FORTRAN95并行.rar_FORTRAN95 并行

<html xmlns="http://www.w3.org/1999/xhtml"> <head> <meta charset="utf-8"> <meta name="generator" content="pdf2htmlEX"> <meta http-equiv="X-UA-Compatible" content="IE=edge,chrome=1"> <link rel="stylesheet" href="https://static.pudn.com/base/css/base.min.css"> <link rel="stylesheet" href="https://static.pudn.com/base/css/fancy.min.css"> <link rel="stylesheet" href="https://static.pudn.com/prod/directory_preview_static/6250236f74bc5c01055be476/raw.css"> <script src="https://static.pudn.com/base/js/compatibility.min.js"></script> <script src="https://static.pudn.com/base/js/pdf2htmlEX.min.js"></script> <script> try{ pdf2htmlEX.defaultViewer = new pdf2htmlEX.Viewer({}); }catch(e){} </script> <title></title> </head> <body> <div id="sidebar" style="display: none"> <div id="outline"> </div> </div> <div id="pf1" class="pf w0 h0" data-page-no="1"><div class="pc pc1 w0 h0"><img class="bi x0 y0 w1 h1" alt="" src="https://static.pudn.com/prod/directory_preview_static/6250236f74bc5c01055be476/bg1.jpg"><div class="t m0 x1 h2 y1 ff1 fs0 fc0 sc0 ls0 ws0">Parallel Programming in Fortran 95 using OpenMP</div><div class="t m0 x2 h3 y2 ff2 fs1 fc0 sc0 ls0 ws0">Miguel Hermanns</div><div class="t m0 x3 h4 y3 ff3 fs2 fc0 sc0 ls0 ws0">School of Aeronautical Engineering</div><div class="t m0 x4 h4 y4 ff3 fs2 fc0 sc0 ls1 ws0">Departamento de Motopropulsi´on y Termoﬂuidodin´amica</div><div class="t m0 x5 h4 y5 ff3 fs2 fc0 sc0 ls2 ws0">Universidad Polit´ecnica de Madrid</div><div class="t m0 x6 h4 y6 ff3 fs2 fc0 sc0 ls0 ws0">Spain</div><div class="t m0 x7 h4 y7 ff3 fs2 fc0 sc0 ls3 ws0">email: hermanns@tupi.dmt.upm.es</div><div class="t m0 x2 h3 y8 ff2 fs1 fc0 sc0 ls5 ws0">19</div><div class="t m0 x8 h5 y9 ff5 fs3 fc0 sc0 ls6 ws0">th</div><div class="t m0 x9 h3 y8 ff2 fs1 fc0 sc0 ls7 ws0">of April 2002</div></div><div class="pi" data-data='{"ctm":[1.611639,0.000000,0.000000,1.611639,0.000000,0.000000]}'></div></div> </body> </html>

<div id="pf3" class="pf w0 h0" data-page-no="3"><div class="pc pc3 w0 h0"><img class="bi x0 y0 w1 h1" alt="" src="https://static.pudn.com/prod/directory_preview_static/6250236f74bc5c01055be476/bg3.jpg"><div class="t m0 xa h6 ya ff6 fs4 fc0 sc0 ls0 ws0">Contents</div><div class="t m0 xa h7 yb ff6 fs5 fc0 sc0 ls8 ws0">1 OpenMP Fortran Application Program Interface 1</div><div class="t m0 xb h8 yc ff2 fs5 fc0 sc0 ls9 ws0">1.1 Introduction................................... 1</div><div class="t m0 xc h8 yd ff2 fs5 fc0 sc0 ls9 ws0">1.1.1 Historicalremarks ........................... 2</div><div class="t m0 xc h8 ye ff2 fs5 fc0 sc0 ls9 ws0">1.1.2 Whoisparticipating .......................... 2</div><div class="t m0 xc h8 yf ff2 fs5 fc0 sc0 ls9 ws0">1.1.3 Aboutthisdocument.......................... 3</div><div class="t m0 xb h8 y10 ff2 fs5 fc0 sc0 ls9 ws0">1.2 Thebasics.................................... 4</div><div class="t m0 xc h8 y11 ff2 fs5 fc0 sc0 ls8 ws0">1.2.1 The sentinels for OpenMP directives and conditional compilation . 4</div><div class="t m0 xc h8 y12 ff2 fs5 fc0 sc0 ls9 ws0">1.2.2 Theparallelregionconstructor .................... 