diff --git a/src/benchmarks/micro/MicroBenchmarks.csproj b/src/benchmarks/micro/MicroBenchmarks.csproj
index 3d9e9fd4286..867737a3b72 100644
--- a/src/benchmarks/micro/MicroBenchmarks.csproj
+++ b/src/benchmarks/micro/MicroBenchmarks.csproj
@@ -88,6 +88,10 @@
     <PackageReference Include="System.Numerics.Vectors" Version="4.5.0" />
     <PackageReference Include="System.Threading.Tasks.Extensions" Version="4.5.4" />
   </ItemGroup>
+  <ItemGroup Condition="'$(TargetFrameworkIdentifier)' == '.NETCoreApp' And '$(_TargetFrameworkVersionWithoutV)' &gt;= '8.0'">
+    <!-- we want both net8.0 and net9.0 targets to use the 9.0.0 version of the NuGet package (since this is where the generic math APIs were first added) -->
+    <PackageReference Include="System.Numerics.Tensors" Version="$(SystemThreadingChannelsPackageVersion)" />
+  </ItemGroup>
 
   <ItemGroup>
     <ProjectReference Include="..\..\harness\BenchmarkDotNet.Extensions\BenchmarkDotNet.Extensions.csproj" />
@@ -212,9 +216,10 @@
   </ItemGroup>
 
   <ItemGroup Condition=" '$(TargetFrameworkIdentifier)' == '.NETFramework' Or ('$(TargetFrameworkIdentifier)' == '.NETCoreApp' And '$(_TargetFrameworkVersionWithoutV)' &lt; '8.0')">
-	<Compile Remove="libraries\System.Buffers\SearchValuesByteTests.cs" />
-	<Compile Remove="libraries\System.Buffers\SearchValuesCharTests.cs" />
-	<Compile Remove="libraries\System.Text.Encoding\Perf.Ascii.cs" />
+    <Compile Remove="libraries\System.Buffers\SearchValuesByteTests.cs" />
+    <Compile Remove="libraries\System.Buffers\SearchValuesCharTests.cs" />
+    <Compile Remove="libraries\System.Text.Encoding\Perf.Ascii.cs" />
+    <Compile Remove="libraries\System.Numerics.Tensors\*.cs" />
   </ItemGroup>
 
   <!-- This is not removing things from older Net versions, it is removing from newer Net versions -->
diff --git a/src/benchmarks/micro/libraries/System.Numerics.Tensors/Perf_BinaryIntegerTensorPrimitives.cs b/src/benchmarks/micro/libraries/System.Numerics.Tensors/Perf_BinaryIntegerTensorPrimitives.cs
new file mode 100644
index 00000000000..88aa8df5c99
--- /dev/null
+++ b/src/benchmarks/micro/libraries/System.Numerics.Tensors/Perf_BinaryIntegerTensorPrimitives.cs
@@ -0,0 +1,46 @@
+// Licensed to the .NET Foundation under one or more agreements.
+// The .NET Foundation licenses this file to you under the MIT license.
+// See the LICENSE file in the project root for more information.
