Toom-4.5 (aka Toom-5x4, Toom-6x3, Toom-7x2)

[Torbjorn Granlund]

nisse at lysator.liu.se (Niels Möller) writes:

  bodrato at mail.dm.unipi.it writes:
  
  > Yes, they can... evaluating in {1,2,3}, instead of {1,2,4}, gives a
  > division by 3 and one by 15; but it needs a trickier evaluation and a
  > spurious 2/3 of a division by 3...
  
  Evaluation in 3 would use some shifting and adding rather than
  multiplies, but I imagine it's going to be more costly than evaluation
  in 4?
  
Evaluation in 3 could make use of mpn_addlsh1 for multiplying by 3.  It
is faster on some machines and slower on some machines than mpn_lshift.

-- 
Torbjörn