Added new giteaaction

.gitea/workflows/publish-nuget.yaml

@@ -0,0 +1,53 @@
+name: Publish NuGet Packages
+
+on:
+  push:
+    tags:
+      - 'v*.*.*'   # Trigger only on semantic version tags
+
+jobs:
+  publish:
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v3
+
+      - name: Setup .NET
+        uses: actions/setup-dotnet@v3
+        with:
+          dotnet-version: 9.0.x
+
+      - name: Restore dependencies
+        run: dotnet restore
+
+      - name: Build all projects
+        run: dotnet build -c Release
+
+      - name: Pack all projects
+        run: |
+          VERSION=$(dotnet msbuild "$projname/$projname.csproj" \
+            -nologo -v:q -t:GetProperty -p:PropertyName=PackageVersion)
+        for projfile in $(find . -name "*.csproj"); do
+            packable=$(dotnet msbuild $projfile -t:GetProperty -p:PropertyName=IsPackable -nologo -v:q)
+            if [ "$packable" = "true" ]; then
+              dotnet pack $proj/$proj.csproj -c Release
+            fi
+          done
+
+      - name: Publish all packages
+        env:
+          NUGET_API_KEY: ${{ secrets.GITEA_TOKEN }}
+        run: |
+          VERSION=$(dotnet msbuild "$projname/$projname.csproj" \
+            -nologo -v:q -t:GetProperty -p:PropertyName=PackageVersion)
+        for projfile in $(find . -name "*.csproj"); do
+            packable=$(dotnet msbuild $projfile -t:GetProperty -p:PropertyName=IsPackable -nologo -v:q)
+            if [ "$packable" = "true" ]; then
+                projname=$(basename $projfile .csproj)
+                dotnet pack $projfile -c Release
+                dotnet nuget push $projname/bin/Release/$projname.$VERSION.nupkg \
+                    --source "https://git.personalnas.com/api/packages/Misaki/nuget/index.json" \
+                    --api-key $NUGET_API_KEY
+            fi
+        done

Misaki.HighPerformance.Image/Misaki.HighPerformance.Image.csproj

@@ -9,6 +9,7 @@
         <Version>1.0.0</Version>
         <AllowUnsafeBlocks>true</AllowUnsafeBlocks>
         <TargetFramework>net9.0</TargetFramework>
+			<IsPackable>false</IsPackable>
     </PropertyGroup>
 
     <PropertyGroup Condition="'$(Configuration)'=='Release'">

Misaki.HighPerformance.Jobs/AssemblyInfo.cs.cs

@@ -1,2 +1,2 @@
 
-global using unsafe JobExecuteFunc = delegate*<void*, ref Misaki.HighPerformance.Jobs.JobRanges, ref int, int, bool>;
+global using unsafe JobExecutionFunc = delegate*<void*, ref Misaki.HighPerformance.Jobs.JobRanges, ref int, int, bool>;

Misaki.HighPerformance.Jobs/JobInfo.cs

@@ -1,11 +1,29 @@
 namespace Misaki.HighPerformance.Jobs;
 
-public enum JobStatus
+/// <summary>
+/// The state of a job in its lifecycle.
+/// </summary>
+public enum JobState
 {
+    /// <summary>
+    /// The job is in an invalid state, indicating an error or uninitialized state. Or already finished the execution and cleaned up by the system.
+    /// </summary>
     Invalid = -1,
+    /// <summary>
+    /// The job has been created but not yet scheduled for execution.
+    /// </summary>
     Created = 0,
+    /// <summary>
+    /// The job is scheduled and waiting to be executed.
+    /// </summary>
     Scheduled = 1,
+    /// <summary>
+    /// The job is currently being executed.
+    /// </summary>
     Running = 2,
+    /// <summary>
+    /// The job has completed execution.
+    /// </summary>
     Completed = 3
 }
 
@@ -21,9 +39,9 @@ internal unsafe struct JobInfo
     public JobRanges jobRanges;
 
     public void* pJobData;
-    public JobExecuteFunc executeDelegate;
+    public JobExecutionFunc pExecutionFunc;
 
-    public JobStatus status;
+    public JobState state;
     public int remainingBatches;
 
     public int threadIndex; // The preferred thread index to run this job on, -1 means any thread

