[llvm] [LoopInterchange] Add tests for the vectorization profitability (NFC) (PR #133665)

[Madhur Amilkanthwar via llvm-commits]

================
@@ -0,0 +1,108 @@
+; RUN: opt < %s -passes=loop-interchange -cache-line-size=64 \
+; RUN:     -pass-remarks-output=%t -disable-output -loop-interchange-profitabilities=vectorize
+; RUN: FileCheck -input-file %t %s
+
+ at A = dso_local global [256 x [256 x float]] zeroinitializer
+ at B = dso_local global [256 x [256 x float]] zeroinitializer
+ at C = dso_local global [256 x [256 x float]] zeroinitializer
+
+; Check that the below loops are exchanged for vectorization.
+;
+; for (int i = 0; i < 256; i++) {
+;   for (int j = 1; j < 256; j++) {
+;     A[i][j] = A[i][j-1] + B[i][j];
+;     C[i][j] += 1;
+;   }
+; }
+;
+; FIXME: These loops are not exchanged at this time due to the problem of
+; profitablity heuristic for vectorization.
+
+; CHECK:      --- !Missed
+; CHECK-NEXT: Pass:            loop-interchange
+; CHECK-NEXT: Name:            InterchangeNotProfitable
+; CHECK-NEXT: Function:        interchange_necesasry_for_vectorization
+; CHECK-NEXT: Args:
+; CHECK-NEXT:   - String:          Interchanging loops is not considered to improve cache locality nor vectorization.
+; CHECK-NEXT: ...
+define void @interchange_necesasry_for_vectorization() {
+entry:
+  br label %for.i.header
+
+for.i.header:
+  %i = phi i64 [ 1, %entry ], [ %i.next, %for.i.inc ]
+  br label %for.j.body
+
+for.j.body:
+  %j = phi i64 [ 1, %for.i.header ], [ %j.next, %for.j.body ]
+  %j.dec = add nsw i64 %j, -1
+  %a.load.index = getelementptr nuw inbounds [256 x [256 x float]], ptr @A, i64 %i, i64 %j.dec
+  %b.index = getelementptr nuw inbounds [256 x [256 x float]], ptr @B, i64 %i, i64 %j
+  %c.index = getelementptr nuw inbounds [256 x [256 x float]], ptr @C, i64 %i, i64 %j
+  %a = load float, ptr %a.load.index, align 4
+  %b = load float, ptr %b.index, align 4
+  %c = load float, ptr %c.index, align 4
+  %add.0 = fadd float %a, %b
+  %a.store.index = getelementptr nuw inbounds [256 x [256 x float]], ptr @A, i64 %i, i64 %j
+  store float %add.0, ptr %a.store.index, align 4
+  %add.1 = fadd float %c, 1.0
+  store float %add.1, ptr %c.index, align 4
+  %j.next = add nuw nsw i64 %j, 1
+  %cmp.j = icmp eq i64 %j.next, 256
+  br i1 %cmp.j, label %for.i.inc, label %for.j.body
+
+for.i.inc:
+  %i.next = add nuw nsw i64 %i, 1
+  %cmp.i = icmp eq i64 %i.next, 256
+  br i1 %cmp.i, label %exit, label %for.i.header
+
+exit:
+  ret void
+}
+
+; Check that the following innermost loop can be vectorized so that
+; interchangig is unnecessary.
----------------
madhur13490 wrote:

Typo: interchangig

https://github.com/llvm/llvm-project/pull/133665