Velocity SIMD CPU Runtime (Runtime + Scalar x2)

.github/workflows/ci.yml

@@ -154,6 +154,12 @@ jobs:
         os: ${{ fromJson(needs.setup-os-matrix.outputs.os) }}
         library: [ILGPU, ILGPU.Algorithms]
         framework: [net6.0, net7.0]
+        flavor: [CPU, Velocity]
+        exclude:
+          - library: ILGPU.Algorithms
+            flavor: Velocity
+          - os: cuda
+            flavor: Velocity
       fail-fast: false
     runs-on: ${{ matrix.os }}
     steps:
@@ -174,7 +180,7 @@ jobs:
       - name: Set test flavor
         id: test-flavor
         shell: bash
-        run: echo "flavor=$([[ "${{ matrix.os }}" == cuda ]] && echo "Cuda" || echo "CPU")" >> $GITHUB_OUTPUT
+        run: echo "flavor=$([[ "${{ matrix.os }}" == cuda ]] && echo "Cuda" || echo "${{ matrix.flavor }}")" >> $GITHUB_OUTPUT
 
       - name: Build and test
         run: dotnet test Src/${{ matrix.library }}.Tests.${{ steps.test-flavor.outputs.flavor }} --configuration=Release --framework=${{ matrix.framework }} -p:TreatWarningsAsErrors=true

.gitignore

@@ -257,6 +257,7 @@ Src/ILGPU/AtomicFunctions.cs
 Src/ILGPU/Backends/PTX/PTXIntrinsics.Generated.cs
 Src/ILGPU/Backends/PTX/PTXLibDeviceMethods.cs
 Src/ILGPU/Backends/PTX/PTXLibDeviceNvvm.cs
+Src/ILGPU/Backends/Velocity/Scalar/ScalarOperations.cs
 Src/ILGPU/Frontend/Intrinsic/RemappedIntrinsics.Generated.cs
 Src/ILGPU/HalfConversion.cs
 Src/ILGPU/IR/Construction/ArithmeticOperations.cs
@@ -339,6 +340,7 @@ Src/ILGPU.Tests/.test.runsettings
 Src/ILGPU.Tests.CPU/Configurations.cs
 Src/ILGPU.Tests.Cuda/Configurations.cs
 Src/ILGPU.Tests.OpenCL/Configurations.cs
+Src/ILGPU.Tests.Velocity/Configurations.cs
 
 # Generated test source files (Algorithms)
 Src/ILGPU.Algorithms.Tests/Generic/ConfigurationBase.cs

Src/ILGPU.Tests.Velocity/.editorconfig

@@ -0,0 +1,7 @@
+[*.cs]
+
+# CA1707: Identifiers should not contain underscores
+dotnet_diagnostic.CA1707.severity = none
+
+# CA1014: Mark assemblies with CLSCompliant
+dotnet_diagnostic.CA1014.severity = none

Src/ILGPU.Tests.Velocity/Configurations.tt

@@ -0,0 +1,51 @@
+// ---------------------------------------------------------------------------------------
+//                                        ILGPU
+//                           Copyright (c) 2023 ILGPU Project
+//                                    www.ilgpu.net
+//
+// File: Configurations.tt/Configurations.cs
+//
+// This file is part of ILGPU and is distributed under the University of Illinois Open
+// Source License. See LICENSE.txt for details.
+// ---------------------------------------------------------------------------------------
+
+<#@ template debug="false" hostspecific="true" language="C#" #>
+<#@ include file="../ILGPU.Tests/Generic/ConfigurationBase.tt" #>
+<#@ assembly name="System.Core" #>
+<#@ import namespace="System.IO" #>
+using Xunit;
+using Xunit.Abstractions;
+
+<#
+var configurationFile = Host.ResolvePath("../ILGPU.Tests/Configurations.txt");
+var configurations = TestConfig.Parse(configurationFile);
+#>
+namespace ILGPU.Tests.Velocity
+{
+<# foreach (var (test, level, collection) in configurations) { #>
+<#      var name = $"Velocity{test}_{level}"; #>
+    [Collection("VelocityContextCollection<#= collection #>")]
+    public sealed partial class <#= name #> : <#= test #>
+    {
+        public <#= name #>(
+            ITestOutputHelper output,
+            VelocityTestContext<#= collection #> testContext)
+            : base(output, testContext)
+        { }
+    }
+
+<# } #>
+<# foreach (var (config, level) in TestConfig.AllConfigurations) { #>
+    public class VelocityTestContext<#= config #> : VelocityTestContext
+    {
+        public VelocityTestContext<#= config #>()
+            : base(OptimizationLevel.<#= level #>)
+        { }
+    }
+
+    [CollectionDefinition("VelocityContextCollection<#= config #>")]
+    public class VelocityContextCollection<#= config #> :
+        ICollectionFixture<VelocityTestContext<#= config #>> { }
+
+<# } #>
+}

