Matrix Multiplication: All 12 Data/Loop Combinations

Exploring how Transposing B affects every permutation of the triple-nested loops.

Locality Comparison: Standard vs Transposed B

Inner Loop Standard B (Miss Rate) Transposed B (Miss Rate) Result of Transpose
k (ijk, jik) 1.25 (B is Col-wise) 0.50 (Bᵀ is Row-wise) 🎉 Improves
j (kij, ikj) 0.50 (B is Row-wise) 1.25 (Bᵀ is Col-wise) ⚠️ Worsens
i (jki, kji) 2.00 (B is Fixed) 2.00 (Bᵀ is Fixed) No Change

Simulation Control

Iteration: 0 / 63

Live Stride Analysis & Cache Stats

-
A Access
Row-wise
Miss Rate: 0.25
B Access
Column-wise
Miss Rate: 1.0
C Access
Fixed
Miss Rate: 0.0
Avg Misses / Iteration -
Step Counter
1 / 64
Current Step Misses
0
None
Total Physical Misses
0
A
×
B
=
C
i = 0j = 0k = 0
Iteration Trace
Stride Analysis
[ Waiting for selection... ]