Less is More dataset
====================

Overview of files:

Two zip-compressed archives each containing 71 CSV files holding data for experiments on a given architecture.
- Final_Measurements_CortexM0.zip: experiments on Cortex M0 using LLVM3.8
- Final_Measurements_CortexM3.zip: experiments on Cortex M3 using LLVM5.0

CSV files:

There is one CSV file for each of the 71 benchmarks studied in the paper. Each file has the following columns:
- ID which represents a tested configuration, see below
- Executables code size in KB
- Energy consumption in Joules
- Execution time measured in Seconds

The ID is computed as follows. From the first ID (ID 0), the ID represents the list of flags of -O0 and then each next consecutive ID adds the next optimization flag that exists in the -O2 list of flags. The last ID in the file represents the complete -O2 list of flags for the specified version of the compiler. For the Cortex-M0 the compiler version is LLVM3.8 and there are 49 optimization configurations tested, and for the Cortex-M3 the compiler version is LLVM5.0 and there are 65 configurations tested.