+
+using BenchmarkDotNet.Attributes;
+using BenchmarkDotNet.Extensions;
+using MicroBenchmarks;
+
+namespace System.Numerics.Tensors.Tests
+{
+    [BenchmarkCategory(Categories.Libraries, Categories.SIMD, Categories.JIT)]
+    [GenericTypeArguments(typeof(int))]
+    public class Perf_BinaryIntegerTensorPrimitives<T>
+        where T : unmanaged, IBinaryInteger<T>
+    {
+        [Params(128, 6 * 512 + 7)]
+        public int BufferLength;
+
+        private T[] _source1;
+        private T[] _destination;
+
+        [GlobalSetup]
+        public void Init()
+        {
+            _source1 = ValuesGenerator.Array<T>(BufferLength, seed: 42);
+            _destination = new T[BufferLength];
+        }
+
+        #region Unary Operations
+        [Benchmark]
+        public void LeadingZeroCount() => TensorPrimitives.LeadingZeroCount<T>(_source1, _destination);
+
+        [Benchmark]
+        public void OnesComplement() => TensorPrimitives.OnesComplement<T>(_source1, _destination);
+
+        [Benchmark]
+        public void PopCount() => TensorPrimitives.PopCount<T>(_source1, _destination);
+
+        [Benchmark]
+        public void ShiftLeft() => TensorPrimitives.ShiftLeft<T>(_source1, shiftAmount: 3, _destination);
+
+        [Benchmark]
+        public void TrailingZeroCount() => TensorPrimitives.TrailingZeroCount<T>(_source1, _destination);
+        #endregion
+    }
+}
\ No newline at end of file
diff --git a/src/benchmarks/micro/libraries/System.Numerics.Tensors/Perf_FloatingPointTensorPrimitives.cs b/src/benchmarks/micro/libraries/System.Numerics.Tensors/Perf_FloatingPointTensorPrimitives.cs
new file mode 100644
index 00000000000..63212e36d44
--- /dev/null
+++ b/src/benchmarks/micro/libraries/System.Numerics.Tensors/Perf_FloatingPointTensorPrimitives.cs
@@ -0,0 +1,105 @@
+// Licensed to the .NET Foundation under one or more agreements.
+// The .NET Foundation licenses this file to you under the MIT license.
+// See the LICENSE file in the project root for more information.
+
+using BenchmarkDotNet.Attributes;
+using BenchmarkDotNet.Extensions;
+using MicroBenchmarks;
+using System.Linq;
+
+namespace System.Numerics.Tensors.Tests
+{
+    [BenchmarkCategory(Categories.Libraries, Categories.SIMD, Categories.JIT)]
+    [GenericTypeArguments(typeof(float))]
+    [GenericTypeArguments(typeof(double))]
+    public class Perf_FloatingPointTensorPrimitives<T>
+        where T : unmanaged, IFloatingPointIeee754<T>
+    {
+        [Params(128, 6 * 512 + 7)]
+        public int BufferLength;
+
+        private T[] _source1;
+        private T[] _source2;
+        private T[] _source3;
+        private T _scalar1;
+        private T[] _ones;
+        private T[] _destination;
+
+        [GlobalSetup]
+        public void Init()
+        {
+            _source1 = ValuesGenerator.Array<T>(BufferLength, seed: 42);
+            _source2 = ValuesGenerator.Array<T>(BufferLength, seed: 43);
+            _source3 = ValuesGenerator.Array<T>(BufferLength, seed: 44);
+            _scalar1 = ValuesGenerator.Value<T>(seed: 45);
+            _ones = Enumerable.Repeat(T.One, BufferLength).ToArray();
+            _destination = new T[BufferLength];
+        }
+
+        #region Unary Operations
+        [Benchmark]
+        public void AtanPi() => TensorPrimitives.AtanPi<T>(_source1, _destination);
+
+        [Benchmark]
+        public void Exp() => TensorPrimitives.Exp<T>(_source1, _destination);
+
+        [Benchmark]
+        public void Log() => TensorPrimitives.Log<T>(_source1, _destination);
+
+        [Benchmark]
+        public void Round() => TensorPrimitives.Round<T>(_source1, _destination);
+
+        [Benchmark]
+        public void Sin() => TensorPrimitives.Sin<T>(_source1, _destination);
+
+        [Benchmark]
+        public void Sinh() => TensorPrimitives.Sinh<T>(_source1, _destination);
+
+        [Benchmark]
+        public void Sigmoid() => TensorPrimitives.Sigmoid<T>(_source1, _destination);
+
+        [Benchmark]
+        public void Sqrt() => TensorPrimitives.Sqrt<T>(_source1, _destination);
+
+        [Benchmark]
+        public void Truncate() => TensorPrimitives.Truncate<T>(_source1, _destination);
+        #endregion
+
+        #region Binary/Ternary Operations
+        [Benchmark]
+        public void FusedMultiplyAdd_Vectors() => TensorPrimitives.FusedMultiplyAdd<T>(_source1, _source2, _source3, _destination);
+
+        [Benchmark]
+        public void FusedMultiplyAdd_ScalarAddend() => TensorPrimitives.FusedMultiplyAdd(_source1, _scalar1, _source2, _destination);
+
+        [Benchmark]
+        public void FusedMultiplyAdd_ScalarMultiplier() => TensorPrimitives.FusedMultiplyAdd(_source1, _source2, _scalar1, _destination);
+
+        [Benchmark]
+        public void Ieee754Remainder_Vector() => TensorPrimitives.Ieee754Remainder<T>(_source1, _source2, _destination);
+
+        [Benchmark]
+        public void Ieee754Remainder_ScalarDividend() => TensorPrimitives.Ieee754Remainder(_scalar1, _source1, _destination);
+
+        [Benchmark]
+        public void Ieee754Remainder_ScalarDivisor() => TensorPrimitives.Ieee754Remainder(_source1, _scalar1, _destination);
+
+        [Benchmark]
+        public void Pow_Vectors() => TensorPrimitives.Pow<T>(_source1, _ones, _destination);
+
+        [Benchmark]
+        public void Pow_ScalarBase() => TensorPrimitives.Pow(_scalar1, _ones, _destination);
+
+        [Benchmark]
+        public void Pow_ScalarExponent() => TensorPrimitives.Pow(_source1, T.One, _destination);
+        #endregion
+
+        #region Reducers
+        [Benchmark]
+        public T CosineSimilarity() => TensorPrimitives.CosineSimilarity<T>(_source1, _source2);
+
+        [Benchmark]
+        public T Distance() => TensorPrimitives.Distance<T>(_source1, _source2);
+        #endregion
+    }
+}
diff --git a/src/benchmarks/micro/libraries/System.Numerics.Tensors/Perf_NumberTensorPrimitives.cs b/src/benchmarks/micro/libraries/System.Numerics.Tensors/Perf_NumberTensorPrimitives.cs
new file mode 100644
index 00000000000..a3adab1b054
--- /dev/null
+++ b/src/benchmarks/micro/libraries/System.Numerics.Tensors/Perf_NumberTensorPrimitives.cs
@@ -0,0 +1,103 @@
+// Licensed to the .NET Foundation under one or more agreements.
+// The .NET Foundation licenses this file to you under the MIT license.
+// See the LICENSE file in the project root for more information.
+
+using BenchmarkDotNet.Attributes;
+using BenchmarkDotNet.Extensions;
+using MicroBenchmarks;
+using System.Linq;
+
+namespace System.Numerics.Tensors.Tests
+{
+    [BenchmarkCategory(Categories.Libraries, Categories.SIMD, Categories.JIT)]
+    [GenericTypeArguments(typeof(int))]
+    [GenericTypeArguments(typeof(float))]
+    [GenericTypeArguments(typeof(double))]
+    public class Perf_NumberTensorPrimitives<T>
+        where T : INumber<T>, IBitwiseOperators<T, T, T>
+    {
+        [Params(128, 6 * 512 + 7)]
+        public int BufferLength;
+
+        private T[] _source1;
+        private T[] _source2;
+        private T[] _source3;
+        private T _scalar1;
+        private T[] _ones;
+        private T[] _destination;
+
+        [GlobalSetup]
+        public void Init()
+        {
+            _source1 = ValuesGenerator.Array<T>(BufferLength, seed: 42);
+            _source2 = ValuesGenerator.Array<T>(BufferLength, seed: 43);
+            _source3 = ValuesGenerator.Array<T>(BufferLength, seed: 44);
+            _scalar1 = ValuesGenerator.Value<T>(seed: 45);
+            _ones = Enumerable.Repeat(T.One, BufferLength).ToArray();
+            _destination = new T[BufferLength];
+        }
+
