WorkflowForge Internal Benchmarks

This document presents WorkflowForge’s internal performance benchmarks - comprehensive self-testing that validates the framework’s performance characteristics independently of competitor comparisons.

Test System: Windows 11 (25H2), Intel 11th Gen i7-1185G7, .NET 8.0.23
Benchmark Framework: BenchmarkDotNet v0.15.8
Methodology: 50 iterations per benchmark, 5 warmup iterations
Last Updated: January 2026

Table of Contents

• Executive Summary
• Operation Performance
• Workflow Throughput
• Memory Allocation
• Concurrency Scaling
• Key Insights

Executive Summary

WorkflowForge internal benchmarks demonstrate (50 iterations):

Operation Performance

Tests individual operation types for execution time and memory allocation.

Operation Execution (Median Times)

Operation Creation (Median Times)

Key Findings:

• Custom operations are the most memory-efficient (296 B)
• Logging operations are fastest (10.3μs)
• Operation creation is extremely fast (1.5-1.8μs)
• Exception handling adds minimal overhead

Workflow Throughput

Tests complete workflow execution patterns with varying operation counts.

Custom Operations vs Delegate Operations (Median Times)

*Delegate operations include 15ms delay per operation for testing purposes.

Workflow Patterns (10 Operations, Median Times)

*Includes built-in delays for realistic simulation.

Key Findings:

• ForEach loops are highly optimized (57.6μs for 10 items)
• Memory scales linearly with operation count
• Custom operations are ~1000x faster than delegate with delays

Memory Allocation

Tests memory allocation patterns and GC behavior.

Allocation Patterns (Median Times, 50 Allocations)

Memory Scaling

Key Findings:

• Minimal allocation baseline is constant at 3.04 KB
• No GC pressure for typical workflows (Gen0 only)
• Large object allocations trigger full GC (Gen0/1/2)
• StringBuilder optimization saves ~20KB vs concatenation at 50 allocations

Concurrency Scaling

Tests concurrent workflow execution patterns.

Scaling Factor (10 Operations per Workflow)

Concurrency Patterns (8 Workflows, 10 Operations)

Key Findings:

• Near-linear scaling up to 16 concurrent workflows
• Consistent memory overhead per workflow (~18-19 KB)
• Shared resource patterns maintain good performance
• High contention scenarios serialize as expected

Key Insights

Performance Characteristics

1. Operation Execution: 10-67μs median for CPU-bound operations
2. Operation Creation: Sub-2μs overhead for all operation types
3. Memory Efficiency: 3.04 KB baseline, linear scaling
4. Concurrency: Near-perfect linear scaling (15.7x for 16 workflows)

Recommendations

1. Use Custom Operations for production - most efficient (296 B allocation)
2. Prefer ForEach loops - highly optimized (57.6μs for 10 items)
3. Avoid Large Object Allocations - triggers Gen2 GC
4. Scale Horizontally - linear scaling supports high throughput

Comparison with Competitive Benchmarks

These internal benchmarks validate the competitive benchmark findings:

Related Documentation

• Performance Overview - Summary and production targets
• Competitive Analysis - Head-to-head comparisons