Misaki.HighPerformance.Jobs/JobScheduler.cs

@@ -71,7 +71,7 @@ public unsafe sealed class JobScheduler : IDisposable
 
         if (exist && Volatile.Read(ref jobInfo.dependencyCount) == 0)
         {
-            if (Interlocked.CompareExchange(ref jobInfo.status, JobStatus.Scheduled, JobStatus.Created) != JobStatus.Created)
+            if (Interlocked.CompareExchange(ref jobInfo.state, JobState.Scheduled, JobState.Created) != JobState.Created)
             {
                 return;
             }
@@ -117,7 +117,7 @@ public unsafe sealed class JobScheduler : IDisposable
             lock (_lock)
             {
                 ref var depJobInfo = ref _jobInfoPool.GetElementReferenceAt(dependency._id, dependency._generation, out var exist);
-                if (!exist || Volatile.Read(ref Unsafe.As<JobStatus, int>(ref depJobInfo.status)) == (int)JobStatus.Completed)
+                if (!exist || Volatile.Read(ref Unsafe.As<JobState, int>(ref depJobInfo.state)) == (int)JobState.Completed)
                 {
                     continue;
                 }
@@ -197,7 +197,7 @@ public unsafe sealed class JobScheduler : IDisposable
             return;
         }
 
-        if (Interlocked.CompareExchange(ref info.status, JobStatus.Completed, JobStatus.Running) != JobStatus.Running)
+        if (Interlocked.CompareExchange(ref info.state, JobState.Completed, JobState.Running) != JobState.Running)
         {
             return;
         }
@@ -256,7 +256,7 @@ public unsafe sealed class JobScheduler : IDisposable
         var jobInfo = new JobInfo
         {
             pJobData = jobData,
-            executeDelegate = &JobExecutor.Execute<T>,
+            pExecutionFunc = &JobExecutor.Execute<T>,
 
             remainingBatches = 1,
             threadIndex = threadIndex,
@@ -311,7 +311,7 @@ public unsafe sealed class JobScheduler : IDisposable
         var jobInfo = new JobInfo
         {
             pJobData = jobData,
-            executeDelegate = &JobExecutor.ExecuteParallel<T>,
+            pExecutionFunc = &JobExecutor.ExecuteParallel<T>,
 
             remainingBatches = totalBatches,
             threadIndex = threadIndex,
@@ -352,7 +352,7 @@ public unsafe sealed class JobScheduler : IDisposable
         var jobInfo = new JobInfo
         {
             pJobData = null,
-            executeDelegate = null,
+            pExecutionFunc = null,
 
             remainingBatches = 1,
             threadIndex = -1,
@@ -367,22 +367,22 @@ public unsafe sealed class JobScheduler : IDisposable
     /// Retrieves the current status of a job identified by the specified handle.
     /// </summary>
     /// <param name="handle">The handle representing the job whose status is to be retrieved. The handle must be valid.</param>
-    /// <returns>The current status of the job as a <see cref="JobStatus"/> value.
+    /// <returns>The current status of the job as a <see cref="JobState"/> value.
-    ///     Returns <see cref="JobStatus.Invalid"/> if the handle is invalid or the job does not exist.</returns>
+    ///     Returns <see cref="JobState.Invalid"/> if the handle is invalid or the job does not exist.</returns>
-    public JobStatus GetJobStatus(JobHandle handle)
+    public JobState GetJobStatus(JobHandle handle)
     {
         if (!handle.IsValid)
         {
-            return JobStatus.Invalid;
+            return JobState.Invalid;
         }
 
         ref var jobInfo = ref _jobInfoPool.GetElementReferenceAt(handle._id, handle._generation, out var exist);
         if (!exist)
         {
-            return JobStatus.Invalid;
+            return JobState.Invalid;
         }
 
-        return (JobStatus)Volatile.Read(ref Unsafe.As<JobStatus, int>(ref jobInfo.status));
+        return (JobState)Volatile.Read(ref Unsafe.As<JobState, int>(ref jobInfo.state));
     }
 
     /// <summary>
@@ -399,7 +399,7 @@ public unsafe sealed class JobScheduler : IDisposable
         var spin = new SpinWait();
         while (_jobInfoPool.TryGetElement(handle._id, handle._generation, out var jobInfo))
         {
-            if (jobInfo.status == JobStatus.Completed)
+            if (jobInfo.state == JobState.Completed)
             {
                 return;
             }

