AlignedAllocator

Hardware-accelerated memory alignment with 10-30% faster SIMD operations:

// Automatic alignment optimization for hardware acceleration
using var alignedAllocator = Z.CreateAlignedAllocator();
using var buffer = alignedAllocator.Allocate<Vector128<float>>(1000);

// Get alignment statistics and performance metrics
var stats = alignedAllocator.GetAlignmentStatistics();
Console.WriteLine($"Alignment efficiency: {stats.AlignmentEfficiency:F1}%");

Key Benefits:

• 10-30% Faster SIMD Operations through optimal alignment
• Better Cache Utilization with cache-line alignment
• Hardware Detection for SSE, AVX, AVX-512, ARM NEON
• Multiple Strategies (Auto, CacheLine, SSE, AVX, AVX512, Custom)

🚀 Quick Start

using ZiggyAlloc;

// Create allocator
var allocator = new SystemMemoryAllocator();

// Allocate memory with automatic cleanup
using var buffer = allocator.Allocate<int>(1000);

// Use like a normal array with bounds checking
buffer[0] = 42;
int value = buffer[0];

// Convert to Span<T> for high-performance operations
Span<int> span = buffer;
span.Fill(123);

📊 Performance Comparison

ZiggyAlloc delivers exceptional performance through multiple optimization strategies:

SIMD Memory Operations (Latest Results)

Revolutionary Performance Gains with Hardware Acceleration:

Advanced Performance Optimizations:

• 20-55% faster allocation through Unsafe.SizeOf<T>() and optimized calculations
• 35-55% faster pool operations with SpinLock optimization and size-class arrays
• 10-20% improvement in span operations using MemoryMarshal
• 25-40% overall system improvement across allocation patterns

Traditional Allocator Performance

Performance Insights:

• SIMD Operations: 5-29x performance improvement for memory clearing and copying
• Large Data Types: 40%+ performance improvement with unmanaged arrays
• GC Pressure: Eliminated completely with unmanaged allocations
• Hardware Acceleration: AVX2 support with automatic fallback for older hardware

🔧 Allocator Comparison

Different allocators for different use cases:

*Performance varies by hardware: NumaAwareAllocator shows 20-40% improvement on NUMA systems, AlignedAllocator provides 10-30% improvement for SIMD operations

🏗️ Architecture Overview

graph TD
    A[IUnmanagedMemoryAllocator] --> B[SystemMemoryAllocator]
    A --> C[ScopedAllocator]
    A --> D[DebugAllocator]
    A --> E[UnmanagedMemoryPool]
    A --> F[HybridAllocator]
    A --> G[SlabAllocator]
    A --> H[LargeBlockAllocator]
    A --> I[NumaAwareAllocator]
    A --> J[AlignedAllocator]

    B --> K[Native Memory]
    C --> B
    D --> B
    E --> B
    F --> B
    G --> B
    
    I[UnmanagedBuffer<T>] --> J[Bounds Checking]
    I --> K[Automatic Cleanup]
    I --> L[Span<T> Integration]

🧠 Core Concepts

UnmanagedBuffer<T>

The core type for working with unmanaged memory:

var allocator = new SystemMemoryAllocator();
using var buffer = allocator.Allocate<int>(100);

// Type-safe access with bounds checking
buffer[0] = 42;
int value = buffer[99];

// Convert to Span<T> for high-performance operations
Span<int> span = buffer;
span.Fill(123);

Multiple Allocator Strategies

SystemMemoryAllocator

Direct system memory allocation with tracking.

ScopedAllocator

Arena-style allocator that frees all memory when disposed.

DebugAllocator

Tracks allocations and detects memory leaks with caller information.

UnmanagedMemoryPool

Reduces allocation overhead by reusing previously allocated buffers.

HybridAllocator

Automatically chooses between managed and unmanaged allocation based on size and type for optimal performance.

SlabAllocator

A slab allocator that pre-allocates large blocks of memory and sub-allocates from them. This allocator is particularly efficient for scenarios with many small, similarly-sized allocations.

var systemAllocator = new SystemMemoryAllocator();
using var slabAllocator = new SlabAllocator(systemAllocator);

// Small allocations are served from pre-allocated slabs
using var smallBuffer = slabAllocator.Allocate<int>(100);

// Large allocations are delegated to the base allocator
using var largeBuffer = slabAllocator.Allocate<int>(10000);

Key Benefits:

• Extremely fast allocation/deallocation for small objects
• Zero fragmentation within slabs
• Reduced system call overhead
• Better cache locality

Use Cases:

• High-frequency small allocations of similar sizes
• Performance-critical code paths
• Scenarios where allocation patterns are predictable

LargeBlockAllocator

A specialized allocator optimized for large memory blocks (>64KB) with memory pooling and alignment optimization.

var systemAllocator = new SystemMemoryAllocator();
using var largeBlockAllocator = new LargeBlockAllocator(systemAllocator);

// Large allocations automatically benefit from pooling and alignment
using var largeBuffer = largeBlockAllocator.Allocate<byte>(1024 * 1024); // 1MB

