Benchmarks

Environment

OS	Windows 10 Pro
Processor	11th Gen Intel(R) Core(TM) i7-1165G7 @ 2.80GHz
RAM	16,0 GB
Disk	SSD

Benchmark runs across different scopes of java bytecode with following meanings:

• runtime - only java runtime without any additional dependencies
• runtime + guava - java runtime with one jar for guava
• runtime + project classpath - java runtime with all visible dependencies of JacoDB project
• runtime + Idea community - java runtime with all visible dependencies of IDEA community project

JacoDB benchmarks also include scheme when Usages feature is installed

JCDB

./gradlew jcdbBenchmark

Benchmark	Repeats	Avg time per operation
runtime	5	741 ms
runtime + guava	5	740 ms
runtime + project dependencies	5	1034 ms
runtime + IDEA community dependecies	5	2324 ms
runtime + Usages	5	896 ms
runtime + guava + Usages	5	929 ms
runtime + project dependencies + Usages	5	1298 ms

Soot

Run

./gradlew sootBenchmark

Benchmark	Repeats	Avg time per operation
runtime	5	827 ms
runtime + guava	5	733 ms
runtime + project dependencies	5	1839 ms
runtime + IDEA community dependecies	5	10091 ms

Soot comparison and differences

Soot just read all jar files and stores bytecode in memory. Jcdb works in a bit different way. Jcdb read available .class from jars and build folders in a parallel and as a result is faster in a situation when there are a lot of jar files (because of parallel execution). Keep in mind that Jcdb has a huge job processed in a background while almost all api is ready (hierarchy and usages requires background activity to be finished).

Jcdb performance for background activities without Usages feature:

./gradlew awaitBackgroundBenchmark

Benchmark	Repeats	Avg time per operation
runtime: wait for background jobs	5	3182 ms
runtime + project dependencies: wait for background jobs	5	6696 ms
runtime + IDEA community dependencies: wait for background jobs	2	76734 ms

For Idea Community code base result sqlite database file size is ~3.5Gb