Home Page Syllabus Marcin Paprzycki


      SC 730 PARALLEL ALGORITHMS

      Material covered:

      Introduction to high-performance computing (based on: J.J. Dongarra, I.S. Duff, D.C. Sorensen, H.A. van der Vorst, Numerical Linear Algebra for High-Performance Computers, pages 1-69)

      Trends in computer design

      Traditional computers
        parallelism within a single processor
          multiple functional units
          pipelining
          RISC
          VLIW
          vector processing
          chaining
          memory "management" inside the processor
        memory organization
          main memory
          cache memory
          main memory
        memory management
      Parallel computers
        Flynt taxonomy
        processor arrays
        shared memory computers and their limitations
        message passing
          interconnection topologies
          virtual shared memory
          cluster computers


      Sorting
      designing parallel merge-sort

      HOMEWORK 1. Implement parallel merge-sort and present the performance data. Due by Thursday, March 4th.