[llvm] d204fdc - [LV] Add countable multi-exit test to vec.stats.ll

[Florian Hahn via llvm-commits]

Author: Florian Hahn
Date: 2025-08-01T16:39:28+01:00
New Revision: d204fdcc231667090a327c444612ff0798f987e1

URL: https://github.com/llvm/llvm-project/commit/d204fdcc231667090a327c444612ff0798f987e1
DIFF: https://github.com/llvm/llvm-project/commit/d204fdcc231667090a327c444612ff0798f987e1.diff

LOG: [LV] Add countable multi-exit test to vec.stats.ll

Currently the multi-exit loops with all countable exits are considered
vectorized with early exit, while that terminology is used for loops we
performed early-exit vectorization, instead of requring a scalar
epilogue.

Added: 
    

Modified: 
    llvm/test/Transforms/LoopVectorize/vect.stats.ll

Removed: 
    


################################################################################
diff  --git a/llvm/test/Transforms/LoopVectorize/vect.stats.ll b/llvm/test/Transforms/LoopVectorize/vect.stats.ll
index 018e2c213ddf2..033907e079a7d 100644
--- a/llvm/test/Transforms/LoopVectorize/vect.stats.ll
+++ b/llvm/test/Transforms/LoopVectorize/vect.stats.ll
@@ -4,9 +4,9 @@
 ; We have 3 loops, two of them are vectorizable (with one being early-exit
 ; vectorized) and the third one is not.
 
-; CHECK: 3 loop-vectorize               - Number of loops analyzed for vectorization
-; CHECK: 1 loop-vectorize               - Number of early exit loops vectorized
-; CHECK: 2 loop-vectorize               - Number of loops vectorized
+; CHECK: 4 loop-vectorize               - Number of loops analyzed for vectorization
+; CHECK: 2 loop-vectorize               - Number of early exit loops vectorized
+; CHECK: 3 loop-vectorize               - Number of loops vectorized
 
 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
 
@@ -15,19 +15,19 @@ entry:
   %cmp1 = icmp sle i64 %size, 0
   %cmp21 = icmp sgt i64 0, %size
   %or.cond = or i1 %cmp1, %cmp21
-  br i1 %or.cond, label %for.end, label %for.body
+  br i1 %or.cond, label %exit, label %loop
 
-for.body:                                         ; preds = %entry, %for.body
-  %indvars.iv2 = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ]
-  %arrayidx = getelementptr inbounds float, ptr %a, i64 %indvars.iv2
+loop:
+  %iv = phi i64 [ %iv.next, %loop ], [ 0, %entry ]
+  %arrayidx = getelementptr inbounds float, ptr %a, i64 %iv
   %0 = load float, ptr %arrayidx, align 4
   %mul = fmul float %0, %0
   store float %mul, ptr %arrayidx, align 4
-  %indvars.iv.next = add nuw nsw i64 %indvars.iv2, 1
-  %cmp2 = icmp sgt i64 %indvars.iv.next, %size
-  br i1 %cmp2, label %for.end, label %for.body
+  %iv.next = add nuw nsw i64 %iv, 1
+  %cmp2 = icmp sgt i64 %iv.next, %size
+  br i1 %cmp2, label %exit, label %loop
 
-for.end:                                          ; preds = %entry, %for.body
+exit:                                          ; preds = %entry, %loop
   ret void
 }
 
@@ -38,9 +38,9 @@ entry:
   call void @init_mem(ptr %p1, i64 1024)
   call void @init_mem(ptr %p2, i64 1024)
   %end.clamped = and i64 %end, 1023
-  br label %for.body
+  br label %loop
 
-for.body:
+loop:
   %ind = phi i64 [ %ind.next, %for.inc ], [ 0, %entry ]
   %arrayidx1 = getelementptr inbounds i32, ptr %p1, i64 %ind
   %0 = load i32, ptr %arrayidx1, align 4
@@ -52,7 +52,7 @@ for.body:
 for.inc:
   %ind.next = add i64 %ind, 1
   %cmp = icmp ult i64 %ind.next, %end.clamped
-  br i1 %cmp, label %for.body, label %exit
+  br i1 %cmp, label %loop, label %exit
 
 found:
   ret i32 1
@@ -66,25 +66,50 @@ entry:
   %cmp1 = icmp sle i64 %size, 0
   %cmp21 = icmp sgt i64 0, %size
   %or.cond = or i1 %cmp1, %cmp21
-  br i1 %or.cond, label %for.end, label %for.body
+  br i1 %or.cond, label %exit, label %loop
 
-for.body:                                         ; preds = %entry, %for.body
-  %indvars.iv2 = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ]
-  %0 = add nsw i64 %indvars.iv2, -5
+loop:
+  %iv = phi i64 [ %iv.next, %loop ], [ 0, %entry ]
+  %0 = add nsw i64 %iv, -5
   %arrayidx = getelementptr inbounds float, ptr %a, i64 %0
   %1 = load float, ptr %arrayidx, align 4
-  %2 = add nsw i64 %indvars.iv2, 2
+  %2 = add nsw i64 %iv, 2
   %arrayidx2 = getelementptr inbounds float, ptr %a, i64 %2
   %3 = load float, ptr %arrayidx2, align 4
   %mul = fmul float %1, %3
-  %arrayidx4 = getelementptr inbounds float, ptr %a, i64 %indvars.iv2
+  %arrayidx4 = getelementptr inbounds float, ptr %a, i64 %iv
   store float %mul, ptr %arrayidx4, align 4
-  %indvars.iv.next = add nuw nsw i64 %indvars.iv2, 1
-  %cmp2 = icmp sgt i64 %indvars.iv.next, %size
-  br i1 %cmp2, label %for.end, label %for.body
+  %iv.next = add nuw nsw i64 %iv, 1
+  %cmp2 = icmp sgt i64 %iv.next, %size
+  br i1 %cmp2, label %exit, label %loop
 
-for.end:                                          ; preds = %entry, %for.body
+exit:
   ret void
 }
 
+define i1 @multiple_countable_exits_multiple_exit_block(ptr %A, ptr %B, i32 %N) {
+entry:
+  br label %loop.header
+
+loop.header:
+  %iv = phi i32 [ 0, %entry ], [ %iv.next, %loop.latch ]
+  %cond.0 = icmp eq i32 %iv, %N
+  br i1 %cond.0, label %exit.0, label %loop.latch
+
+loop.latch:
+  %A.gep = getelementptr inbounds i32, ptr %A, i32 %iv
+  %lv = load i32, ptr %A.gep, align 4
+  %B.gep = getelementptr inbounds i32, ptr %B, i32 %iv
+  store i32 %lv, ptr %B.gep, align 4
+  %iv.next = add nuw i32 %iv, 1
+  %cond.1 = icmp ult i32 %iv.next, 1000
+  br i1 %cond.1, label %loop.header, label %exit.1
+
+exit.0:
+  ret i1 false
+
+exit.1:
+  ret i1 true
+}
+
 declare void @init_mem(ptr, i64);