Blocking for Cache (tiling)

• Blocking for cache:

  • An optimization that applies to data sets that do not fit into the (2nd level) data cache
  • A way to increase spatial locality of reference (i.e. exploit full cache lines)
  • A way to increase temporal locality of reference (i.e. to improve data re-use)
  • It is beneficial mostly with multi-dimensional arrays

-LNO:blocking=[on|off] (default on)

-LNO:blocking_size=n1,n2 (for L1 and L2)

By default L1=32KB and L2=1MB

use -LNO:cs2=8M to specify the 8MB L2 cache

Previous slide

Next slide

Back to first slide

View graphic version