Repository navigation
JIT: (bug) floating-point vector ~v + 1 is folded to -v #133522
Description
Repro
using System;
using System.Runtime.CompilerServices;
using System.Runtime.Intrinsics;
class Program {
[MethodImpl(MethodImplOptions.NoInlining | MethodImplOptions.AggressiveOptimization)]
static Vector128<float> Test(Vector128<float> v) => ~v + Vector128.Create(1.0f);
static void Main() {
Console.WriteLine(Test(Vector128.Create(1.0f)).AsUInt32().GetElement(0).ToString("X8"));
}
}Expected
C03FFFFF
~1.0f is 0xC07FFFFF; adding 1.0f gives 0xC03FFFFF.
Actual
BF800000
That is -1.0f, i.e. the input was negated instead.
Platform
Windows x64, .NET 11.0.0-rc.1.26425.128. Also reproduces with Vector128<double>: expected C007FFFFFFFFFFFF, actual BFF0000000000000.
Activity
dotnet-policy-service commentedon Sep 9, 2026
Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch
See info in area-owners.md if you want to be subscribed.
[-]JIT: floating-point vector `~v + 1` is folded to `-v`[/-][+]JIT: (bug) floating-point vector `~v + 1` is folded to `-v`[/+]Fix vector complement-add folding for floating-point types (#133539)
EgorBo commentedon Sep 18, 2026
Affected SDK versions: