BiR SANAL PARALEL MAKiNE

Baz: uygulamalar icin yuksek perforrnansh bilgisayarlara ihtiyac vardrr. Buuygulamalar gercek zarnanh kontrol sistemleri, gercek zaman kriterlerine esnekancak yine de zamana karst hassas olan slrnulatorler veya icrasi interaktif olmadrg:halde cok uzun suren hesaplarna agirhkh prograrnlardir. Bu calismada bu turihtiyaclara cevap verebilecek, dagrtrk hesaplama mirnarisi alnnda, bir isletirnsisternine ek olarak tasarlanrrus sanal bir paralel bilgisayar (PVM: Parallel VirtualMachine) ve grafiksel arayuzu XPVM tamulrmsnr. PVM, birbirlerine ag ile bagh.farkh mimari ve islerirn sistemleri ile cahsan dagiruk halde duran bilgisayarlan biramac icin ve ayrn anda cahsnrrnaya yarayan yazihrn pakedidir. PVM ile sisternekanlan bilgisayarlar gorunurde birer i~ istasyonu ya da kisisel bilgisayar olsalar da,sistem sonunda sanal bir paralel makineye donusmektedir. Dolayrsiyla PVM.oldukca pahah yuksek perforrnansh bilgisayar sistemleri alamayan iiniversiteler vearasnrrna kurumlan icin ekonornik bir cozurndur.

PARALLEL ViRTUAL MACHiNE (PVM)

Some applications require high performance computing environment. Theseapplications can be real-time control systems, soft real-time simulators or longrunning non-interactive compute bound tasks. In this paper, a parallel virtualmachine (PVM) and its graphical interface (XPVM) have been introduced to meethigh performance computing demands. PVM is a software package designed as alibrary extension to an operating system. A virtual parallel system can easily bebuild from networked computers using PVM. These computers can also beheterogeneous according to their architecture and operating systems. Therefore,PVM makes it possible to build affordable high performance computers foruniversities and research companies using ordinary workstations and personalcomputers.

___

  • [1] Geist A., Beguclin A., Dongarra J., Jiang W., Manchek R., Sunderman V. : PVM: Parallel Virtual Machine- A User's Guide and Tutorial
  • [2] Griffioen J., Yavatkar R., Finkel R., Unify: A Scalable, Loosely- Coupled, Distributed Shared Memory Multicomputer, University of Kentucky, Computer Science Dept. Technical Report, Ocak 1993.
  • [3] Manchek B., Heterogeneous Distributed Computing, 1995.
  • [4] Manchek R., "Workstation Solutions: Opening the Door to Heterogeneous Network Supercomputing", Supercomputing Review Volume 4, Number 9, Utah Department of Mathematics - University of Utah, USA, September, 1991, pp 44-46.
  • [5] Ozrnen A, Durmus B, "Dagink Sistemlerde Paralel Performans ve Karmasikhk Analizi", 1. Ulusal Yiiksek Perforrnansh Bilisim Sempozyumu, Gebze-Kocaeli, 24-25 Ekim 2002, pp 45-49.