Go backward to SIMD Mesh Matrix Multiplication Go up to Top Go forward to SIMD Mesh Matrix Multiplication

SIMD Mesh Matrix Multiplication

1. Precondition array
   • Shift row i by i-1 elements west,
   • Shift column j by j-1 elements north.
2. Multiply and add

   On processor <i,j >:

   c = SUM_k a_ik*b_kj

• Inverted dimensions
  • Matrix V i, ->j.
  • Processor array V iyproc, ->ixproc.
• 4n shift and n arithmetic operations.
• n² processors.

Maspar program: see slide.