Misaki.HighPerformance.Jobs/Misaki.HighPerformance.Jobs.csproj

@@ -6,6 +6,7 @@
     <ImplicitUsings>enable</ImplicitUsings>
     <Nullable>enable</Nullable>
     <AllowUnsafeBlocks>True</AllowUnsafeBlocks>
+		<IsPackable>false</IsPackable>
   </PropertyGroup>
 
   <ItemGroup>

Misaki.HighPerformance.Jobs/WorkerThread.cs

@@ -75,27 +75,14 @@ internal class WorkerThread : IDisposable
                 continue;
             }
 
-            //var handle = JobHandle.Invalid;
-
-            //// Always try the local thread and main thread queue first.
-            //if (!_localQueue.TryDequeue(out handle)
-            //    && !_scheduler.TryStealJob(-1, out handle))
-            //{
-            //    var randomIndex = _random.Next(0, _scheduler.WorkerCount);
-            //    if (_scheduler.TryStealJob(randomIndex, out var tempHandle))
-            //    {
-            //        handle = tempHandle;
-            //    }
-            //}
-
             //DoWork:
             var handle = FindJob();
             ref var jobInfo = ref _scheduler.GetJobInfoReference(handle, out var exist);
 
             if (exist)
             {
-                Interlocked.CompareExchange(ref jobInfo.status, JobStatus.Running, JobStatus.Scheduled);
+                Interlocked.CompareExchange(ref jobInfo.state, JobState.Running, JobState.Scheduled);
-                var executeDelegate = jobInfo.executeDelegate;
+                var executeDelegate = jobInfo.pExecutionFunc;
 
                 if (executeDelegate == null
                     || executeDelegate(jobInfo.pJobData, ref jobInfo.jobRanges, ref jobInfo.remainingBatches, _index))

Misaki.HighPerformance.LowLevel/Misaki.HighPerformance.LowLevel.csproj

@@ -5,6 +5,7 @@
     <ImplicitUsings>enable</ImplicitUsings>
     <Nullable>enable</Nullable>
     <AllowUnsafeBlocks>True</AllowUnsafeBlocks>
+		<IsPackable>false</IsPackable>
   </PropertyGroup>
 
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|AnyCPU'">

Misaki.HighPerformance.Mathematics.CodeGen/Generators/VectorGenerator.cs

@@ -1,6 +1,7 @@
 using Microsoft.CodeAnalysis;
 using System.Collections.Generic;
 using System.Linq;
+using System.Numerics;
 
 namespace Misaki.HighPerformance.Mathematics.CodeGen.Generators
 {
@@ -332,7 +333,7 @@ namespace Misaki.HighPerformance.Mathematics.CodeGen.Generators
             var asResult = $"As{typeSimpleName}()";
 
             // Add
-            sourceBuilder.AppendLine($@"
+            sourceBuilder.Append($@"
         {INLINE_METHOD_ATTRIBUTE}
         public static {typeName} operator +({typeName} lhs, {typeName} rhs)
         {{
@@ -350,7 +351,22 @@ namespace Misaki.HighPerformance.Mathematics.CodeGen.Generators
         public static {typeName} operator +({componentType} lhs, {typeName} rhs)
         {{
             return new {typeName}(lhs) + rhs;
-        }}");
+        }}
+
+#if NET10_0_OR_GREATER
+        // Use scaler here to let JIT handle the simd optimization since we can not do a in-place vectorlization manually.
+        {INLINE_METHOD_ATTRIBUTE}
+        public void operator +=({typeName} other)
+        {{");
+
+            for (var i = 0; i < typeInfo.Row; i++)
+            {
+                sourceBuilder.Append($@"
+                this.{_vectorComponents[i]} += other.{_vectorComponents[i]};");
+            }
+            sourceBuilder.AppendLine($@"
+        }}
+#endif");
 
