Intervals.NET.Benchmarks.Benchmarks.FixedStepExtensionsBenchmarks-report-github.md

BenchmarkDotNet v0.13.12, Windows 10 (10.0.19045.6456/22H2/2022Update)
Intel Core i7-1065G7 CPU 1.30GHz, 1 CPU, 8 logical and 4 physical cores
.NET SDK 8.0.403
  [Host]     : .NET 8.0.11 (8.0.1124.51707), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI
  DefaultJob : .NET 8.0.11 (8.0.1124.51707), X64 RyuJIT AVX-512F+CD+BW+DQ+VL+VBMI

Method	Mean	Error	StdDev	Ratio	RatioSD	Completed Work Items	Lock Contentions	Allocated	Alloc Ratio
IntegerDomain_Span_Small	7.023 ns	0.4041 ns	1.1851 ns	1.00	0.00	-	-	-	NA
IntegerDomain_Span_Medium	8.150 ns	0.2118 ns	0.2521 ns	1.47	0.10	-	-	-	NA
IntegerDomain_Span_Large	7.989 ns	0.1333 ns	0.1113 ns	1.49	0.07	-	-	-	NA
DateTimeDomain_Span_Small	19.872 ns	0.4412 ns	0.5579 ns	3.57	0.21	-	-	-	NA
DateTimeDomain_Span_Medium	21.864 ns	0.2959 ns	0.2471 ns	4.07	0.18	-	-	-	NA
DateTimeDomain_Span_Large	20.113 ns	0.4554 ns	0.4472 ns	3.68	0.25	-	-	-	NA
IntegerDomain_Shift_Small	16.887 ns	0.2709 ns	0.2262 ns	3.14	0.15	-	-	-	NA
IntegerDomain_Shift_Medium	17.048 ns	0.3180 ns	0.3905 ns	3.06	0.20	-	-	-	NA
IntegerDomain_Shift_Large	17.088 ns	0.2378 ns	0.2108 ns	3.16	0.17	-	-	-	NA
DateTimeDomain_Shift	16.985 ns	0.2463 ns	0.2304 ns	3.12	0.20	-	-	-	NA
IntegerDomain_Expand_Small	17.560 ns	0.2967 ns	0.2776 ns	3.22	0.19	-	-	-	NA
IntegerDomain_Expand_Medium	17.956 ns	0.2634 ns	0.2199 ns	3.34	0.17	-	-	-	NA
IntegerDomain_Expand_Large	17.961 ns	0.2616 ns	0.2319 ns	3.32	0.18	-	-	-	NA
DateTimeDomain_Expand	17.393 ns	0.3220 ns	0.3579 ns	3.16	0.20	-	-	-	NA
IntegerDomain_ExpandByRatio_Small	23.665 ns	0.2922 ns	0.2440 ns	4.41	0.20	-	-	-	NA
IntegerDomain_ExpandByRatio_Medium	23.569 ns	0.2486 ns	0.2076 ns	4.39	0.19	-	-	-	NA
IntegerDomain_ExpandByRatio_Large	23.963 ns	0.3390 ns	0.3005 ns	4.43	0.24	-	-	-	NA
DateTimeDomain_ExpandByRatio	48.474 ns	0.6099 ns	0.5406 ns	8.96	0.48	-	-	-	NA
IntegerDomain_ExpandByRatio_Asymmetric	24.806 ns	0.3612 ns	0.3016 ns	4.62	0.20	-	-	-	NA
IntegerDomain_ExpandByRatio_Contraction	24.246 ns	0.3533 ns	0.3304 ns	4.45	0.27	-	-	-	NA

Summary

What This Measures

Fixed-step extension method performance—the high-level operations that combine domain logic with range manipulation. Tests Span, Shift, Expand, and ExpandByRatio across integer and DateTime domains with various range sizes.