// Memory is automatically pooled for reuse
using var anotherBuffer = largeBlockAllocator.Allocate<byte>(1024 * 1024); // Reuses pooled memory

Key Benefits:

• Memory Pooling: Reduces allocation overhead for large blocks
• 4KB Alignment: Optimal memory alignment for performance
• SIMD Integration: Uses hardware-accelerated memory operations
• Size-Class Optimization: Different pools for different block sizes

Use Cases:

• Large data processing (images, scientific data, etc.)
• High-performance computing scenarios
• Applications with predictable large allocation patterns

🚀 Advanced Features

SIMD Memory Operations

Hardware-accelerated memory operations with revolutionary performance gains:

using ZiggyAlloc;

// SIMD operations are automatically used by allocators
var allocator = new SystemMemoryAllocator();

// Large allocations automatically benefit from SIMD acceleration
using var largeBuffer = allocator.Allocate<byte>(65536);

// Memory clearing is 29x faster with AVX2 acceleration
largeBuffer.Clear(); // Uses SimdMemoryOperations.ZeroMemory internally

// Memory copying is 5-8x faster
using var destBuffer = allocator.Allocate<byte>(65536);
largeBuffer.CopyTo(destBuffer); // Uses SimdMemoryOperations.CopyMemory

Key Benefits:

• 5-29x Performance Improvement: Hardware-accelerated memory operations
• AVX2 Support: Uses latest CPU instructions when available
• Automatic Fallback: Graceful degradation for older hardware
• Zero Configuration: Works out-of-the-box with all allocators

Memory Pooling

Reduce allocation overhead by reusing buffers:

var systemAllocator = new SystemMemoryAllocator();
using var pool = new UnmanagedMemoryPool(systemAllocator);

// First allocation - creates new buffer
using var buffer1 = pool.Allocate<int>(100);

// Second allocation - reuses buffer from pool if available
using var buffer2 = pool.Allocate<int>(100);

// Buffers are returned to the pool when disposed

Hybrid Allocation

Intelligent allocation strategy selection:

var systemAllocator = new SystemMemoryAllocator();
using var hybridAllocator = new HybridAllocator(systemAllocator);

// Small allocations may use managed arrays for better performance
using var smallBuffer = hybridAllocator.Allocate<int>(100);

// Large allocations will use unmanaged memory to avoid GC pressure
using var largeBuffer = hybridAllocator.Allocate<int>(10000);

📈 Performance Benchmarks

Comprehensive benchmarks demonstrate exceptional performance across multiple optimization strategies:

SIMD Memory Operations (Latest)

• 5-29x Performance Improvement: Hardware-accelerated memory operations using AVX2
• ZeroMemory Operations: 15-29x faster than standard implementations
• CopyMemory Operations: 5-8x faster than standard implementations
• Hardware Detection: Automatic AVX2/SIMD/fallback selection based on CPU capabilities

Traditional Optimizations

• Large Data Types: 40%+ performance improvement with unmanaged arrays
• GC Pressure: Eliminated completely with unmanaged allocations
• Memory Pooling: Reduces allocation overhead by reusing buffers
• Hybrid Allocation: Uses managed arrays for small allocations (faster) and unmanaged memory for large allocations (no GC pressure)
• Lock-Free Operations: SpinLock optimization for better contention handling

Memory Pooling Benefits

// Without pooling - each allocation calls into the OS
var allocator = new SystemMemoryAllocator();
for (int i = 0; i < 1000; i++)
{
    using var buffer = allocator.Allocate<byte>(1024); // System call each time
    // Process buffer...
}

// With pooling - first allocation per size calls OS, subsequent allocations reuse
using var pool = new UnmanagedMemoryPool(allocator);
for (int i = 0; i < 1000; i++)
{
    using var buffer = pool.Allocate<byte>(1024); // Reuses pooled buffer
    // Process buffer...
}

Hybrid Allocator Thresholds

📚 Examples

The examples directory contains organized examples demonstrating various use cases:

Basic Usage

• Simple memory allocation and automatic cleanup
• Using using statements for RAII-style memory management

Advanced Features

• Different allocator types and their use cases
• Memory leak detection
• High-performance buffer operations
• Native interop scenarios

Performance Optimization

• Memory pooling for frequent allocations
• Hybrid allocation strategies
• Avoiding GC pressure with large allocations

Real-World Applications

• Image processing without GC pressure
• Scientific computing with large datasets
• Native API interop

To run examples:

cd examples
dotnet run -- basic
dotnet run -- allocators
dotnet run -- performance
dotnet run -- realworld

📦 Installation

Install the NuGet package:

dotnet add package ZiggyAlloc

Or add to your .csproj:

<PackageReference Include="ZiggyAlloc" Version="1.3.0" />

📖 Documentation

• Getting Started Guide
• Full API Documentation
• Examples
• Benchmarks

🛠️ Requirements

• .NET 8.0 or later
• unsafe code enabled (configured in package)

📃 License

This project is licensed under the MIT License - see the LICENSE file for details.