[llvm] be37f13 - [SLP]Add an extra test for vectorization of non-pow-2 trees, NFC.

[Alexey Bataev via llvm-commits]

Author: Alexey Bataev
Date: 2020-07-22T09:13:30-04:00
New Revision: be37f13e2ddeeb191264c03295aa8925f2ce2452

URL: https://github.com/llvm/llvm-project/commit/be37f13e2ddeeb191264c03295aa8925f2ce2452
DIFF: https://github.com/llvm/llvm-project/commit/be37f13e2ddeeb191264c03295aa8925f2ce2452.diff

LOG: [SLP]Add an extra test for vectorization of non-pow-2 trees, NFC.

Added: 
    llvm/test/Transforms/SLPVectorizer/X86/geps-non-pow-2.ll

Modified: 
    

Removed: 
    


################################################################################
diff  --git a/llvm/test/Transforms/SLPVectorizer/X86/geps-non-pow-2.ll b/llvm/test/Transforms/SLPVectorizer/X86/geps-non-pow-2.ll
new file mode 100644
index 000000000000..596543880d36
--- /dev/null
+++ b/llvm/test/Transforms/SLPVectorizer/X86/geps-non-pow-2.ll
@@ -0,0 +1,94 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt -slp-vectorizer -S -o - -mtriple=x86_64-unknown-linux -mcpu=haswell < %s | FileCheck %s
+ at e = dso_local local_unnamed_addr global i32 0, align 4
+ at f = dso_local local_unnamed_addr global i32 0, align 4
+
+; Function Attrs: nofree norecurse nounwind uwtable
+define dso_local i32 @g() local_unnamed_addr {
+; CHECK-LABEL: @g(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = load i32, i32* @e, align 4
+; CHECK-NEXT:    [[TOBOOL_NOT19:%.*]] = icmp eq i32 [[TMP0]], 0
+; CHECK-NEXT:    br i1 [[TOBOOL_NOT19]], label [[WHILE_END:%.*]], label [[WHILE_BODY:%.*]]
+; CHECK:       while.body:
+; CHECK-NEXT:    [[C_022:%.*]] = phi i32* [ [[C_022_BE:%.*]], [[WHILE_BODY_BACKEDGE:%.*]] ], [ undef, [[ENTRY:%.*]] ]
+; CHECK-NEXT:    [[TMP1:%.*]] = phi <2 x i32*> [ [[TMP14:%.*]], [[WHILE_BODY_BACKEDGE]] ], [ undef, [[ENTRY]] ]
+; CHECK-NEXT:    [[INCDEC_PTR:%.*]] = getelementptr inbounds i32, i32* [[C_022]], i64 1
+; CHECK-NEXT:    [[TMP2:%.*]] = ptrtoint i32* [[C_022]] to i64
+; CHECK-NEXT:    [[TMP3:%.*]] = trunc i64 [[TMP2]] to i32
+; CHECK-NEXT:    [[TMP4:%.*]] = getelementptr i32, <2 x i32*> [[TMP1]], <2 x i64> <i64 1, i64 1>
+; CHECK-NEXT:    switch i32 [[TMP3]], label [[WHILE_BODY_BACKEDGE]] [
+; CHECK-NEXT:    i32 2, label [[SW_BB:%.*]]
+; CHECK-NEXT:    i32 4, label [[SW_BB6:%.*]]
+; CHECK-NEXT:    ]
+; CHECK:       sw.bb:
+; CHECK-NEXT:    [[TMP5:%.*]] = extractelement <2 x i32*> [[TMP4]], i32 0
+; CHECK-NEXT:    [[TMP6:%.*]] = ptrtoint i32* [[TMP5]] to i64
+; CHECK-NEXT:    [[TMP7:%.*]] = trunc i64 [[TMP6]] to i32
+; CHECK-NEXT:    [[TMP8:%.*]] = getelementptr i32, <2 x i32*> [[TMP1]], <2 x i64> <i64 2, i64 2>
+; CHECK-NEXT:    [[TMP9:%.*]] = extractelement <2 x i32*> [[TMP4]], i32 1
+; CHECK-NEXT:    store i32 [[TMP7]], i32* [[TMP9]], align 4
+; CHECK-NEXT:    [[INCDEC_PTR5:%.*]] = getelementptr inbounds i32, i32* [[C_022]], i64 2
+; CHECK-NEXT:    br label [[WHILE_BODY_BACKEDGE]]
+; CHECK:       sw.bb6:
+; CHECK-NEXT:    [[INCDEC_PTR8:%.*]] = getelementptr inbounds i32, i32* [[C_022]], i64 2