Src/ILGPU.Tests.Velocity/ILGPU.Tests.Velocity.csproj

@@ -0,0 +1,59 @@
+<Project Sdk="Microsoft.NET.Sdk">
+
+  <PropertyGroup>
+    <TargetFrameworks>$(LibraryUnitTestTargetFrameworks)</TargetFrameworks>
+    <IsPackable>false</IsPackable>
+  </PropertyGroup>
+
+  <PropertyGroup>
+    <RunSettingsFilePath>$(MSBuildProjectDirectory)\..\ILGPU.Tests\.test.runsettings</RunSettingsFilePath>
+  </PropertyGroup>
+
+  <PropertyGroup>
+    <EnableNETAnalyzers>true</EnableNETAnalyzers>
+    <AnalysisMode>AllEnabledByDefault</AnalysisMode>
+  </PropertyGroup>
+
+  <ItemGroup>
+    <PackageReference Include="Microsoft.NET.Test.Sdk" Version="17.7.2" />
+    <PackageReference Include="xunit" Version="2.5.1" />
+    <PackageReference Include="xunit.runner.visualstudio" Version="2.5.1">
+      <PrivateAssets>all</PrivateAssets>
+      <IncludeAssets>runtime; build; native; contentfiles; analyzers; buildtransitive</IncludeAssets>
+    </PackageReference>
+    <DotNetCliToolReference Include="dotnet-xunit" Version="2.3.1" />
+    <PackageReference Include="T4.Build" Version="0.2.4" PrivateAssets="All" />
+    <PackageReference Include="Xunit.SkippableFact" Version="1.4.13" />
+  </ItemGroup>
+  <ItemGroup>
+    <ProjectReference Include="..\..\Src\ILGPU\ILGPU.csproj" />
+    <ProjectReference Include="..\ILGPU.Tests\ILGPU.Tests.csproj" />
+  </ItemGroup>
+
+  <ItemGroup>
+    <None Include="Configurations.cs">
+      <DesignTime>True</DesignTime>
+      <AutoGen>True</AutoGen>
+      <DependentUpon>Configurations.tt</DependentUpon>
+    </None>
+  </ItemGroup>
+
+  <ItemGroup>
+    <None Update="Configurations.tt">
+      <Generator>TextTemplatingFileGenerator</Generator>
+      <LastGenOutput>Configurations.cs</LastGenOutput>
+    </None>
+  </ItemGroup>
+
+  <ItemGroup>
+    <Service Include="{508349b6-6b84-4df5-91f0-309beebad82d}" />
+  </ItemGroup>
+
+  <ItemGroup>
+    <Compile Update="Configurations.cs">
+      <DesignTime>True</DesignTime>
+      <AutoGen>True</AutoGen>
+      <DependentUpon>Configurations.tt</DependentUpon>
+    </Compile>
+  </ItemGroup>
+</Project>