             // Subtract
             sourceBuilder.AppendLine($@"
@@ -371,7 +387,22 @@ namespace Misaki.HighPerformance.Mathematics.CodeGen.Generators
         public static {typeName} operator -({componentType} lhs, {typeName} rhs)
         {{
             return new {typeName}(lhs) - rhs;
-        }}");
+        }}
+
+#if NET10_0_OR_GREATER
+        // Use scaler here to let JIT handle the simd optimization since we can not do a in-place vectorlization manually.
+        {INLINE_METHOD_ATTRIBUTE}
+        public void operator -=({typeName} other)
+        {{");
+
+            for (var i = 0; i < typeInfo.Row; i++)
+            {
+                sourceBuilder.Append($@"
+                this.{_vectorComponents[i]} -= other.{_vectorComponents[i]};");
+            }
+            sourceBuilder.AppendLine($@"
+        }}
+#endif");
 
             // Multiply
             sourceBuilder.AppendLine($@"
@@ -393,7 +424,22 @@ namespace Misaki.HighPerformance.Mathematics.CodeGen.Generators
         public static {typeName} operator *({componentType} lhs, {typeName} rhs)
         {{
             return new {typeName}(lhs) * rhs;
-        }}");
+        }}
+
+#if NET10_0_OR_GREATER
+        // Use scaler here to let JIT handle the simd optimization since we can not do a in-place vectorlization manually.
+        {INLINE_METHOD_ATTRIBUTE}
+        public void operator *=({typeName} other)
+        {{");
+
+            for (var i = 0; i < typeInfo.Row; i++)
+            {
+                sourceBuilder.Append($@"
+                this.{_vectorComponents[i]} *= other.{_vectorComponents[i]};");
+            }
+            sourceBuilder.AppendLine($@"
+        }}
+#endif");
 
             // Divide
             sourceBuilder.AppendLine($@"
@@ -415,7 +461,22 @@ namespace Misaki.HighPerformance.Mathematics.CodeGen.Generators
         public static {typeName} operator /({componentType} lhs, {typeName} rhs)
         {{
             return new {typeName}(lhs) / rhs;
-        }}");
+        }}
+
+#if NET10_0_OR_GREATER
+        // Use scaler here to let JIT handle the simd optimization since we can not do a in-place vectorlization manually.
+        {INLINE_METHOD_ATTRIBUTE}
+        public void operator /=({typeName} other)
+        {{");
+
+            for (var i = 0; i < typeInfo.Row; i++)
+            {
+                sourceBuilder.Append($@"
+                this.{_vectorComponents[i]} /= other.{_vectorComponents[i]};");
+            }
+            sourceBuilder.AppendLine($@"
+        }}
+#endif");
 
             // Modulus
             sourceBuilder.AppendLine($@"
@@ -435,7 +496,22 @@ namespace Misaki.HighPerformance.Mathematics.CodeGen.Generators
         public static {typeName} operator %({componentType} lhs, {typeName} rhs)
         {{
             return new {typeName}(lhs) % rhs;
-        }}");
+        }}
+
+#if NET10_0_OR_GREATER
+        // Use scaler here to let JIT handle the simd optimization since we can not do a in-place vectorlization manually.
+        {INLINE_METHOD_ATTRIBUTE}
+        public void operator %=({typeName} other)
+        {{");
+
+            for (var i = 0; i < typeInfo.Row; i++)
+            {
+                sourceBuilder.Append($@"
+                this.{_vectorComponents[i]} %= other.{_vectorComponents[i]};");
+            }
+            sourceBuilder.AppendLine($@"
+        }}
+#endif");
 
             // Unary operators
             sourceBuilder.AppendLine($@"
@@ -594,9 +670,18 @@ namespace Misaki.HighPerformance.Mathematics.CodeGen.Generators
         {INLINE_METHOD_ATTRIBUTE}
         public static global::System.Runtime.Intrinsics.Vector{_vectorBitsSize}<{componentType}> AsVector{_vectorBitsSize}(this {typeName} value)
         {{");
-            var fullTypeName = $"{typeInfo.TypePrefix}{typeInfo.Row + _missingComponents}";
+
+            if (typeInfo.Row == 4)
+            {
+                sourceBuilder.Append($@"
+            return global::System.Runtime.CompilerServices.Unsafe.BitCast<{typeName}, global::System.Runtime.Intrinsics.Vector{_vectorBitsSize}<{componentType}>>(value);");
+            }
+            else
+            {
+                sourceBuilder.Append($@"
+            return global::System.Runtime.Intrinsics.Vector{_vectorBitsSize}.Create({string.Join(", ", Enumerable.Range(0, typeInfo.Row + _missingComponents).Select(i => i < typeInfo.Row ? $"value.{_vectorComponents[i]}" : "1"))});");
+            }
             sourceBuilder.Append($@"
-            return global::System.Runtime.Intrinsics.Vector{_vectorBitsSize}.Create({string.Join(", ", Enumerable.Range(0, typeInfo.Row + _missingComponents).Select(i => i < typeInfo.Row ? $"value.{_vectorComponents[i]}" : "1"))});
         }}
 
