This repository holds a number of MLKit benchmark programs. The repository makes use of the kitbench tool provided by the MLKit compiler framework. This tool makes use of the environment variable MLKIT_ROOT for pointing to the root directory of the MLKit repository holding the mlkit executable and libraries.

To run a number of benchmarks, execute the following command:

$ make clean bench

This command will compile and execute a number of benchmark programs (with and without generational garbage collection enabled). Each executable will be executed 10 times and the report bench.html will show average numbers over the 10 runs.

For details about the use of the kitbench tool, execute:

$ kitbench -help