Parallel Wiedemann

Iterative Algorithm

• Loop :

  • vi+1 = A vi
  • si+1 = <u, vi+1>
  • update of the generating polynomial with s0, …, si+1

Desynchronization :

• Block cutting of the matrix

• Several products in a row

• Si are used by groups

• Induces some Parallelism

• Matrix-vector product is dominant

Diapositive précédente

Diapositive suivante

Revenir à la première diapositive

Afficher la version graphique