Src/ILGPU.Tests.Velocity/TestContext.cs

@@ -0,0 +1,45 @@
+// ---------------------------------------------------------------------------------------
+//                                        ILGPU
+//                           Copyright (c) 2023 ILGPU Project
+//                                    www.ilgpu.net
+//
+// File: TestContext.cs
+//
+// This file is part of ILGPU and is distributed under the University of Illinois Open
+// Source License. See LICENSE.txt for details.
+// ---------------------------------------------------------------------------------------
+
+using ILGPU.Runtime.Velocity;
+using System;
+
+namespace ILGPU.Tests.Velocity
+{
+    /// <summary>
+    /// An abstract test context for Velocity accelerators.
+    /// </summary>
+    public abstract class VelocityTestContext : TestContext
+    {
+        /// <summary>
+        /// Creates a new test context instance.
+        /// </summary>
+        /// <param name="optimizationLevel">The optimization level to use.</param>
+        /// <param name="prepareContext">The context preparation handler.</param>
+        protected VelocityTestContext(
+            OptimizationLevel optimizationLevel,
+            Action<Context.Builder> prepareContext)
+            : base(
+                  optimizationLevel,
+                  builder => prepareContext(
+                      builder.Velocity(VelocityDeviceType.Scalar2)),
+                  context => context.CreateVelocityAccelerator())
+        { }
+
+        /// <summary>
+        /// Creates a new test context instance.
+        /// </summary>
+        /// <param name="optimizationLevel">The optimization level to use.</param>
+        protected VelocityTestContext(OptimizationLevel optimizationLevel)
+            : this(optimizationLevel, _ => { })
+        { }
+    }
+}

Src/ILGPU.sln

@@ -25,6 +25,8 @@ Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "ILGPU.Algorithms.Tests.Open
 EndProject
 Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "ILGPU.Algorithms.Tests", "ILGPU.Algorithms.Tests\ILGPU.Algorithms.Tests.csproj", "{18F2225C-82FD-4B01-8AF9-CF746D16EDA1}"
 EndProject
+Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "ILGPU.Tests.Velocity", "ILGPU.Tests.Velocity\ILGPU.Tests.Velocity.csproj", "{4AFD2AAD-FA52-43EA-B9A8-10E948F9A139}"
+EndProject
 Global
 	GlobalSection(SolutionConfigurationPlatforms) = preSolution
 		Debug|Any CPU = Debug|Any CPU
@@ -71,6 +73,10 @@ Global
 		{18F2225C-82FD-4B01-8AF9-CF746D16EDA1}.Debug|Any CPU.Build.0 = Debug|Any CPU
 		{18F2225C-82FD-4B01-8AF9-CF746D16EDA1}.Release|Any CPU.ActiveCfg = Release|Any CPU
 		{18F2225C-82FD-4B01-8AF9-CF746D16EDA1}.Release|Any CPU.Build.0 = Release|Any CPU
+		{4AFD2AAD-FA52-43EA-B9A8-10E948F9A139}.Debug|Any CPU.ActiveCfg = Debug|Any CPU
+		{4AFD2AAD-FA52-43EA-B9A8-10E948F9A139}.Debug|Any CPU.Build.0 = Debug|Any CPU
+		{4AFD2AAD-FA52-43EA-B9A8-10E948F9A139}.Release|Any CPU.ActiveCfg = Release|Any CPU
+		{4AFD2AAD-FA52-43EA-B9A8-10E948F9A139}.Release|Any CPU.Build.0 = Release|Any CPU
 	EndGlobalSection
 	GlobalSection(SolutionProperties) = preSolution
 		HideSolutionNode = FALSE
@@ -84,6 +90,7 @@ Global
 		{AA73D3B1-873F-4A79-8347-E0781E382FF8} = {7701FE3C-4187-401C-9612-44667203B0E5}
 		{6ED7EDA1-E6DA-4867-8084-C0327EEFB7A9} = {7701FE3C-4187-401C-9612-44667203B0E5}
 		{18F2225C-82FD-4B01-8AF9-CF746D16EDA1} = {7701FE3C-4187-401C-9612-44667203B0E5}
+		{4AFD2AAD-FA52-43EA-B9A8-10E948F9A139} = {7701FE3C-4187-401C-9612-44667203B0E5}
 	EndGlobalSection
 	GlobalSection(ExtensibilityGlobals) = postSolution
 		SolutionGuid = {22270DEE-D42D-479D-A76F-B2E7A5F7C949}

Src/ILGPU/Backends/Backend.cs

@@ -66,6 +66,11 @@ public enum BackendType
         /// </summary>
         IL,
 
+        /// <summary>
+        /// A Velocity backend.
+        /// </summary>
+        Velocity,
+
         /// <summary>
         /// A PTX backend.
         /// </summary>
@@ -123,7 +128,7 @@ public void OptimizedKernelContext(
         /// <summary>
         /// Represents the current kernel context in scope of a backend instance.
         /// </summary>
-        protected readonly ref struct BackendContext
+        protected internal readonly ref struct BackendContext
         {
             #region Nested Types