[llvm] [VPlan] Add VPValue for VF, use it for VPWidenIntOrFpInductionRecipe. (PR #95305)

[David Sherwood via llvm-commits]

================
@@ -73,7 +74,8 @@ define void @induction_i3_zext(ptr %dst) #0 {
 ; CHECK:       vector.ph:
 ; CHECK:         %ind.end = trunc i64 %n.vec to i3
 ; CHECK-NEXT:    [[TMP4:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT:    [[TMP5:%.*]] = mul i64 [[TMP4]], 4
+; CHECK-NEXT:    [[TMP40:%.*]] = mul i64 [[TMP4]], 2
+; CHECK-NEXT:    [[TMP5:%.*]] = mul i64 [[TMP40]], 2
----------------
david-arm wrote:

This is interesting. I guess I was expecting that since this patch only creates a runtime VF on demand that total lines of IR would only decrease, rather than increase which is happening here.

https://github.com/llvm/llvm-project/pull/95305