Nexus 6P Benchmarks - Pt. 3

Austin Haedicke bio photo By Austin Haedicke Comment
Part 1 Part 2 Part 3 Part 4

Review:

After rooting my Nexus 6P I began collecting data on various ROMs and schedulers. Specifically, the data I looked at were:

  • The average score between: AnTuTu Benchmark, GeekBench (single), GeekBench (multi), and Quadrant.
  • Estimated maximum screen on time (SOT) as calculated by average screen on percentage in trials multiplied by estimated maximum battery life.

My results thus far have been compiled on this Google Sheet. The third sheet (I/O Compare) is for comparing I/O schedulers only. The fourth sheet (Profile Data) is a beta page for the data I’m using to make a custom Interactive governor profile (post forthcoming – I know there are many publised settings, but I wanted to roll-my-own).

For further information on the process thus far, please see the previous Part One and Part Two posts.


Kernel Testing:

I had to start somewhere and obviously can’t cover every possible combination of ROM + Kernel + Scheduler + Settings. So, the ROMs selected for kernel testing were Chroma and Slim. I based this on a combination of personal preference and (average) benchmark scores. To narrow the list, in order to be considered the ROM (and respective configurations) needed to score +1 standard deviation above the mean benchmark score for all tests.

After polling the Google + Community, I began testing Slim and Chroma with the ElementalX, Franco, and Kylo kernels. Benchmarks were ran and the top scorer (plus some others) for each ROM/Kernel was then run through a ~50 hour real-world-use trial where I noted uptime, percent battery used, average CPU frequency (LITTLE only), and percent deep sleep time.

Results:

Both Chroma and Slim seemed to benefit in performance from Kylo and Franco kernels in benchmark scores. Chroma faired better with Kylo while Slim worked better with Franco. ElementalX seemed to perform underwhelmingly. I did try to overclock ElementalX, but the performance boost was negligible, so I nixed it.

ElementalX did give Slim an almost +1 SD benefit in terms of estimated SOT. Even more impressive though, Kylo and Chroma together churned out an estimated 5.8 hours of SOT (that’s average over 50 hours of on-battery run time!). A distant second, but there was improvement for Slim when using ElementalX, which brought 4.9 hours of SOT (stil +1 SD above the mean, very good).

Going Forward:

I’ve already started tuning a custom profile for the Interactive governor based on my usage (as you can see on the linked spreadsheet). I will pick two (so that there can be comparison) ROM/Kernel combinations to test with the following Interactive profiles:

Thanks for reading, stay tuned!

NOTE:  In the interest of time I'm trying to wrap my head
around the Interactive governor and may or may not get to
ElementalX, which I realize is kind of the point of naming
an entire kernel after a featured governor.
comments powered by Disqus