5</div><div class="t m0 xa h7 y13 ff6 fs5 fc0 sc0 ls8 ws0">2 OpenMP constructs 9</div><div class="t m0 xb h8 y14 ff2 fs5 fc0 sc0 lsa ws0">2.1 Work-sharingconstructs ............................ 9</div><div class="t m0 xc h8 y15 ff2 fs5 fc0 sc0 lsb ws0">2.1.1 !$OMP DO/!$OMP END DO ........................ 10</div><div class="t m0 xc h8 y16 ff2 fs5 fc0 sc0 lsb ws0">2.1.2 !$OMP SECTIONS/!$OMP END SECTIONS ................ 15</div><div class="t m0 xc h8 y17 ff2 fs5 fc0 sc0 lsb ws0">2.1.3 !$OMP SINGLE/!$OMP END SINGLE .................. 16</div><div class="t m0 xc h8 y18 ff2 fs5 fc0 sc0 lsb ws0">2.1.4 !$OMP WORKSHARE/!$OMP END WORKSHARE .............. 17</div><div class="t m0 xb h8 y19 ff2 fs5 fc0 sc0 ls9 ws0">2.2 Combinedparallelwork-sharingconstructs.................. 20</div><div class="t m0 xc h8 y1a ff2 fs5 fc0 sc0 lsb ws0">2.2.1 !$OMP PARALLEL DO/!$OMP END PARALLEL DO ............ 21</div><div class="t m0 xc h8 y1b ff2 fs5 fc0 sc0 lsb ws0">2.2.2 !$OMP PARALLEL SECTIONS/!$OMP END PARALLEL SECTIONS .... 21</div><div class="t m0 xc h8 y1c ff2 fs5 fc0 sc0 lsb ws0">2.2.3 !$OMP PARALLEL WORKSHARE/!$OMP END PARALLEL WORKSHARE .. 21</div><div class="t m0 xb h8 y1d ff2 fs5 fc0 sc0 ls9 ws0">2.3 Synchronizationconstructs........................... 22</div><div class="t m0 xc h8 y1e ff2 fs5 fc0 sc0 lsb ws0">2.3.1 !$OMP MASTER/!$OMP END MASTER .................. 22</div><div class="t m0 xc h8 y1f ff2 fs5 fc0 sc0 lsb ws0">2.3.2 !$OMP CRITICAL/!$OMP END CRITICAL ................ 22</div><div class="t m0 xc h8 y20 ff2 fs5 fc0 sc0 lsb ws0">2.3.3 !$OMP BARRIER ............................. 24</div><div class="t m0 xc h8 y21 ff2 fs5 fc0 sc0 lsb ws0">2.3.4 !$OMP ATOMIC .............................. 26</div><div class="t m0 xc h8 y22 ff2 fs5 fc0 sc0 lsb ws0">2.3.5 !$OMP FLUSH .............................. 27</div><div class="t m0 xc h8 y23 ff2 fs5 fc0 sc0 lsb ws0">2.3.6 !$OMP ORDERED/!$OMP END ORDERED ................. 30</div><div class="t m0 xb h8 y24 ff2 fs5 fc0 sc0 ls9 ws0">2.4 Dataenvironmentconstructs.......................... 32</div><div class="t m0 xc h8 y25 ff2 fs5 fc0 sc0 lsb ws0">2.4.1 !$OMP THREADPRIVATE (list) .................... 32</div><div class="t m0 xa h7 y26 ff6 fs5 fc0 sc0 ls0 ws0">3 PRIVATE, SHARED & Co. 37</div><div class="t m0 xb h8 y27 ff2 fs5 fc0 sc0 ls8 ws0">3.1 Data scope attribute clauses . . . . . . . . . . . . . . . . . . . . . . . . . . 37</div><div class="t m0 xc h8 y28 ff2 fs5 fc0 sc0 lsb ws0">3.1.1 PRIVATE(list) ............................. 37</div><div class="t m0 xc h8 y29 ff2 fs5 fc0 sc0 lsb ws0">3.1.2 SHARED(list) .............................. 38</div><div class="t m0 xc h8 y2a ff2 fs5 fc0 sc0 lsb ws0">3.1.3 DEFAULT( PRIVATE | SHARED | NONE ) ............... 39</div><div class="t m0 xd h8 y2b ff2 fs5 fc0 sc0 ls0 ws0">i</div></div><div class="pi" data-data='{"ctm":[1.611639,0.000000,0.000000,1.611639,0.000000,0.000000]}'></div></div>

