[llvm] [VPlan] Recursively match operands of interleave group (PR #164839)

[David Sherwood via llvm-commits]

================
@@ -0,0 +1,182 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --version 6
+; RUN: opt %s -passes=loop-vectorize -force-vector-interleave=1 -force-vector-width=4 -enable-interleaved-mem-accesses -S | FileCheck %s
+
+define void @pr128062(ptr %dst.start, i8 %a, i16 %b) {
+; CHECK-LABEL: define void @pr128062(
+; CHECK-SAME: ptr [[DST_START:%.*]], i8 [[A:%.*]], i16 [[B:%.*]]) {
+; CHECK-NEXT:  [[ENTRY:.*:]]
+; CHECK-NEXT:    br label %[[VECTOR_PH:.*]]
+; CHECK:       [[VECTOR_PH]]:
+; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <4 x i16> poison, i16 [[B]], i64 0
+; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <4 x i16> [[BROADCAST_SPLATINSERT]], <4 x i16> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT:    [[BROADCAST_SPLATINSERT1:%.*]] = insertelement <4 x i8> poison, i8 [[A]], i64 0
+; CHECK-NEXT:    [[BROADCAST_SPLAT2:%.*]] = shufflevector <4 x i8> [[BROADCAST_SPLATINSERT1]], <4 x i8> poison, <4 x i32> zeroinitializer
+; CHECK-NEXT:    br label %[[VECTOR_BODY:.*]]
+; CHECK:       [[VECTOR_BODY]]:
+; CHECK-NEXT:    [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ]
+; CHECK-NEXT:    [[OFFSET_IDX:%.*]] = mul i64 [[INDEX]], 4
+; CHECK-NEXT:    [[NEXT_GEP:%.*]] = getelementptr i8, ptr [[DST_START]], i64 [[OFFSET_IDX]]
+; CHECK-NEXT:    [[WIDE_VEC:%.*]] = load <16 x i8>, ptr [[NEXT_GEP]], align 1
----------------
david-arm wrote:

Why are we still performing a wider load here? Perhaps I'm missing something, but it doesn't seem right that we're loading more than we're storing. The original scalar loop has matching loads and stores to the same pointers.

https://github.com/llvm/llvm-project/pull/164839