Refresh of this old benchmark:
https://osu-ppy-sh.jeuxcrack.net/forum/p/3908475
This time the timeinterval of the statistics are almost perfectly equal and syncronized.
What I noticed new:
Fullscreen reduced the time/amount of slowest frames. 0.1% slowest frames average 3.78ms vs 4.66ms. BANZAI
Cutting edge is slowest than beta, couldn't think of any reason. 0.1% slowest frames average 4.19ms vs 3.85ms or avg 1.83ms vs 1.56ms.

???
DX performace is almost the same, that is dx vs opengl is now:
0.1% slowest frames average 3.24ms vs 3.85ms or avg 0.96ms vs 1.56ms
Ignoring the fullscreen improvement everything seems the same.
2015-02-23 open gl not full screen
http://puu.sh/hRGc5/08781f3fd5.png
2015-05-18 opengl beta fullscreen
http://puu.sh/hRGaf/39ddf70734.png
2015-05-18 opengl edge fullscreen
http://puu.sh/hRGby/803e362a75.png
2015-02-23 DX not fullscreen
http://puu.sh/hRGfl/ec874f239a.png
2015-05-18 DX not fullscreen
http://puu.sh/hRGgp/7ae203cba4.png
I tested frame limiter now... For the porpuse of checking if the slowest 0.1% frames improved.
double screen refresh rate
http://puu.sh/hRGj7/6977631ba6.png
I noticed the 1ms intervals... and reminded of windows timer

. Checked it and it was at 1ms. After I forced to 0.5ms:
double screen refresh rate 2
http://puu.sh/hRGok/0900a72a3f.png
Not sure if exists a real difference or it is just mesurement issues.
Oced my u by 1.097 and got 1.100 more average frames lol 0.1 percentil only improved by 1.031 times.
Tested the new Vsync that now really vsyncs beyond 65hz~. It was A LOT laggier than just frame limiting to 144fps. Any possibility of a low latency V-sync?