<div id="pf4" class="pf w0 h0" data-page-no="4"><div class="pc pc4 w0 h0"><img class="bi x0 y0 w1 h1" alt="" src="https://static.pudn.com/prod/directory_preview_static/6250236f74bc5c01055be476/bg4.jpg"><div class="t m0 xa h8 y2c ff2 fs5 fc0 sc0 lsf ws0">ii CONTENTS</div><div class="t m0 xc h8 y2d ff2 fs5 fc0 sc0 lsb ws0">3.1.4 FIRSTPRIVATE(list) .......................... 40</div><div class="t m0 xc h8 y2e ff2 fs5 fc0 sc0 lsb ws0">3.1.5 LASTPRIVATE(list) .......................... 41</div><div class="t m0 xc h8 y2f ff2 fs5 fc0 sc0 lsb ws0">3.1.6 COPYIN(list) .............................. 42</div><div class="t m0 xc h8 y30 ff2 fs5 fc0 sc0 lsb ws0">3.1.7 COPYPRIVATE(list) .......................... 43</div><div class="t m0 xc h8 y31 ff2 fs5 fc0 sc0 lsb ws0">3.1.8 REDUCTION(operator:list) ..................... 43</div><div class="t m0 xb h8 y32 ff2 fs5 fc0 sc0 ls9 ws0">3.2 Otherclauses .................................. 46</div><div class="t m0 xc h8 y33 ff2 fs5 fc0 sc0 lsb ws0">3.2.1 IF(scalar</div><div class="t m0 x3 h8 y34 ff8 fs5 fc0 sc0 lsb ws0">logical expression) .................. 46</div><div class="t m0 xc h8 y35 ff2 fs5 fc0 sc0 lsb ws0">3.2.2 NUM</div><div class="t m0 xe h8 y36 ff7 fs5 fc1 sc0 lsc ws0">THREADS(scalar integer expression) ............. 47</div><div class="t m0 xc h8 y37 ff2 fs5 fc0 sc0 lsb ws0">3.2.3 NOWAIT .................................. 47</div><div class="t m0 xc h8 y38 ff2 fs5 fc0 sc0 lsb ws0">3.2.4 SCHEDULE(t ype, chunk) ....................... 48</div><div class="t m0 xc h8 y39 ff2 fs5 fc0 sc0 lsb ws0">3.2.5 ORDERED ................................. 52</div><div class="t m0 xa h7 y3a ff6 fs5 fc0 sc0 lsf ws0">4 The OpenMP run-time library 55</div><div class="t m0 xb h8 y3b ff2 fs5 fc0 sc0 ls9 ws0">4.1 Executionenvironmentroutines........................ 55</div><div class="t m0 xc h8 y3c ff2 fs5 fc0 sc0 lsb ws0">4.1.1 OMP</div><div class="t m0 xe h8 y3d ff7 fs5 fc0 sc0 lsc ws0">set num threads .......................... 55</div><div class="t m0 xc h8 y3e ff2 fs5 fc0 sc0 lsb ws0">4.1.2 OMP</div><div class="t m0 xe h8 y3f ff7 fs5 fc0 sc0 lsc ws0">get num threads .......................... 56</div><div class="t m0 xc h8 y40 ff2 fs5 fc0 sc0 lsb ws0">4.1.3 OMP</div><div class="t m0 xe h8 y41 ff7 fs5 fc0 sc0 lsc ws0">get max threads .......................... 56</div><div class="t m0 xc h8 y42 ff2 fs5 fc0 sc0 lsb ws0">4.1.4 OMP</div><div class="t m0 xe h8 y43 ff7 fs5 fc0 sc0 lsc ws0">get thread num .......................... 56</div><div class="t m0 xc h8 y44 ff2 fs5 fc0 sc0 lsb ws0">4.1.5 OMP</div><div class="t m0 xe h8 y45 ff7 fs5 fc0 sc0 lsc ws0">get num procs ........................... 57</div><div class="t m0 xc h8 y46 ff2 fs5 fc0 sc0 lsb ws0">4.1.6 OMP</div><div class="t m0 xe h8 y47 ff7 fs5 fc0 sc0 lsc ws0">in parallel ............................ 57</div><div class="t m0 xc h8 y48 ff2 fs5 fc0 sc0 lsb ws0">4.1.7 OMP</div><div class="t m0 xe h8 y49 ff7 fs5 fc0 sc0 lsc ws0">set dynamic ............................ 57</div><div class="t m0 xc h8 y4a ff2 fs5 fc0 sc0 lsb ws0">4.1.8 OMP</div><div class="t m0 xe h8 y4b ff7 fs5 fc0 sc0 lsc ws0">get dynamic ............................ 58</div><div class="t m0 xc h8 y4c ff2 fs5 fc0 sc0 lsb ws0">4.1.9 OMP</div><div class="t m0 xe h8 y4d ff7 fs5 fc0 sc0 lsc ws0">set nested ............................. 58</div><div class="t m0 xc h8 y4e ff2 fs5 fc0 sc0 lsb ws0">4.1.10 OMP</div><div class="t m0 xe h8 y4f ff7 fs5 fc0 sc0 lsc ws0">get nested ............................. 