rewrite of f32 fma, and other optmization of simd kernel ops

StgSharp.Mathematics/Numeric/MatrixParallelHandle.cs

@@ -1,6 +1,6 @@
 //-----------------------------------------------------------------------
 // -----------------------------------------------------------------------
-// file="MatrixParallelTask"
+// file="MatrixParallelHandle"
 // Project: StgSharp
 // AuthorGroup: Nitload
 // Copyright (c) Nitload. All rights reserved.
@@ -34,22 +34,6 @@ using System.Runtime.InteropServices;
 
 namespace StgSharp.Mathematics.Numeric
 {
-    public unsafe struct ScalarPacket
-    {
-
-        private fixed ulong data[8];
-
-        public ref T Data<T>(int index) where T: unmanaged,INumber<T>
-        {
-            if (index is < 0 or >= 8) {
-                throw new IndexOutOfRangeException();
-            }
-
-            return ref Unsafe.As<ulong, T>(ref data[index]);
-        }
-
-    }
-
     [StructLayout(LayoutKind.Sequential)]
     internal unsafe struct MatrixParallelTaskPackageNonGeneric
     {
@@ -59,11 +43,8 @@ namespace StgSharp.Mathematics.Numeric
                                   MatrixParallelTaskPackageNonGeneric* source,
                                   MatrixParallelTaskPackageNonGeneric* target)
         {
-            Buffer.MemoryCopy(
-                source,
-                target,
-                sizeof(MatrixParallelTaskPackageNonGeneric),
-                sizeof(MatrixParallelTaskPackageNonGeneric));
+            Buffer.MemoryCopy(source, target, sizeof(MatrixParallelTaskPackageNonGeneric),
+                              sizeof(MatrixParallelTaskPackageNonGeneric));
         }
 
         #region source
@@ -148,11 +129,8 @@ namespace StgSharp.Mathematics.Numeric
         [MethodImpl(MethodImplOptions.AggressiveInlining)]
         public static unsafe void Copy(MatrixParallelTaskPackage<T>* source, MatrixParallelTaskPackage<T>* target)
         {
-            Buffer.MemoryCopy(
-                source,
-                target,
-                sizeof(MatrixParallelTaskPackage<T>),
-                sizeof(MatrixParallelTaskPackage<T>));
+            Buffer.MemoryCopy(source, target, sizeof(MatrixParallelTaskPackage<T>),
+                              sizeof(MatrixParallelTaskPackage<T>));
         }
 
         #region source