1. Before calibration, ensure that the CPU is running at maximum frequency and set the governor to "performance" if necessary.
2. Run the calibration at least two times to check if the same values are being obtained. From the calibration, the value to be used with the -c option is the "Time per iteration" from the output. For the sleep calibration, the value of "Scaling factor" can be used. Ideally this should be close to 1.
3. When the actual run is made, make sure that the loop_length is close to the time per loop shown. This can be tried with different combinations, but at least full CPUness MEMness and IOness have to be verified. If it is not, this is most probably because of the "per-loop" time being too large which means that the 100%-loop count does not take 1ms as it should.
4. If PAPI is not installed on your system, before making please comment out the $LIB line in the Makefile. I haven't gotten down to writing a configure for this yet.

-C, --cpu
	Specifies the desired CPUness.

-M, --mem
	Specifies the desired MEMness.

-l, --loop_length
	Specifies the smallest loop length in microseconds.

-n, --loops
	Specifies the number of times to iterate over the loop.

-c, --cpu_calib
	If calibration is not desired, then these values have to be
	specified. This is the result of calibration.

-m, --mem_calib
	If calibration is not desired, then these values have to be
	specified. This is the result of calibration.

-i, --sleep_calib
	If calibration is not desired, then these values have to be
	specified. This is the result of calibration.

--calibrate
	Perform calibration before running the actual loops. This will
	calculate how many internal loops are required for 100% of each
	of CPUness, MEMness and IOness for the given loop length.

--no_papi
	Do not initialize or use the PAPI library. If the PAPI library
	has not been installed on the system, then this is a required
	option. The benchmark assumes that PAPI is enabled by default.

--papi_periodic
	Use this option if the benchmark is being used as a target 
	workload. This requies PAPI to be installed. It initializes
	the library and periodically prints out any value that needs to
	be monitored.

./wileE --calibrate

Starting Calibration:

CPU Calibration: Working set size used: 1024 Time per iteration: 12.103000 Number of loops for 100% CPUness: 82 Time for above loops: 988 L2DCM:13 for 1000 loops L2DCM:1 for 82 loops Randomizing...44739242

Memory Calibration: Working set size used: 128 Time per iteration: 20.566000 Number of loops for 100% Memory-ness: 48 Time for above loops: 1001 L2DCM:326456 for 1000 loops L2DCM:15658 for 48 loops

IO Calibration: 1000: 1038 11000: 11027 21000: 21028 31000: 31027 41000: 41028 51000: 51027 61000: 61028 71000: 71028 81000: 81029 91000: 91027 Scaling factor used: 1.000297

./wileE -C 0.0 -M 1.0 -l 100000 -n 100 -c 12.103 -m 20.566 -i 1.00 --no_papi