[llvm] [LoopVectorize] Refine runtime memory check costs when there is an outer loop (PR #76034)

[Florian Hahn via llvm-commits]

================
@@ -0,0 +1,193 @@
+; RUN: opt -p loop-vectorize -mattr=+sve -debug-only=loop-vectorize -S -disable-output < %s 2>&1 | FileCheck %s
+
+target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
+target triple = "aarch64-unknown-linux-gnu"
+
+
+define void @outer_no_tc(ptr nocapture noundef %a, ptr nocapture noundef readonly %b, i64 noundef %m, i64 noundef %n) vscale_range(1,16) {
+; CHECK-LABEL: LV: Checking a loop in 'outer_no_tc'
+; CHECK:      Calculating cost of runtime checks:
+; CHECK:      Total cost of runtime checks: 3
+; CHECK-NEXT: LV: Minimum required TC for runtime checks to be profitable:32
+entry:
+  %cmp24 = icmp sgt i64 %m, 0
+  %cmp222 = icmp sgt i64 %n, 0
+  %or.cond = and i1 %cmp24, %cmp222
+  br i1 %or.cond, label %for.cond1.preheader.us, label %for.cond.cleanup
+
+for.cond1.preheader.us:
----------------
fhahn wrote:

Would be good to clean up the block and value names to improve readability of the tests

https://github.com/llvm/llvm-project/pull/76034