         {INLINE_METHOD_ATTRIBUTE}

Misaki.HighPerformance.Mathematics/Misaki.HighPerformance.Mathematics.csproj

@@ -5,6 +5,9 @@
     <ImplicitUsings>enable</ImplicitUsings>
     <Nullable>enable</Nullable>
     <AllowUnsafeBlocks>True</AllowUnsafeBlocks>
+    <Title>$(AssemblyName)</Title>
+    <Authors>Misaki</Authors>
+		<IsPackable>false</IsPackable>
   </PropertyGroup>
 
   <ItemGroup>

Misaki.HighPerformance.Test/Benchmark/MathematicsBenchmark.cs

@@ -6,10 +6,10 @@ namespace Misaki.HighPerformance.Test.Benchmark;
 
 public class MathematicsBenchmark
 {
-    [Params(10, 100)]
+    [Params(10)]
     public int count = 10;
 
-    [Benchmark(Baseline = true)]
+    [Benchmark]
     public void Vector2Add()
     {
         var a = new Vector2(1, 2);
@@ -25,8 +25,8 @@ public class MathematicsBenchmark
     [Benchmark]
     public void Float2Add()
     {
-        var a = new float2(1);
+        var a = new float2(1, 2);
-        var b = new float2(5);
+        var b = new float2(5, 6);
         var result = new float2();
 
         for (var i = 0; i < count; i++)
@@ -34,4 +34,30 @@ public class MathematicsBenchmark
             result += a + b;
         }
     }
+
+    [Benchmark]
+    public void Vector4Add()
+    {
+        var a = new Vector4(1, 2, 3, 4);
+        var b = new Vector4(5, 6, 7, 8);
+        var result = new Vector4();
+
+        for (var i = 0; i < count; i++)
+        {
+            result += a + b;
+        }
+    }
+
+    [Benchmark]
+    public void Float4Add()
+    {
+        var a = new float4(1, 2, 3, 4);
+        var b = new float4(5, 6, 7, 8);
+        var result = new float4();
+
+        for (var i = 0; i < count; i++)
+        {
+            result += a + b;
+        }
+    }
 }

Misaki.HighPerformance.Test/Misaki.HighPerformance.Test.csproj

@@ -2,7 +2,7 @@
 
   <PropertyGroup>
     <OutputType>Exe</OutputType>
-    <TargetFramework>net9.0</TargetFramework>
+    <TargetFramework>net10.0</TargetFramework>
     <ImplicitUsings>enable</ImplicitUsings>
     <Nullable>enable</Nullable>
     <PublishAot>True</PublishAot>

Misaki.HighPerformance.Test/Program.cs

@@ -20,27 +20,4 @@
 
 using Misaki.HighPerformance.Test.Benchmark;
 
-//BenchmarkDotNet.Running.BenchmarkRunner.Run<ParallelNoiseBenchmark>();
+BenchmarkDotNet.Running.BenchmarkRunner.Run<MathematicsBenchmark>();
-
-var benchmark = new ParallelNoiseBenchmark();
-var sw = new System.Diagnostics.Stopwatch();
-
-benchmark.Setup();
-
-for (var i = 0; i < 1024; i++)
-{
-    benchmark.JobSystem();
-}
-
-sw.Start();
-
-for (var i = 0; i < 1024; i++)
-{
-    benchmark.JobSystem();
-}
-
-sw.Stop();
-
-benchmark.Cleanup();
-
-Console.WriteLine($"JobSystem: {sw.Elapsed.TotalMilliseconds / 1024.0} ms");

Misaki.HighPerformance/Misaki.HighPerformance.csproj

@@ -4,6 +4,7 @@
     <TargetFramework>net9.0</TargetFramework>
     <ImplicitUsings>enable</ImplicitUsings>
     <Nullable>enable</Nullable>
+		<IsPackable>false</IsPackable>
   </PropertyGroup>
 
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|AnyCPU'">