[PATCH] D34078: Enable support for floating-point division reductions

[Yichao Yu via Phabricator via llvm-commits]

yuyichao created this revision.

Similar to fsub, fdiv can also be vectorized using fmul.

Also http://llvm.org/viewvc/llvm-project?view=revision&revision=215200


https://reviews.llvm.org/D34078

Files:
  lib/Transforms/Utils/LoopUtils.cpp
  test/Transforms/LoopVectorize/float-reduction.ll


Index: test/Transforms/LoopVectorize/float-reduction.ll
===================================================================
--- test/Transforms/LoopVectorize/float-reduction.ll
+++ test/Transforms/LoopVectorize/float-reduction.ll
@@ -44,3 +44,25 @@
 for.end:                                          ; preds = %for.body
   ret float %sub
 }
+
+;CHECK-LABEL: @foodiv(
+;CHECK: fdiv fast <4 x float>
+;CHECK: ret
+define float @foodiv(float* nocapture %A, i32* nocapture %n) nounwind uwtable readonly ssp {
+entry:
+  br label %for.body
+
+for.body:                                         ; preds = %for.body, %entry
+  %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
+  %sum.04 = phi float [ 1.000000e+00, %entry ], [ %sub, %for.body ]
+  %arrayidx = getelementptr inbounds float, float* %A, i64 %indvars.iv
+  %0 = load float, float* %arrayidx, align 4
+  %sub = fdiv fast float %sum.04, %0
+  %indvars.iv.next = add i64 %indvars.iv, 1
+  %lftr.wideiv = trunc i64 %indvars.iv.next to i32
+  %exitcond = icmp eq i32 %lftr.wideiv, 200
+  br i1 %exitcond, label %for.end, label %for.body
+
+for.end:                                          ; preds = %for.body
+  ret float %sub
+}
Index: lib/Transforms/Utils/LoopUtils.cpp
===================================================================
--- lib/Transforms/Utils/LoopUtils.cpp
+++ lib/Transforms/Utils/LoopUtils.cpp
@@ -450,6 +450,7 @@
     return InstDesc(Kind == RK_IntegerOr, I);
   case Instruction::Xor:
     return InstDesc(Kind == RK_IntegerXor, I);
+  case Instruction::FDiv:
   case Instruction::FMul:
     return InstDesc(Kind == RK_FloatMult, I, UAI);
   case Instruction::FSub:


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D34078.102111.patch
Type: text/x-patch
Size: 1663 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20170610/e7c43806/attachment.bin>