[llvm] [LV][NFC] Make ARM/mve-selectandorcost.ll test more robust (PR #197388)

[via llvm-commits]

llvmorg-github-actions[bot] wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-llvm-transforms

Author: David Sherwood (david-arm)

<details>
<summary>Changes</summary>

The test currently has an fdiv in the loop, which leads to vector costs that are almost identical to the scalar cost. This makes the test fragile and future cost model work for VPDerivedIVRecipe will cause the test to favour the scalar version. I've changed the fdiv to fmul to make the test more robust.

---
Full diff: https://github.com/llvm/llvm-project/pull/197388.diff


1 Files Affected:

- (modified) llvm/test/Transforms/LoopVectorize/ARM/mve-selectandorcost.ll (+3-3) 


``````````diff
diff --git a/llvm/test/Transforms/LoopVectorize/ARM/mve-selectandorcost.ll b/llvm/test/Transforms/LoopVectorize/ARM/mve-selectandorcost.ll
index abc77b26d5053..917dae8ff2de1 100644
--- a/llvm/test/Transforms/LoopVectorize/ARM/mve-selectandorcost.ll
+++ b/llvm/test/Transforms/LoopVectorize/ARM/mve-selectandorcost.ll
@@ -43,7 +43,7 @@ define float @test(ptr nocapture readonly %pA, ptr nocapture readonly %pB, i32 %
 ; CHECK-NEXT:    [[TMP10:%.*]] = fadd fast <4 x float> [[TMP9]], [[TMP8]]
 ; CHECK-NEXT:    [[TMP11:%.*]] = fsub fast <4 x float> [[WIDE_LOAD]], [[WIDE_LOAD3]]
 ; CHECK-NEXT:    [[TMP12:%.*]] = call fast <4 x float> @llvm.fabs.v4f32(<4 x float> [[TMP11]])
-; CHECK-NEXT:    [[TMP13:%.*]] = fdiv fast <4 x float> [[TMP12]], [[TMP10]]
+; CHECK-NEXT:    [[TMP13:%.*]] = fmul fast <4 x float> [[TMP12]], [[TMP10]]
 ; CHECK-NEXT:    [[TMP14:%.*]] = select ninf <4 x i1> [[TMP7]], <4 x float> [[TMP13]], <4 x float> splat (float -0.000000e+00)
 ; CHECK-NEXT:    [[PREDPHI]] = fadd reassoc arcp contract afn <4 x float> [[VEC_PHI]], [[TMP14]]
 ; CHECK-NEXT:    [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4
@@ -78,7 +78,7 @@ define float @test(ptr nocapture readonly %pA, ptr nocapture readonly %pB, i32 %
 ; CHECK-NEXT:    [[ADD:%.*]] = fadd fast float [[TMP20]], [[TMP19]]
 ; CHECK-NEXT:    [[SUB:%.*]] = fsub fast float [[TMP17]], [[TMP18]]
 ; CHECK-NEXT:    [[TMP21:%.*]] = tail call fast float @llvm.fabs.f32(float [[SUB]])
-; CHECK-NEXT:    [[DIV:%.*]] = fdiv fast float [[TMP21]], [[ADD]]
+; CHECK-NEXT:    [[DIV:%.*]] = fmul fast float [[TMP21]], [[ADD]]
 ; CHECK-NEXT:    [[ADD4:%.*]] = fadd fast float [[DIV]], [[ACCUM_017]]
 ; CHECK-NEXT:    br label %[[IF_END]]
 ; CHECK:       [[IF_END]]:
@@ -117,7 +117,7 @@ if.then:
   %add = fadd fast float %3, %2
   %sub = fsub fast float %0, %1
   %4 = tail call fast float @llvm.fabs.f32(float %sub)
-  %div = fdiv fast float %4, %add
+  %div = fmul fast float %4, %add
   %add4 = fadd fast float %div, %accum.017
   br label %if.end
 

``````````

</details>


https://github.com/llvm/llvm-project/pull/197388