Initial integration of intrinsics

src/NaCl.Core/Base/ChaCha20Base.cs

@@ -64,6 +64,19 @@ public override void ProcessKeyStreamBlock(ReadOnlySpan<byte> nonce, int counter
             ArrayUtils.StoreArray16UInt32LittleEndian(block, 0, state);
         }
 
+#if INTRINSICS
+        public override unsafe void ProcessStream(ReadOnlySpan<byte> nonce, Span<byte> output, ReadOnlySpan<byte> input, int initialCounter, int offset = 0)
+        {
+            Span<uint> state = stackalloc uint[BLOCK_SIZE_IN_INTS];
+            SetInitialState(state, nonce, initialCounter);
+            fixed(uint* x = state)
+            fixed (byte* m = input, c = output.Slice(offset))
+            {
+                ChaCha20BaseIntrinsics.ChaCha20(x, m, c, (ulong)input.Length);
+            }
+        }
+#endif
+
         /// <summary>
         /// Process a pseudorandom keystream block, converting the key and part of the <paramref name="nonce"> into a <paramref name="subkey">, and the remainder of the <paramref name="nonce">.
         /// </summary>

src/NaCl.Core/Base/Snuffle.cs

@@ -59,6 +59,10 @@ public Snuffle(ReadOnlyMemory<byte> key, int initialCounter)
         /// <returns>ByteBuffer.</returns>
         public abstract void ProcessKeyStreamBlock(ReadOnlySpan<byte> nonce, int counter, Span<byte> block);
 
+#if INTRINSICS
+        public abstract void ProcessStream(ReadOnlySpan<byte> nonce, Span<byte> output, ReadOnlySpan<byte> input, int initialCounter, int offset = 0);
+#endif
+
         /// <summary>
         /// The size of the randomly generated nonces.
         /// ChaCha20 uses 12-byte nonces, but XSalsa20 and XChaCha20 use 24-byte nonces.
@@ -193,6 +197,10 @@ public virtual void Decrypt(ReadOnlySpan<byte> ciphertext, ReadOnlySpan<byte> no
             Process(nonce, plaintext, ciphertext);
         }
 
+
+#if INTRINSICS
+        private void Process(ReadOnlySpan<byte> nonce, Span<byte> output, ReadOnlySpan<byte> input, int offset = 0) => ProcessStream(nonce, output, input, InitialCounter, offset);
+#else
         /// <summary>
         /// Processes the Encryption/Decryption function.
         /// </summary>
@@ -237,6 +245,7 @@ private void Process(ReadOnlySpan<byte> nonce, Span<byte> output, ReadOnlySpan<b
                 }
             }
         }
+#endif
 
         /// <summary>
         /// Formats the nonce length exception message.

src/NaCl.Core/NaCl.Core.csproj

@@ -1,8 +1,8 @@
-<Project Sdk="Microsoft.NET.Sdk">
+<Project Sdk="Microsoft.NET.Sdk">
 
   <PropertyGroup>
-    <TargetFrameworks>netstandard1.6;netstandard2.0;netstandard2.1;netcoreapp3.1</TargetFrameworks>
-    <TargetFrameworks Condition="'$(OS)' != 'Unix'">netstandard1.6;netstandard2.0;netstandard2.1;netcoreapp3.1;net45;net48</TargetFrameworks>
+    <TargetFrameworks>netstandard1.6;netstandard2.0;netstandard2.1;netcoreapp3.1;netcoreapp5.0</TargetFrameworks>
+    <TargetFrameworks Condition="'$(OS)' != 'Unix'">netstandard1.6;netstandard2.0;netstandard2.1;netcoreapp3.1;netcoreapp5.0;net45;net48</TargetFrameworks>
     <AllowUnsafeBlocks>true</AllowUnsafeBlocks>
     <Version>1.2.0</Version>
     <Authors>David De Smet</Authors>
@@ -26,10 +26,14 @@
     <PackageReference Include="System.Memory" Version="4.5.3" />
   </ItemGroup>
 
-  <PropertyGroup Condition="$(TargetFramework) == 'netcoreapp3.1'">
+  <PropertyGroup Condition="$(TargetFramework) == 'netcoreapp3.1' OR $(TargetFramework) == 'netcoreapp5.0'">
     <DefineConstants>FCL_BITOPS</DefineConstants>
   </PropertyGroup>
 
+  <PropertyGroup Condition="$(TargetFramework) == 'netcoreapp3.1' OR $(TargetFramework) == 'netcoreapp5.0'">
+    <DefineConstants>INTRINSICS</DefineConstants>
+  </PropertyGroup>
+
   <Target Name="LogDebugInfo">
     <Message Text="Building for $(TargetFramework) on $(OS)" Importance="High" />
   </Target>

test/NaCl.Core.Benchmarks/ChaCha20Benchmark.cs

@@ -8,8 +8,11 @@
     using Internal;
 
     using BenchmarkDotNet.Attributes;
+    using BenchmarkDotNet.Jobs;
 
     [BenchmarkCategory("Stream Cipher")]
+    [SimpleJob(RuntimeMoniker.NetCoreApp21, baseline: true)]
+    [SimpleJob(RuntimeMoniker.NetCoreApp31)]
     [MemoryDiagnoser]
     [RPlotExporter, RankColumn]
     public class ChaCha20Benchmark

test/NaCl.Core.Benchmarks/NaCl.Core.Benchmarks.csproj

@@ -1,8 +1,8 @@
-<Project Sdk="Microsoft.NET.Sdk">
+<Project Sdk="Microsoft.NET.Sdk">
 
   <PropertyGroup>
     <OutputType>Exe</OutputType>
-    <TargetFramework>netcoreapp3.1</TargetFramework>
+    <TargetFrameworks>netcoreapp3.1;netcoreapp2.1</TargetFrameworks>
   </PropertyGroup>
 
   <ItemGroup>

test/NaCl.Core.Benchmarks/Program.cs

@@ -11,11 +11,11 @@ static void Main(string[] args)
             // Execute following code:
             // $ dotnet run -c release --framework netcoreapp3.1
             //BenchmarkSwitcher.FromAssembly(typeof(Program).Assembly).Run(args);
-            BenchmarkRunner.Run<Poly1305Benchmark>();
+            //BenchmarkRunner.Run<Poly1305Benchmark>();
             BenchmarkRunner.Run<ChaCha20Benchmark>();
-            BenchmarkRunner.Run<ChaCha20Poly1305Benchmark>();
-            BenchmarkRunner.Run<XChaCha20Benchmark>();
-            BenchmarkRunner.Run<XChaCha20Poly1305Benchmark>();
+            //BenchmarkRunner.Run<ChaCha20Poly1305Benchmark>();
+            //BenchmarkRunner.Run<XChaCha20Benchmark>();
+            //BenchmarkRunner.Run<XChaCha20Poly1305Benchmark>();
 
             Console.ReadLine();
         }

test/NaCl.Core.Tests/NaCl.Core.Tests.csproj

@@ -1,7 +1,7 @@
 <Project Sdk="Microsoft.NET.Sdk">
 
   <PropertyGroup>
-    <TargetFramework>netcoreapp3.1</TargetFramework>
+    <TargetFrameworks>netcoreapp3.1;netcoreapp2.1</TargetFrameworks>
     <LangVersion>latest</LangVersion>
     <IsTestProject>true</IsTestProject>
   </PropertyGroup>