Skip to content

Latest commit

 

History

History
103 lines (76 loc) · 4.09 KB

README.md

File metadata and controls

103 lines (76 loc) · 4.09 KB
Raw

Simple wasm benchmark sample app

Runs various performance measurements. It is intended as a quick tool to measure mono/wasm performance in dotnet/runtime repo, using in-tree runtime. Can be used to check performance impact of runtime changes and the summary is provided in .md markdown format, suitable for commit messages and PR descriptions.

Browser and console versions are available.

Running the benchmark

To run the benchmark on linux/mac:

> make build
> make run

can be used in the browser-bench/ and also in the browser-bench/Console/.

To run the benchmark on windows:

> dotnet build /t:RunSample

Blazor startup measurements

To run the benchmark with blazor startup measurements, set BlazorStartup property to true, like:

> dotnet build /t:RunSample /p:BlazorStartup=true

Additional build arguments

The benchmark project is built in a separate process, so to pass additional msbuild arguments, use BuildAdditionalArgs property, like:

> dotnet build /t:RunSample  /p:BuildAdditionalArgs="/p:WasmEnableSIMD=false"

Example console output:

> make run
console.debug: MONO_WASM: Initializing mono runtime
console.debug: MONO_WASM: ICU data archive(s) loaded, disabling invariant mode
console.info: Initializing.....
Benchmark started
Exceptions, NoExceptionHandling count: 8344090, per call: 6.807213249138013E-05ms, total: 0.568s
Exceptions, NoExceptionHandling count: 8500347, per call: 6.940893118833855E-05ms, total: 0.59s
...
.md
| measurement | time |
|-:|-:|
|        Exceptions, NoExceptionHandling |     0.0680us |
|                   Exceptions, TryCatch |     0.0723us |
|              Exceptions, TryCatchThrow |     0.0035ms |
|             Exceptions, TryCatchFilter |     0.0848us |
|       Exceptions, TryCatchFilterInline |     0.0659us |
|        Exceptions, TryCatchFilterThrow |     0.0046ms |
| Exceptions, TryCatchFilterThrowApplies |     0.0036ms |
|         Json, non-ASCII text serialize |    15.3855ms |
|       Json, non-ASCII text deserialize |    24.7299ms |
|                  Json, small serialize |     0.2980ms |
|                Json, small deserialize |     0.5080ms |
|                  Json, large serialize |    95.3333ms |
|                Json, large deserialize |   141.4737ms |

The .md output will look like this:

measurement time
Exceptions, NoExceptionHandling 0.0680us
Exceptions, TryCatch 0.0723us
Exceptions, TryCatchThrow 0.0035ms
Exceptions, TryCatchFilter 0.0848us
Exceptions, TryCatchFilterInline 0.0659us
Exceptions, TryCatchFilterThrow 0.0046ms
Exceptions, TryCatchFilterThrowApplies 0.0036ms
Json, non-ASCII text serialize 15.3855ms
Json, non-ASCII text deserialize 24.7299ms
Json, small serialize 0.2980ms
Json, small deserialize 0.5080ms
Json, large serialize 95.3333ms
Json, large deserialize 141.4737ms

Multiple results can be also easily combined together in text editor to have a table with measurements next to each other for comparison.

Filter jobs/measurements

The -t option can be used to run subset of jobs/measurements like this, 1st line linux/mac, 2nd line windows:

> make run ARGS="-t Json:non,Exceptions:Inline"
> dotnet build /v:n /t:RunSample /p:Args="-t Json:non%2cExceptions:Inline"

Note the escaped , character (%2c) in the dotnet property above.

Console options

> make run ARGS=-h
...
Simple mono wasm benchmark

Copyright 2021 Microsoft Corporation

Options:
  -h, --help, -?             Show this message and exit
  -t, --tasks=VALUE          Filter comma separated tasks and its measurements
                               matching, TASK[:REGEX][,TASK[:REGEX],...].
                               Example: -t Json:non,Exceptions:Inline