Theoretical performance?

• The number of operations in the matrix multiply is:

  • 4 Loads (read [1x4] input matrix)
  • 4 Stores (write [1x4] output matrix)
  • 16 Mults (matrix products)
  • 12 Adds (column sums)

• Total ops is 36. Theoretical number of required instruction packets is 36/EUMAX.

  • If EUMAX = 4, then should only take 9 instruction packets!

• You will be graded on how close you get to this limit.

Previous slide

Next slide

Back to first slide

View graphic version