Fastest way to transpose 4x4 byte matrix

Question

I have a 4x4 block of bytes that I\'d like to transpose using general purpose hardware. In other words, for bytes A-P, I\'m looking for the most efficient (in terms of numbe

Answer 1

An efficient solution is possible on a 64 bits machine, if you accept that. First shift the 32 bits integer constants by (0,) 1, 2 and 3 bytes respectively [3 shitfs]. Then mask out the unwanted bits and perform logical ORs [12 ANDs with a constant, 12 ORs]. Finally, shift back to 32 bits [3 shifts] and read out the 32 bits.

ABCD
EFGH
IJKL
MNOP

ABCD
 EFGH
  IJKL
   MNOP

A---
 E---
  I---
   MNOP
=======
AEIMNOP
AEIM

AB--
 -F--
  -J--
   -NOP
=======
ABFJNOP
BFJN

ABC-
 --G-
  --K-
   --OP
=======
ABCGKOP
CGKO

ABCD
 ---H
  ---L
   ---P
=======
ABCDHLP
DHLP