Key Performance Insights

📏 Span: Constant-Time Range Measurement (O(1))

• Integer domain (all sizes): 7-8 ns
• DateTime domain (all sizes): 19-22 ns
• Result: Performance independent of range size (proves O(1))
• Use case: Count discrete points in a range

↔️ Shift: Move Range by Offset (O(1))

• Integer domain: 16.9-17.1 ns
• DateTime domain: 17.0 ns
• Result: Consistent across range sizes and domain types
• Use case: Move a time window forward/backward by N steps

📐 Expand: Extend Boundaries (O(1))

• Integer domain: 17.6-18.0 ns
• DateTime domain: 17.4 ns
• Result: Fixed cost regardless of expansion amount
• Use case: Add margin to a range (e.g., buffer time)

🔢 ExpandByRatio: Proportional Expansion (O(1))

• Integer domain: 23.6-24.0 ns (includes Span calculation)
• DateTime domain: 48.5 ns (heavier DateTime arithmetic)
• Asymmetric expansion: 24.8 ns (same cost)
• Contraction (negative ratio): 24.2 ns (same cost)
• Use case: Zoom in/out by percentage

Performance by Operation

Operation           Integer (ns)    DateTime (ns)    Complexity
───────────────────────────────────────────────────────────────
Span                7-8             19-22            O(1)
Shift               16.9-17.1       17.0             O(1)
Expand              17.6-18.0       17.4             O(1)
ExpandByRatio       23.6-24.8       48.5             O(1)

Range Size Independence

Critical insight: Performance is identical across small/medium/large ranges:

• Small range (10 steps): 7.0 ns
• Medium range (1000 steps): 8.2 ns
• Large range (1,000,000 steps): 8.0 ns

This proves O(1) complexity—the hallmark of fixed-step domains.

Design Trade-offs (Aligned with README Philosophy)

Why all operations are O(1):

• ✅ Fixed-step domains: Distance is arithmetic, not iteration
• ✅ Add is arithmetic, not loop-based
• ✅ Floor/Ceiling are alignment operations, not searches
• ✅ No need to enumerate range contents

Why ExpandByRatio is slower:

• Calls Span() first (to calculate ratio)
• Then calls Expand() with computed offsets
• Combined cost: Span + Expand (~7ns + 17ns = ~24ns)

Why DateTime is sometimes slower:

• DateTime arithmetic involves tick calculations
• More complex than integer addition/subtraction
• Still O(1), just more CPU instructions

Practical Recommendations

✅ All operations suitable for hot paths:

• Span: 7-22 ns (measure range size)
• Shift: ~17 ns (move time windows)
• Expand: ~18 ns (add margins)
• ExpandByRatio: 24-48 ns (proportional scaling)

✅ Use with confidence in loops:

for (int i = 0; i < 10000; i++)
{
    var span = range.Span(domain);      // 7-22 ns per iteration
    var shifted = range.Shift(domain, 5); // 17 ns per iteration
}
// Total: 170,000-240,000 ns = 0.17-0.24 milliseconds

✅ Range size doesn't matter:

• Processing [1, 10] or [1, 1000000] costs the same
• Choose range boundaries based on business logic, not performance

Real-World Impact

Maintenance Window Scheduling (1000 shifts/day):

Operation: range.Shift(dayDomain, offset)
Cost: 17 ns × 1000 = 17 microseconds/day
Negligible overhead

Analytics Dashboard (100 span calculations/second):

Operation: range.Span(intDomain)
Cost: 7 ns × 100 = 700 ns/second
Sub-microsecond impact

Dynamic Range Expansion (zoom controls, 60 FPS):

Operation: range.ExpandByRatio(domain, 0.1, 0.1)
Cost: 24-48 ns × 60 = 1.44-2.88 microseconds/second
Real-time performance guaranteed

Memory Behavior

All extension operations:  0 bytes allocated
Return value:              New Range<T> struct (stack-allocated)
Domain instances:          Reusable, no per-operation cost

Comparison: Fixed vs Variable-Step

These benchmarks show fixed-step extension performance (O(1)):

• Span: 7-22 ns
• All operations: Range-size independent

For variable-step extensions (business days, O(N)):

• See VariableStepDomainBenchmarks
• Span may require iteration: 52-3,600 ns depending on range size
• Performance scales with range duration

Why This Matters

These benchmarks prove that fixed-step extensions have guaranteed O(1) performance:

• ✅ Process any range size in constant time
• ✅ Predictable performance for real-time systems
• ✅ Zero allocations, suitable for hot paths
• ✅ No performance penalty for domain abstraction

Choose fixed-step domains when you need guaranteed constant-time operations.