rebench.conf

# Config file for ReBench
# Config format is YAML (see http://yaml.org/ for detailed spec)

# this run definition will be chosen if no parameters are given to rebench.py
default_experiment: Test
default_data_file:  'test.data'

# general configuration for runs
runs:
    iterations: 100
    invocations: 1
    # min_iteration_time:         100    # give a warning if average runtime is below this value

# definition of benchmark suites
# settings in the benchmark suite will override similar settings of the executor
benchmark_suites:
    TestSuite1:
        gauge_adapter: Test
        # location: /Users/...
        command: TestBenchMarks %(benchmark)s %(input)s %(variable)s something-else
        input_sizes: [1, 2, 10, 100, 1000]
        benchmarks:
            - Bench1
            - Bench2:
                extra_args: 6
        max_invocation_time: 300
        variable_values: # this is an other dimension, over which the runs need to be varied
            - val1
            - val2
    TestSuite2:
        gauge_adapter: Test
        command: TestBenchMarks %(benchmark)s %(input)s %(variable)s another-thing
        input_sizes: [1, 2, 10, 100, 1000]
        cores: [7, 13, 55]
        benchmarks:
            - Bench1:
                extra_args: 3
            - Bench2
        variable_values: # this is an other dimension, over which the runs need to be varied
            - val1
            - val2

# Executors have a name and are specified by a path and the executable to be executed
# optional: the number of cores for which the runs have to be executed
executors:
    TestRunner1:
        path: tests
        executable: test-vm1.py
        cores: [1, 4, 8]
    TestRunner2:
        path: tests
        executable: test-vm2.py

# define the benchmarks to be executed for a re-executable benchmark run
# special definitions done here should override benchmark suite definitions, and
# executor definitions
experiments:
    Test:
        description: >
            This run definition is used for testing.
            It should try all possible settings and the generated out
            will be compared to the expected one by the unit test(s)
        suites:
            - TestSuite1
            - TestSuite2
        input_sizes: [1]
        executions:
            # List of executors and Benchmarks/Benchmark Suites to be run on them
            # benchmarks define here will override the ones defined for the whole run

            #the following example is equivalent to the global run definition,
            #but needs to be tested...
            - TestRunner1:
                suites:
                  - TestSuite1
            - TestRunner1:
                suites:
                  - TestSuite2
            - TestRunner2