Benchmark module

This module contains benchmarks written using JMH from OpenJDK.

Preparation uisng the Ghent PDF Output SuiteV50

As the Ghent PDF Output SuiteV50 can not be downloaded automatically as a license has to be agreed on the following steps need to be done prior to running Rendering.renderGhentCMYK:

select Download on the download page
unpack the ZIP file in the directory target/pdfs so that the directory Ghent_PDF_Output_Suite_V50_Full is on top
keep only Ghent_PDF_Output_Suite_V50_Full/Categories/1-CMYK/Test pages/Ghent_PDF-Output-Test-V50_CMYK_X4.pdf

Running benchmarks

The default behavior is to run all benchmarks:

java -jar target/benchmarks.jar

List all available benchmarks:

java -jar target/benchmarks.jar -l

Pass a pattern or name after the command to select the benchmarks:

java -jar target/benchmarks.jar LoadAndSave.loadMediumFile

Check which benchmarks match the provided pattern:

java -jar target/benchmarks.jar -l LoadAndSave

Run a specific test and override the number of forks, iterations and warm-up iteration to 2:

java -jar target/benchmarks.jar  -f 2 -i 2 -wi 2 LoadAndSave.loadMediumFile

Get a list of available profilers:

java -jar target/benchmarks.jar -lprof

The following sections cover async profiler and GC profilers in more detail.

Using JMH with async-profiler

JMH includes async-profiler. After download run JMH using the async-profiler:

java -jar target/benchmarks.jar -prof async:libPath=/path/to/libasyncProfiler.so

With flame graph output (the semicolon is escaped to ensure it is not treated as a command separator):

java -jar target/benchmarks.jar -prof async:libPath=/path/to/libasyncProfiler.so\;output=flamegraph

To get help on options to be used for the async-profiler use the following command:

java -jar target/benchmarks.jar -prof async:help

Using JMH with GC profiler

To measure the allocation rate run the benchmark with -prof gc:

java -jar target/benchmarks.jar -prof gc

For profiling the norm alloc rates are important as this which measure the allocations per operation rather than allocations per second which can increase/decrease with faster/slower code.

Writing benchmarks

Examples for writing JMH tests are available from the projects samples provided by the JMH project.

Tutorials are available at