+; CHECK-NEXT:    [[TMP10:%.*]] = ptrtoint i32* [[INCDEC_PTR]] to i64
+; CHECK-NEXT:    [[TMP11:%.*]] = trunc i64 [[TMP10]] to i32
+; CHECK-NEXT:    [[TMP12:%.*]] = getelementptr i32, <2 x i32*> [[TMP1]], <2 x i64> <i64 2, i64 2>
+; CHECK-NEXT:    [[TMP13:%.*]] = extractelement <2 x i32*> [[TMP4]], i32 0
+; CHECK-NEXT:    store i32 [[TMP11]], i32* [[TMP13]], align 4
+; CHECK-NEXT:    br label [[WHILE_BODY_BACKEDGE]]
+; CHECK:       while.body.backedge:
+; CHECK-NEXT:    [[C_022_BE]] = phi i32* [ [[INCDEC_PTR]], [[WHILE_BODY]] ], [ [[INCDEC_PTR8]], [[SW_BB6]] ], [ [[INCDEC_PTR5]], [[SW_BB]] ]
+; CHECK-NEXT:    [[TMP14]] = phi <2 x i32*> [ [[TMP4]], [[WHILE_BODY]] ], [ [[TMP12]], [[SW_BB6]] ], [ [[TMP8]], [[SW_BB]] ]
+; CHECK-NEXT:    br label [[WHILE_BODY]]
+; CHECK:       while.end:
+; CHECK-NEXT:    ret i32 undef
+;
+entry:
+  %0 = load i32, i32* @e, align 4
+  %tobool.not19 = icmp eq i32 %0, 0
+  br i1 %tobool.not19, label %while.end, label %while.body
+
+while.body:                                       ; preds = %entry, %while.body.backedge
+  %c.022 = phi i32* [ %c.022.be, %while.body.backedge ], [ undef, %entry ]
+  %b.021 = phi i32* [ %b.021.be, %while.body.backedge ], [ undef, %entry ]
+  %a.020 = phi i32* [ %a.020.be, %while.body.backedge ], [ undef, %entry ]
+  %incdec.ptr = getelementptr inbounds i32, i32* %c.022, i64 1
+  %1 = ptrtoint i32* %c.022 to i64
+  %2 = trunc i64 %1 to i32
+  %incdec.ptr1 = getelementptr inbounds i32, i32* %a.020, i64 1
+  %incdec.ptr2 = getelementptr inbounds i32, i32* %b.021, i64 1
+  switch i32 %2, label %while.body.backedge [
+  i32 2, label %sw.bb
+  i32 4, label %sw.bb6
+  ]
+
+sw.bb:                                            ; preds = %while.body
+  %incdec.ptr3 = getelementptr inbounds i32, i32* %b.021, i64 2
+  %3 = ptrtoint i32* %incdec.ptr2 to i64
+  %4 = trunc i64 %3 to i32
+  %incdec.ptr4 = getelementptr inbounds i32, i32* %a.020, i64 2
+  store i32 %4, i32* %incdec.ptr1, align 4
+  %incdec.ptr5 = getelementptr inbounds i32, i32* %c.022, i64 2
+  br label %while.body.backedge
+
+sw.bb6:                                           ; preds = %while.body
+  %incdec.ptr7 = getelementptr inbounds i32, i32* %a.020, i64 2
+  %incdec.ptr8 = getelementptr inbounds i32, i32* %c.022, i64 2
+  %5 = ptrtoint i32* %incdec.ptr to i64
+  %6 = trunc i64 %5 to i32
+  %incdec.ptr9 = getelementptr inbounds i32, i32* %b.021, i64 2
+  store i32 %6, i32* %incdec.ptr2, align 4
+  br label %while.body.backedge
+
+while.body.backedge:                              ; preds = %sw.bb6, %while.body, %sw.bb
+  %c.022.be = phi i32* [ %incdec.ptr, %while.body ], [ %incdec.ptr8, %sw.bb6 ], [ %incdec.ptr5, %sw.bb ]
+  %b.021.be = phi i32* [ %incdec.ptr2, %while.body ], [ %incdec.ptr9, %sw.bb6 ], [ %incdec.ptr3, %sw.bb ]
+  %a.020.be = phi i32* [ %incdec.ptr1, %while.body ], [ %incdec.ptr7, %sw.bb6 ], [ %incdec.ptr4, %sw.bb ]
+  br label %while.body
+
+while.end:                                        ; preds = %entry
+  ret i32 undef
+}
+