan, R.P. Ghimire 2 Deparmen of Mahemaics, Tri-Chandra

More information

Less Pessimistic Worst-Case Delay Analysis for Packet-Switched Networks

Less Pessimistic Worst-Case Delay Analysis for Packet-Switched Networks Less Pessimisic Wors-Case Delay Analysis for Packe-Swiched Neworks Maias Wecksén Cenre for Research on Embedded Sysems P O Box 823 SE-31 18 Halmsad maias.wecksen@hh.se Magnus Jonsson Cenre for Research

More information

BI-TEMPORAL INDEXING

BI-TEMPORAL INDEXING BI-TEMPORAL INDEXING Mirella M. Moro Uniersidade Federal do Rio Grande do Sul Poro Alegre, RS, Brazil hp://www.inf.ufrgs.br/~mirella/ Vassilis J. Tsoras Uniersiy of California, Rierside Rierside, CA 92521,

More information

Achieving Security Assurance with Assertion-based Application Construction

Achieving Security Assurance with Assertion-based Application Construction Achieving Securiy Assurance wih Asserion-based Applicaion Consrucion Carlos E. Rubio-Medrano and Gail-Joon Ahn Ira A. Fulon Schools of Engineering Arizona Sae Universiy Tempe, Arizona, USA, 85282 {crubiome,

More information

MOBILE COMPUTING 3/18/18. Wi-Fi IEEE. CSE 40814/60814 Spring 2018

MOBILE COMPUTING 3/18/18. Wi-Fi IEEE. CSE 40814/60814 Spring 2018 MOBILE COMPUTING CSE 40814/60814 Spring 2018 Wi-Fi Wi-Fi: name is NOT an abbreviaion play on Hi-Fi (high fideliy) Wireless Local Area Nework (WLAN) echnology WLAN and Wi-Fi ofen used synonymous Typically

More information

MOBILE COMPUTING. Wi-Fi 9/20/15. CSE 40814/60814 Fall Wi-Fi:

MOBILE COMPUTING. Wi-Fi 9/20/15. CSE 40814/60814 Fall Wi-Fi: MOBILE COMPUTING CSE 40814/60814 Fall 2015 Wi-Fi Wi-Fi: name is NOT an abbreviaion play on Hi-Fi (high fideliy) Wireless Local Area Nework (WLAN) echnology WLAN and Wi-Fi ofen used synonymous Typically

More information

A GRAPHICS PROCESSING UNIT IMPLEMENTATION OF THE PARTICLE FILTER

A GRAPHICS PROCESSING UNIT IMPLEMENTATION OF THE PARTICLE FILTER A GRAPHICS PROCESSING UNIT IMPLEMENTATION OF THE PARTICLE FILTER Gusaf Hendeby, Jeroen D. Hol, Rickard Karlsson, Fredrik Gusafsson Deparmen of Elecrical Engineering Auomaic Conrol Linköping Universiy,

More information

Construction Process. Transactional Process Scheduler. Production Process. 2.3 Transactional Subsystems. Test. CAD Documentation. Conflict!

Construction Process. Transactional Process Scheduler. Production Process. 2.3 Transactional Subsystems. Test. CAD Documentation. Conflict! Philadelphia, Pennsylvania, USA, May 31 - June 2, 1999. Concurrency Conrol and Recovery in Transacional Process Managemen Heo Schuld Gusavo Alonso Insiue of Informaion Sysems Swiss Federal Insiue of Technology

More information

Network management and QoS provisioning - QoS in Frame Relay. . packet switching with virtual circuit service (virtual circuits are bidirectional);

Network management and QoS provisioning - QoS in Frame Relay. . packet switching with virtual circuit service (virtual circuits are bidirectional); QoS in Frame Relay Frame relay characerisics are:. packe swiching wih virual circui service (virual circuis are bidirecional);. labels are called DLCI (Daa Link Connecion Idenifier);. for connecion is

More information

Axiomatic Foundations and Algorithms for Deciding Semantic Equivalences of SQL Queries

