To run from project root directory:

sbt bench/bench

or from an sbt session:

geotrelllis> project bench
bench> bench

For running the benchmarks in a single file, you can use the benchOnly <path> input task *from the bench project, which supports tab completion:

geotrellis> project bench
bench> benchOnly benchOnly geotrellis/raster/GenericRasterBench.scala

Results will be in bench/target/jmh-results-<datestamp>.csv.

The configuration and execution of the the JMH task is implemented through a custom sbt plugin located at <base.dir>/project/GTBenchmarkPlugin.scala. It defines a number of settings that can be configured on a per-subproject basis:

• jmhOutputFormat: Output format: {text|csv|scsv|json|latex}
• jmhOutputDir: Directory for writing JMH results
• jmhFileRegex: Filename regular expression for selecting files to benchmark
• jmhThreads: Number of JMH worker threads
• jmhFork: How many times to fork a single JMH benchmark
• jmhIterations: Number of measurement iterations to do
• jmhWarmupIterations: Number of warmup iterations to do
• jmhTimeUnit: Benchmark results time unit: {m|s|ms|us|ns}
• jmhExtraOptions: Additional arguments to jmh:run before the filename regex

(For an up-to-date list of settings, refer to the plugin source.)

If you wish to bypass this managed configuration and execute the jmh command directly, you can do so with the jmh:run task. For example

jmh:run -t 1 -f 1 -wi 10 -i 10 .*Bench.*

See the sbt-jmh documentation for details on running directly.