As previously stated, the poll has been closed automatically a few hours before. Everybody has an access to the results.
The mysterious change was the fact that every component compiled as target thumb
(major portion of an android) has been optimized for speed (O3)
in build #1 (experimental), and optimized for size (Os)
in build #2 (normal behaviour).
Commits for reference:
Nearly everybody has pointed a finger on build #1 in terms of performance, and this was right, as build #1 was optimized for speed and it had to be faster, this was the main purpose of the test.
Battery is a matter of taste, I can see a draw between build #1 and #2. This is OK I think, one person reported battery drain due to no deepsleep in build #1, this may be possible due to totally different compilation approach, however both builds should have the same battery drain.
Stability was the nice question. Most of you have pointed a finger on build #1, and this is interesting because build #1 was the "experimental" build, and I expected totally different results. Most like a placebo here, because definitely stability would be better in build #2 or the same in both builds in best scenario.
Your choice was a matter of taste, same as the battery test. I think that most of you chose build #1 because of performance.
Thank you for participating in my mysterious builds test. I highly appreciate that. Now when you know the difference, feel free to make any additional tests if you wish