Reduce LogEventLevel boxing

[Pliner]

Hi!

What issue does this PR address?

During the profiling I found that LogEventLevel was boxed during formatting. You can see the evidence of the boxing in the screenshot :

Does this PR introduce a breaking change?
Nope

Please check if the PR fulfills these requirements

• The commit follows our guidelines
• Unit Tests for the changes have been added (for bug fixes / features)

Other information:

[nblumhardt]

Nice spotting 👍 - thanks for raising this, and sorry about the slow response over the end-of-year holiday period.

Did you collect any before/after performance numbers? There are some benchmarks in the repository that should give a decent basis for measurement.

[Pliner]

Here are two benchmark results: the baseline(commit hash 5c3a782) and the PR.

As far as I understand cases of the benchmark, the most relevant is OutputTemplateRenderingBenchmark.

net46

BenchmarkDotNet=v0.10.6, OS=Windows 10 Redstone 2 (10.0.15063)
Processor=Intel Core i7-6700 CPU 3.40GHz (Skylake), ProcessorCount=8
Frequency=3328125 Hz, Resolution=300.4695 ns, Timer=TSC
  [Host]     : Clr 4.0.30319.42000, 32bit LegacyJIT-v4.7.3190.0
  DefaultJob : Clr 4.0.30319.42000, 32bit LegacyJIT-v4.7.3190.0

Method	Mean	Error	StdDev	Gen 0	Allocated
FormatToOutput	2.280 us	0.0203 us	0.0180 us	0.1144	492 B

Method	Mean	Error	StdDev	Gen 0	Allocated
FormatToOutput	1.729 us	0.0077 us	0.0060 us	0.1068	456 B

netcoreapp2.0

BenchmarkDotNet=v0.10.6, OS=Windows 10 Redstone 2 (10.0.15063)
Processor=Intel Core i7-6700 CPU 3.40GHz (Skylake), ProcessorCount=8
Frequency=3328125 Hz, Resolution=300.4695 ns, Timer=TSC
dotnet cli version=2.1.4
  [Host]     : .NET Core 4.6.26614.01, 64bit RyuJIT
  DefaultJob : .NET Core 4.6.26614.01, 64bit RyuJIT

Method	Mean	Error	StdDev	Gen 0	Allocated
FormatToOutput	1.315 us	0.0236 us	0.0221 us	0.1163	488 B

Method	Mean	Error	StdDev	Gen 0	Allocated
FormatToOutput	1.221 us	0.0116 us	0.0109 us	0.1087	464 B

It seems there is no perfomance degradation 😅

PS Also here are full results.

baseline.zip
pr.zip

[Pliner]

@nblumhardt Is it possible to include the PR to 2.8 release?

[nblumhardt]

Thanks for the nudge @Pliner - sorry about the slow turnaround on these. Optimisations especially need a close look and I am still a little way from having time to dig into these in detail. Should be low risk to roll them into a 2.8.1, though!

[Pliner]

Thanks!