Axiomatic Foundations and Algorithms for Deciding Semantic Equivalences of SQL Queries Axiomaic Foundaions and Algorihms for Deciding Semanic Equivalences of SQL Queries Shumo Chu, Brendan Murphy, Jared Roesch, Alvin Cheung, Dan Suciu Paul G. Allen School of Compuer Science and Engineering

More information

Test - Accredited Configuration Engineer (ACE) Exam - PAN-OS 6.0 Version

Test - Accredited Configuration Engineer (ACE) Exam - PAN-OS 6.0 Version Tes - Accredied Configuraion Engineer (ACE) Exam - PAN-OS 6.0 Version ACE Exam Quesion 1 of 50. Which of he following saemens is NOT abou Palo Alo Neworks firewalls? Sysem defauls may be resored by performing

More information

Quantitative macro models feature an infinite number of periods A more realistic (?) view of time

Quantitative macro models feature an infinite number of periods A more realistic (?) view of time INFINIE-HORIZON CONSUMPION-SAVINGS MODEL SEPEMBER, Inroducion BASICS Quaniaive macro models feaure an infinie number of periods A more realisic (?) view of ime Infinie number of periods A meaphor for many

More information

Y. Tsiatouhas. VLSI Systems and Computer Architecture Lab

Y. Tsiatouhas. VLSI Systems and Computer Architecture Lab CMOS INEGRAED CIRCUI DESIGN ECHNIQUES Universiy of Ioannina Clocking Schemes Dep. of Compuer Science and Engineering Y. siaouhas CMOS Inegraed Circui Design echniques Overview 1. Jier Skew hroughpu Laency

More information

Why Waste a Perfectly Good Abstraction?

Why Waste a Perfectly Good Abstraction? Why Wase a Perfecly Good Absracion? Arie Gurfinkel and Marsha Chechik Deparmen of Compuer Science, Universiy of Torono, Torono, ON M5S 3G4, Canada. Email: arie,chechik@cs.orono.edu Absrac. Sofware model-checking

More information

An efficient approach to improve throughput for TCP vegas in ad hoc network

An efficient approach to improve throughput for TCP vegas in ad hoc network Inernaional Research Journal of Engineering and Technology (IRJET) e-issn: 395-0056 Volume: 0 Issue: 03 June-05 www.irje.ne p-issn: 395-007 An efficien approach o improve hroughpu for TCP vegas in ad hoc

More information

Video-Based Face Recognition Using Probabilistic Appearance Manifolds

Video-Based Face Recognition Using Probabilistic Appearance Manifolds Video-Based Face Recogniion Using Probabilisic Appearance Manifolds Kuang-Chih Lee Jeffrey Ho Ming-Hsuan Yang David Kriegman klee10@uiuc.edu jho@cs.ucsd.edu myang@honda-ri.com kriegman@cs.ucsd.edu Compuer

More information

Probabilistic Detection and Tracking of Motion Discontinuities

Probabilistic Detection and Tracking of Motion Discontinuities Probabilisic Deecion and Tracking of Moion Disconinuiies Michael J. Black David J. Flee Xerox Palo Alo Research Cener 3333 Coyoe Hill Road Palo Alo, CA 94304 fblack,fleeg@parc.xerox.com hp://www.parc.xerox.com/fblack,fleeg/

More information

The Beer Dock: Three and a Half Implementations of the Beer Distribution Game

The Beer Dock: Three and a Half Implementations of the Beer Distribution Game The Beer Dock 2002-08-13 17:55:44-0700 The Beer Dock: Three and a Half Implemenaions of he Beer Disribuion Game Michael J. Norh[1] and Charles M. Macal Argonne Naional Laboraory, Argonne, Illinois Absrac

More information

SOT: Compact Representation for Triangle and Tetrahedral Meshes

SOT: Compact Representation for Triangle and Tetrahedral Meshes SOT: Compac Represenaion for Triangle and Terahedral Meshes Topraj Gurung and Jarek Rossignac School of Ineracive Compuing, College of Compuing, Georgia Insiue of Technology, Alana, GA ABSTRACT The Corner

More information

MB86297A Carmine Timing Analysis of the DDR Interface

MB86297A Carmine Timing Analysis of the DDR Interface Applicaion Noe MB86297A Carmine Timing Analysis of he DDR Inerface Fujisu Microelecronics Europe GmbH Hisory Dae Auhor Version Commen 05.02.2008 Anders Ramdahl 0.01 Firs draf 06.02.2008 Anders Ramdahl

