================ @@ -188,3 +188,87 @@ loop.i.latch: exit: ret void } + +; for (i = 0; i < 10; i++) +; for (j = 0; j < 10; j++) +; A[100*i + j] = B[100*i + j] + C[i + 100*j] + C[i + 99*j] + C[i + 98*j]; +; +; The above loop should NOT be interchanged in terms of locality of reference. ---------------- sjoerdmeijer wrote:
Nit: I know that this is part of a patch set, and some of the other patches talk about the heuristics, but it would be good to make this a little bit more self-contained. I.e., instead of 3 unfriendly memory accesses there is only 1? And maybe a quick mention why the 100? https://github.com/llvm/llvm-project/pull/193476 _______________________________________________ llvm-branch-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
