blocked mm cycles/iteration (for varying matrix sizes)
  n  b    ijk    ikj
 25  25  8.34  7.28 
 50  25  8.07  7.14 
 75  25  7.91  7.03 
100  25  8.25  7.38 
125  25  8.52  7.46 
150  25  8.52  7.57 
175  25  8.64  7.71 
200  25  8.75  7.84 
225  25  8.89  7.92 
250  25  9.02  8.01 
275  25  9.06  8.07 
300  25  9.11  8.10 
325  25  9.14  8.13 
350  25  9.16  8.15 
375  25  9.16  8.15 
400  25  9.18  8.17 

blocked mm cycles/iteration (for varying block sizes)
  n  b    ijk    ikj
400  25  9.17  8.17 
400  50  8.93 10.32 
400  75  8.54 10.26 
400 100 10.02 10.69 
400 125 11.89 11.62 
400 150 13.08 13.21 
400 175 14.62 15.71 
400 200 17.11 21.12 
400 225 17.05 21.06 
400 250 16.92 21.09 
400 275 16.63 21.12 
400 300 16.34 21.41 
400 325 16.30 21.64 
400 350 16.55 22.20 
400 375 17.17 23.35 
400 400 18.09 25.11