Tweaking Ubuntu 17.10 To Try To Run Like Clear Linux

First up was the Parboil benchmark where Clear Linux was about 7% faster out-of-the-box than Ubuntu 17.10 freshly installed. Switching to the performance governor from powersave had shaved off a few seconds, but the other tweaks hadn't caused any difference in performance.

With the CUTCP (Distance-Cutoff Coulombic Potential) test it's a much quicker test where one might think with it taking just a few seconds the powersave vs. performance governor difference might be more noticeable, but that was not the case.

With the Rodinia benchmark, Clear Linux was more than twice as fast as Ubuntu 17.10 out-of-the-box and none of the tweaks had shorten the run-time for this scientific test.

With the Numpy Python benchmark obviously the compiler flag changes won't have any impact on the performance of the distribution-provided Python package, but with the tests being automated, they were run for each configuration anyhow. For Ubuntu 17.10 there wasn't a significant difference in performance.

In not all tests does Clear Linux run faster, in some workloads the performance is quite similar to Ubuntu 17.10 depending upon the conditions. In the case of x264, their performance-sensitive code paths are already hand-tuned Assembly.