StopwatchBenchmarks

Quick and dirty solution for benchmarking this dotnet PR.

Has 4 benchmarks:

• ElapsedBenchmark - calling the ElapsedXXX properties both when the stopwatch is running and when stopped
• RandomizedElapsedBenchmark - similar to ElapsedBenchmark but over an array of stopwatches which are randomly either running or stopped
  • this aims to confuse any branch prediction that's going on
• StartNewStopBenchmark - calls StartNew() followed by Stop
• StartStopBenchmark - calls Start() followed by Stop on pre-allocated stopwatches

Main goal is to reduce the allocation of size of stopwatches, which is best illustrated by StartNewStopBenchmark. All other benchmarks aim for minimal regressions in performance.

Results

Environment

BenchmarkDotNet=v0.13.1, OS=Windows 10.0.22572
AMD Ryzen 7 Microsoft Surface Edition, 1 CPU, 16 logical and 8 physical cores
.NET SDK=6.0.200
  [Host]     : .NET 6.0.2 (6.0.222.6406), X64 RyuJIT
  DefaultJob : .NET 6.0.2 (6.0.222.6406), X64 RyuJIT

ElapsedBenchmark

Method	LeaveRunning	Mean	Error	StdDev	Ratio	Allocated
Current	False	2.556 μs	0.0188 μs	0.0166 μs	1.00	-
Updated	False	2.566 μs	0.0262 μs	0.0245 μs	1.00	-
Current	True	56.362 μs	0.0158 μs	0.0132 μs	1.00	-
Updated	True	56.153 μs	0.0370 μs	0.0328 μs	1.00	-

RandomizedElapsedBenchmark

Method	Mean	Error	StdDev	Ratio	Allocated
Current	9.194 μs	0.0050 μs	0.0042 μs	1.00	-
Updated	9.202 μs	0.0302 μs	0.0283 μs	1.00	-

StartNewStopBenchmark

Method	Mean	Error	StdDev	Ratio	Gen 0	Allocated
Current	45.94 μs	0.354 μs	0.314 μs	1.00	19.1040	39 KB
Updated	44.76 μs	0.078 μs	0.069 μs	0.97	11.4746	23 KB

StartStopBenchmark

Method	Mean	Error	StdDev	Ratio	Allocated
Current	41.91 μs	0.028 μs	0.022 μs	1.00	-
Updated	44.33 μs	0.144 μs	0.135 μs	1.06	-