Vectorization on NEC supercomputers

14.01.02


Start Lecture by clicking here


Content

 Author: Dr. M. Galle

  1. Vectorization on NEC supercomputers
  2.     Copyright
  3.     Overview
  4.     NEC SX-5
  5.     The Specifications
  6.     Vector Unit Architecture
  7.     Scalar Architecture
  8.     SX-5 Scalar Unit Block Diagram
  9.     IEEE Format (float0) !!!
  10. Vectorization
  11.     Segmentation, Pipelining
  12.     Segmentation, Pipelining (2)
  13.     Segmentation, Pipelining (3)
  14.     Segmentation, Pipelining (4)
  15.     Data Parallelism
  16. Vectorization examples
  17.     ex. 1: v = s + v
  18.     ex. 1: v = s + v (cont.)
  19.     Chaining
  20.     ex. 1: v = s + v (cont.)
  21.     ex. 1: v = s + v (cont.)
  22.     ex. 2: v = v + v
  23.     ex. 2: v = v + v (cont.)
  24.     ex. 3: v = v + s * v
  25.     ex. 3: v = v + s * v (cont.)
  26.     ex. 4: s = s + v * v
  27.     ex. 4: s = s + v * v (cont.)
  28.     Startup and Short Vectors
  29.     Measurements (SX-4)
  30.     vectorization of if-blocks
  31.     vectorization of if-blocks
  32.     vectorization of if-blocks
  33.     vectorization of if-blocks
  34.     Special SX features
  35.     ex. s1: matrix multiply
  36.     ex. s1: matrix multiply (2)
  37.     ex. s1: matrix multiply(3)
  38.     ex. s1: matrix multiply (4)
  39.     Basic Rules for Performance
  40.     FORTRAN 90 !cdir
  41.     Optimization examples
  42.     loop interchange
  43.     loop expansion
  44.     loop division
  45.     call to function
  46.     2D recursion
  47.     2D recursion (2)
  48. indirect addressing
  49.     indirect addressing (2)
  50.     indirect addressing (3)
  51.     indirect addressing (4)
  52.     indirect addressing (5)
  53.     indirect addressing (6)
  54.     indirect addressing (7)
  55.     indirect addressing (8)
  56.     indirect addressing (9)
  57.     indirect addressing (10)
  58.     indirect addressing (11)
  59.     indirect addressing (12)
  60.     indirect addressing example
  61.     indirect addressing example(2)
  62.     indirect addressing example(3)
  63.     indirect addressing example(4)
  64.     indirect addressing example(5)
  65.     vectorization example
  66.     vectorization example (2)
  67.     vectorization example (3)
  68.     vectorization example (4)
  69.     vectorization example (4)
  70. Performance Analysis Tools
  71.     F/C_PROGINF
  72.     prof
  73.     loopprof
  74.     ftrace
  75.     perf-tool

Back to the Parallel Programming Workshop Overview