58</div><div class="t m0 xb h8 y50 ff2 fs5 fc0 sc0 lsa ws0">4.2 Lockroutines .................................. 59</div><div class="t m0 xc h8 y51 ff2 fs5 fc0 sc0 lsb ws0">4.2.1 OMP</div><div class="t m0 xe h8 y52 ff7 fs5 fc0 sc0 lsc ws0">init lock and OMP init nest lock ............... 60</div><div class="t m0 xc h8 y53 ff2 fs5 fc0 sc0 lsb ws0">4.2.2 OMP</div><div class="t m0 xe h8 y54 ff7 fs5 fc0 sc0 lsc ws0">set lock and OMP set nest lock ................ 60</div><div class="t m0 xc h8 y55 ff2 fs5 fc0 sc0 lsb ws0">4.2.3 OMP</div><div class="t m0 xe h8 y56 ff7 fs5 fc0 sc0 lsc ws0">unset lock and OMP unset nest lock ............. 61</div><div class="t m0 xc h8 y57 ff2 fs5 fc0 sc0 lsb ws0">4.2.4 OMP</div><div class="t m0 xe h8 y58 ff7 fs5 fc0 sc0 lsc ws0">test lock and OMP test nest lock ............... 61</div><div class="t m0 xc h8 y59 ff2 fs5 fc0 sc0 lsb ws0">4.2.5 OMP</div><div class="t m0 xe h8 y5a ff7 fs5 fc0 sc0 lsc ws0">destroy lock and OMP destroy nest lock ........... 62</div><div class="t m0 xc h8 y5b ff2 fs5 fc0 sc0 ls9 ws0">4.2.6 Examples ................................ 62</div><div class="t m0 xb h8 y5c ff2 fs5 fc0 sc0 lsb ws0">4.3 Timing routines ................................. 65</div><div class="t m0 xc h8 y5d ff2 fs5 fc0 sc0 lsb ws0">4.3.1 OMP</div><div class="t m0 xe h8 y5e ff7 fs5 fc1 sc0 lsc ws0">get wtime ............................. 65</div><div class="t m0 xc h8 y5f ff2 fs5 fc0 sc0 lsb ws0">4.3.2 OMP</div><div class="t m0 xe h8 y60 ff7 fs5 fc1 sc0 lsc ws0">get wtick ............................. 66</div><div class="t m0 xb h8 y61 ff2 fs5 fc0 sc0 lsb ws0">4.4 The Fortran 90 module omp</div><div class="t m0 xf h8 y62 ff7 fs5 fc1 sc0 lsc ws0">lib ........................ 66</div><div class="t m0 xa h7 y63 ff6 fs5 fc0 sc0 ls8 ws0">5 The environment variables 69</div><div class="t m0 xb h8 y64 ff2 fs5 fc0 sc0 lsb ws0">5.1 OMP</div><div class="t m0 x10 h8 y65 ff7 fs5 fc0 sc0 lsc ws0">NUM THREADS ................................ 70</div><div class="t m0 xb h8 y66 ff2 fs5 fc0 sc0 lsb ws0">5.2 OMP</div><div class="t m0 x10 h8 y67 ff7 fs5 fc0 sc0 lsc ws0">SCHEDULE .................................. 70</div><div class="t m0 xb h8 y68 ff2 fs5 fc0 sc0 lsb ws0">5.3 OMP</div><div class="t m0 x10 h8 y69 ff7 fs5 fc0 sc0 lsc ws0">DYNAMIC ................................... 71</div><div class="t m0 xb h8 y6a ff2 fs5 fc0 sc0 lsb ws0">5.4 OMP</div><div class="t m0 x10 h8 y6b ff7 fs5 fc0 sc0 lsc ws0">NESTED ................................... 71</div></div><div class="pi" data-data='{"ctm":[1.611639,0.000000,0.000000,1.611639,0.000000,0.000000]}'></div></div>

<div id="pf5" class="pf w0 h0" data-page-no="5"><div class="pc pc5 w0 h0"><img class="bi x0 y0 w1 h1" alt="" src="https://static.pudn.com/prod/directory_preview_static/6250236f74bc5c01055be476/bg5.jpg"><div class="t m0 xa h6 y6c ff6 fs4 fc0 sc0 ls0 ws0">Chapter 1</div><div class="t m0 xa h6 y6d ff6 fs4 fc0 sc0 ls0 ws0">OpenMP Fortran Application</div><div class="t m0 xa h6 y6e ff6 fs4 fc0 sc0 ls10 ws0">Program Interface</div><div class="t m0 xa h9 y6f ff6 fs6 fc0 sc0 ls11 ws0">1.1 Introduction</div><div class="t m0 xa h8 y70 ff2 fs5 fc0 sc0 ls0 ws0">In the necessity of more and more computational power, the developers of computing</div><div class="t m0 xa h8 y71 ff2 fs5 fc0 sc0 ls0 ws0">systems started to think on using several of their existing computing machines in a joined</div><div class="t m0 xa h8 y72 ff2 fs5 fc0 sc0 ls8 ws0">manner. This is the origin of parallel machines and the start of a new ﬁeld for programmers</div><div class="t m0 xa h8 y73 ff2 fs5 fc0 sc0 ls0 ws0">and researches.</div><div class="t m0 xb h8 y74 ff2 fs5 fc0 sc0 ls8 ws0">Nowadays parallel computers are very common in research facilities as well as compa-</div><div class="t m0 xa h8 y75 ff2 fs5 fc0 sc0 ls0 ws0">nies all over the world and are extensively used for complex computations, like simulations</div><div class="t m0 xa h8 y76 ff2 fs5 fc0 sc0 ls8 ws0">of atomic explosions, folding of proteins or turbulent ﬂows.</div><div class="t m0 xb h8 y77 ff2 fs5 fc0 sc0 ls0 ws0">A challenge in parallel machines is the development of codes able of using the capabil-</div><div class="t m0 xa h8 y78 ff2 fs5 fc0 sc0 ls8 ws0">ities of the available hardware in order to solve larger problems in less time. But parallel</div><div class="t m0 xa h8 y79 ff2 fs5 fc0 sc0 ls8 ws0">programming is not an easy task, since a large variety of architectures exist. Mainly two</div><div class="t m0 xa h8 y7a ff2 fs5 fc0 sc0 ls8 ws0">families of parallel machines can be identiﬁed:</div><div class="t m0 xa h8 y7b ff6 fs5 fc0 sc0 lsf ws0">Shared-memory architecture : these parallel machines are build up on a set of proces-</div><div class="t m0 x11 h8 y7c ff2 fs5 fc0 sc0 ls8 ws0">sors which have access to a common memory. Usually the name of SMP machines</div><div class="t m0 x11 h8 y7d ff2 fs5 fc0 sc0 ls0 ws0">is used for computers based on this architecture, where SMP stands for Symmetric</div><div class="t m0 x11 h8 y7e ff6 fs5 fc0 sc0 ls8 ws0">Multi Processing.</div><div class="t m0 xa h8 y7f ff6 fs5 fc0 sc0 lsf ws0">Distributed-memory architecture : in these parallel machines each processor has its</div><div class="t m0 x11 h8 y80 ff2 fs5 fc0 sc0 ls0 ws0">own private memory and information is interchanged between the processors through</div><div class="t m0 x11 h8 y81 ff2 fs5 fc0 sc0 ls8 ws0">messages. The name of clusters is commonly used for this type of computing</div><div class="t m0 x11 h8 y82 ff2 fs5 fc0 sc0 ls8 ws0">devices.</div><div class="t m0 xb h8 y83 ff2 fs5 fc0 sc0 ls8 ws0">Each of the two families has its advantages and disadvantages and the actual parallel</div><div class="t m0 xa h8 y84 ff2 fs5 fc0 sc0 ls0 ws0">programming standards try to exploit these advantages by focusing only on one of these</div><div class="t m0 xa h8 y85 ff2 fs5 fc0 sc0 ls0 ws0">architectures.</div><div class="t m0 xb h8 y86 ff2 fs5 fc0 sc0 ls0 ws0">In the last years a new industry standard has been created with the aim to serve</div><div class="t m0 xa h8 y87 ff2 fs5 fc0 sc0 ls8 ws0">as a good basis for the development of parallel programs on shared-memory machines:</div><div class="t m0 xa h8 y88 ff6 fs5 fc0 sc0 ls0 ws0">OpenMP.</div><div class="t m0 x12 h8 y89 ff2 fs5 fc0 sc0 ls0 ws0">1</div></div><div class="pi" data-data='{"ctm":[1.611639,0.000000,0.000000,1.611639,0.000000,0.000000]}'></div></div>