+        #region Unary Operations
+        [Benchmark]
+        public void Abs() => TensorPrimitives.Abs<T>(_source1, _destination);
+
+        [Benchmark]
+        public void Negate() => TensorPrimitives.Abs<T>(_source1, _destination);
+        #endregion
+
+        #region Binary/Ternary Operations
+        [Benchmark]
+        public void Add_Vector() => TensorPrimitives.Add<T>(_source1, _source2, _destination);
+
+        [Benchmark]
+        public void Add_Scalar() => TensorPrimitives.Add(_source1, _scalar1, _destination);
+
+        [Benchmark]
+        public void AddMultiply_Vectors() => TensorPrimitives.AddMultiply<T>(_source1, _source2, _source3, _destination);
+
+        [Benchmark]
+        public void AddMultiply_ScalarAddend() => TensorPrimitives.AddMultiply(_source1, _scalar1, _source2, _destination);
+
+        [Benchmark]
+        public void AddMultiply_ScalarMultiplier() => TensorPrimitives.AddMultiply(_source1, _source2, _scalar1, _destination);
+
+        [Benchmark]
+        public void BitwiseAnd_Vector() => TensorPrimitives.BitwiseAnd<T>(_source1, _source2, _destination);
+
+        [Benchmark]
+        public void BitwiseAnd_Scalar() => TensorPrimitives.BitwiseAnd(_source1, _scalar1, _destination);
+
+        [Benchmark]
+        public void Divide_Vector() => TensorPrimitives.Divide<T>(_source1, _ones, _destination);
+
+        [Benchmark]
+        public void Divide_Scalar() => TensorPrimitives.Divide(_source1, T.One, _destination);
+
+        [Benchmark]
+        public void Max_Vector() => TensorPrimitives.Max<T>(_source1, _source2, _destination);
+
+        [Benchmark]
+        public void Max_Scalar() => TensorPrimitives.Max(_source1, _scalar1, _destination);
+
+        [Benchmark]
+        public void MaxMagnitude_Vector() => TensorPrimitives.MaxMagnitude<T>(_source1, _source2, _destination);
+
+        [Benchmark]
+        public void MaxMagnitude_Scalar() => TensorPrimitives.MaxMagnitude(_source1, _scalar1, _destination);
+        #endregion
+
+        #region Reducers
+        [Benchmark]
+        public T Max() => TensorPrimitives.Max<T>(_source1);
+
+        [Benchmark]
+        public T SumOfMagnitudes() => TensorPrimitives.SumOfMagnitudes<T>(_ones);
+
+        [Benchmark]
+        public T SumOfSquares() => TensorPrimitives.SumOfSquares<T>(_ones);
+
+        [Benchmark]
+        public int IndexOfMax() => TensorPrimitives.IndexOfMax<T>(_source1);
+        #endregion
+    }
+}
diff --git a/src/harness/BenchmarkDotNet.Extensions/ValuesGenerator.cs b/src/harness/BenchmarkDotNet.Extensions/ValuesGenerator.cs
index fa36e3c17a1..1b0bc42c87e 100644
--- a/src/harness/BenchmarkDotNet.Extensions/ValuesGenerator.cs
+++ b/src/harness/BenchmarkDotNet.Extensions/ValuesGenerator.cs
@@ -67,11 +67,11 @@ public static T[] ArrayOfUniqueValues<T>(int count)
         /// For byte and sbyte values are built from Random.NextBytes, for other types GenerateValue is used
         /// to generate a random value in the appropriate range
         /// </summary>
-        public static T[] Array<T>(int count)
+        public static T[] Array<T>(int count, int? seed = null)
         {
             var result = new T[count];
 
-            var random = new Random(Seed);
+            var random = new Random(seed ?? Seed);
 
             if (typeof(T) == typeof(byte) || typeof(T) == typeof(sbyte))
             {
@@ -88,6 +88,12 @@ public static T[] Array<T>(int count)
             return result;
         }
 
+        public static T Value<T>(int? seed = null)
+        {
+            var random = new Random(seed ?? Seed);
+            return GenerateValue<T>(random);
+        }
+
         public static readonly byte[] s_encodingMap = {
             65, 66, 67, 68, 69, 70, 71, 72,         //A..H
             73, 74, 75, 76, 77, 78, 79, 80,         //I..P