More information

Dynamic Route Planning and Obstacle Avoidance Model for Unmanned Aerial Vehicles

Dynamic Route Planning and Obstacle Avoidance Model for Unmanned Aerial Vehicles Volume 116 No. 24 2017, 315-329 ISSN: 1311-8080 (prined version); ISSN: 1314-3395 (on-line version) url: hp://www.ijpam.eu ijpam.eu Dynamic Roue Planning and Obsacle Avoidance Model for Unmanned Aerial

More information

Location. Electrical. Loads. 2-wire mains-rated. 0.5 mm² to 1.5 mm² Max. length 300 m (with 1.5 mm² cable). Example: Belden 8471

Location. Electrical. Loads. 2-wire mains-rated. 0.5 mm² to 1.5 mm² Max. length 300 m (with 1.5 mm² cable). Example: Belden 8471 Produc Descripion Insallaion and User Guide Transiser Dimmer (454) The DIN rail mouned 454 is a 4channel ransisor dimmer. I can operae in one of wo modes; leading edge or railing edge. All 4 channels operae

More information

FLORIDA INTERNATIONAL UNIVERSITY. Miami, Florida DIMUSE: AN INTEGRATED FRAMEWORK FOR DISTRIBUTED MULTIMEDIA

FLORIDA INTERNATIONAL UNIVERSITY. Miami, Florida DIMUSE: AN INTEGRATED FRAMEWORK FOR DISTRIBUTED MULTIMEDIA FLORIDA INTERNATIONAL UNIVERSITY Miami, Florida DIMUSE: AN INTEGRATED FRAMEWORK FOR DISTRIBUTED MULTIMEDIA SYSTEM WITH DATABASE MANAGEMENT AND SECURITY SUPPORT A disseraion submied in parial fulfillmen

More information

Optimizing the Processing Performance of a Smart DMA Controller for LTE Terminals

Optimizing the Processing Performance of a Smart DMA Controller for LTE Terminals Opimizing he Processing Performance of a Smar DMA Conroller for LTE Terminals David Szczesny, Sebasian Hessel, Shadi Traboulsi, Aila Bilgic Insiue for Inegraed Sysems, Ruhr-Universiä Bochum D-78 Bochum,

More information

MATH Differential Equations September 15, 2008 Project 1, Fall 2008 Due: September 24, 2008

MATH Differential Equations September 15, 2008 Project 1, Fall 2008 Due: September 24, 2008 MATH 5 - Differenial Equaions Sepember 15, 8 Projec 1, Fall 8 Due: Sepember 4, 8 Lab 1.3 - Logisics Populaion Models wih Harvesing For his projec we consider lab 1.3 of Differenial Equaions pages 146 o

More information

Distributed Task Negotiation in Modular Robots

Distributed Task Negotiation in Modular Robots Disribued Task Negoiaion in Modular Robos Behnam Salemi, eer Will, and Wei-Min Shen USC Informaion Sciences Insiue and Compuer Science Deparmen Marina del Rey, USA, {salemi, will, shen}@isi.edu Inroducion

More information

Reinforcement Learning by Policy Improvement. Making Use of Experiences of The Other Tasks. Hajime Kimura and Shigenobu Kobayashi

Reinforcement Learning by Policy Improvement. Making Use of Experiences of The Other Tasks. Hajime Kimura and Shigenobu Kobayashi Reinforcemen Learning by Policy Improvemen Making Use of Experiences of The Oher Tasks Hajime Kimura and Shigenobu Kobayashi Tokyo Insiue of Technology, JAPAN genfe.dis.iech.ac.jp, kobayasidis.iech.ac.jp

More information

Small Spacecraft Software Modeling: A Petri Net-Based Approach

Small Spacecraft Software Modeling: A Petri Net-Based Approach SSC13-VIII-3 Small Spacecraf Sofware Modeling: A Peri Ne-Based Approach Levi, Pasha Missouri Universiy of Science and Technology 400 Wes 13h Sree, Rolla, MO 65409-0050; 573-341-7280 lmnn3@ms.edu Faculy

More information

Improved TLD Algorithm for Face Tracking

Improved TLD Algorithm for Face Tracking Absrac Improved TLD Algorihm for Face Tracking Huimin Li a, Chaojing Yu b and Jing Chen c Chongqing Universiy of Poss and Telecommunicaions, Chongqing 400065, China a li.huimin666@163.com, b 15023299065@163.com,

More information

Shortest Path Algorithms. Lecture I: Shortest Path Algorithms. Example. Graphs and Matrices. Setting: Dr Kieran T. Herley.

Shortest Path Algorithms. Lecture I: Shortest Path Algorithms. Example. Graphs and Matrices. Setting: Dr Kieran T. Herley. Shores Pah Algorihms Background Seing: Lecure I: Shores Pah Algorihms Dr Kieran T. Herle Deparmen of Compuer Science Universi College Cork Ocober 201 direced graph, real edge weighs Le he lengh of a pah

More information

IntentSearch:Capturing User Intention for One-Click Internet Image Search

IntentSearch:Capturing User Intention for One-Click Internet Image Search JOURNAL OF L A T E X CLASS FILES, VOL. 6, NO. 1, JANUARY 2010 1 InenSearch:Capuring User Inenion for One-Click Inerne Image Search Xiaoou Tang, Fellow, IEEE, Ke Liu, Jingyu Cui, Suden Member, IEEE, Fang

More information

MoBAN: A Configurable Mobility Model for Wireless Body Area Networks

MoBAN: A Configurable Mobility Model for Wireless Body Area Networks MoBAN: A Configurable Mobiliy Model for Wireless Body Area Neworks Majid Nabi 1, Marc Geilen 1, Twan Basen 1,2 1 Deparmen of Elecrical Engineering, Eindhoven Universiy of Technology, he Neherlands 2 Embedded

More information

Network Slicing for Ultra-Reliable Low Latency Communication in Industry 4.0 Scenarios

Network Slicing for Ultra-Reliable Low Latency Communication in Industry 4.0 Scenarios 1 Nework Slicing for Ulra-Reliable Low Laency Communicaion in Indusry 4.0 Scenarios Anders Ellersgaard Kalør, René Guillaume, Jimmy Jessen Nielsen, Andreas Mueller, and Pear Popovski arxiv:1708.09132v1

More information

Vulnerability Evaluation of Multimedia Subsystem Based on Complex Network

Vulnerability Evaluation of Multimedia Subsystem Based on Complex Network JOURAL OF MULTIMDIA, VOL. 8, O. 4, AUGUST 23 439 Vulnerabiliy valuaion of Mulimedia Subsysem Based on Complex ewor Xiaoling Tang Insiue of Higher ducaion Research, Jilin Business and Technology College,

More information

Design Alternatives for a Thin Lens Spatial Integrator Array

Design Alternatives for a Thin Lens Spatial Integrator Array Egyp. J. Solids, Vol. (7), No. (), (004) 75 Design Alernaives for a Thin Lens Spaial Inegraor Array Hala Kamal *, Daniel V azquez and Javier Alda and E. Bernabeu Opics Deparmen. Universiy Compluense of

More information

I. INTRODUCTION. Keywords -- Web Server, Perceived User Latency, HTTP, Local Measuring. interchangeably.

I. INTRODUCTION. Keywords -- Web Server, Perceived User Latency, HTTP, Local Measuring. interchangeably. Evaluaing Web User Perceived Laency Using Server Side Measuremens Marik Marshak 1 and Hanoch Levy School of Compuer Science Tel Aviv Universiy, Tel-Aviv, Israel mmarshak@emc.com, hanoch@pos.au.ac.il 1

More information

Representing Non-Manifold Shapes in Arbitrary Dimensions

Representing Non-Manifold Shapes in Arbitrary Dimensions Represening Non-Manifold Shapes in Arbirary Dimensions Leila De Floriani,2 and Annie Hui 2 DISI, Universiy of Genova, Via Dodecaneso, 35-646 Genova (Ialy). 2 Deparmen of Compuer Science, Universiy of Maryland,

More information