Repository navigation
JIT: (bug) complementing a scalar SIMD comparison leaves upper lanes unchanged #133521
Description
Repro
using System;
using System.Runtime.CompilerServices;
using System.Runtime.Intrinsics;
using System.Runtime.Intrinsics.X86;
class Program {
[MethodImpl(MethodImplOptions.NoInlining | MethodImplOptions.AggressiveOptimization)]
static Vector128<float> Test(Vector128<float> a, Vector128<float> b) =>
~Sse.CompareScalarEqual(a, b);
static void Main() {
if (!Sse.IsSupported)
throw new PlatformNotSupportedException("SSE is required.");
Vector128<float> value = Vector128.Create(1.0f, 2.0f, 3.0f, 4.0f);
Console.WriteLine(Test(value, value).AsUInt32().GetElement(1).ToString("X8"));
}
}Expected
BFFFFFFF
CompareScalarEqual copies lane 1 from the first input (2.0f = 0x40000000); the full-vector ~ must complement it.
Actual
40000000
Platform
Windows x64, .NET 11.0.0-rc.1.26425.128. Also reproduces with Avx.CompareScalar.
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: complementing a scalar SIMD comparison leaves upper lanes unchanged[/-][+]JIT: (bug) complementing a scalar SIMD comparison leaves upper lanes unchanged[/+]Fix SIMD complement and scalar FMA folding (#133542)
EgorBo commentedon Sep 18, 2026
Affected SDK versions: