An Experiment to observe Linux reported load averages Vs. Actual loads under high frequency high load conditions. WWW.Smythies.com
This web page provides the details for the data that gives this graph:

The graph is actually data from 4 experiments done on a Ubuntu server 12.04 running two different kernels, as specified in the legend.
To avoid discontinuities in the reported load averages, and to maintain predictable execution loop times in the main program, the cpu frequency governors were set to powersave mode. However, note that the execution time predicability is problematic at high frequencies and high loads. Therefore actual frequncies and loads were recorded during the experiments, which is why some entires look at bit odd and why the results are listed as percentage errors.
The main program, as a text file.
The program, as a text file, that generates the script for execution conditions of the main program for one of the 4 experiments. (the other scripts were manually edited fromthe output of this.)
One script, as a text file, that was used to run the main program under the desired conditions for that experiment.
The raw data as a text file for the 3 processes Kernel 3.5RC2 line
The raw data as a text file for the 8 processes Kernel 3.5RC2 line
The raw data as a text file for the 3 processes Kernel peter35 line
The raw data as a text file for the 8 processes Kernel peter35 line
The following are large files (~~0.5 to 1 megabyte)
The oversampled data as a text file. For the 3 processes, Kernel 3.5RC2 line
The oversampled data as a text file. For the 8 processes, Kernel 3.5RC2 line
The oversampled data as a text file. For the 3 processes, Kernel peter35 line
The oversampled data as a text file. For the 8 processes, Kernel peter35 line