(If my Eniglish is wrong, I'm sorry. I'm not good at Eniglish.)
I am considering separating this game into categories for Windows 7 and Windows 10. I would like to hear others' opinions on whether this decision is appropriate.
I am one of the moderators of this game. However, I am not familiar with the internal systems of this game. Sorry.
・Framerate rulings: https://www.speedrun.com/new_super_marisa_land/thread/h6gyz
There is a suggestion in the above thread that Win10 runs are slower than Win7 runs. The specific frame rate difference is not clear to me. However, I have compared several videos and there certainly seems to be a difference between them.
That is why I am considering separating the categories. (Please also see this thread: https://www.speedrun.com/new_super_marisa_land/thread/tcuzd )
From my time playing the game, it does seem to be a pretty notable difference. Even if someone were to perfectly replicate the world record, they would most likely be about 10-15 seconds behind for no good reason, and so separating depending on the OS seems fair in my opinion. It would be nice if there was some compromise method, like maybe we could just add a certain amount of time to each run to make up the gap, but it’s extremely variable depending on how much time you spend on each stage, so that's not really possible. Also, because the game displays as 60fps on either Windows 7 or Windows 10, it’s extremely hard to tell the exact amount of discrepancy or why it’s even happening at all.
残念ですけど差はすごすぎるので公平のために分けるしかないと思う。多分、全部のスピードランに十秒追加させられますっとか思ったんだけど各のスピードランを付ける方法に対する未知数が多すぎる仕方ない。
I am not familiar with the difference between win10 and win7. I'm sorry. Therefore, I will compare the submitted video with other videos and assign it to the category with the closer framerate.
So I decided to try figuring out where the time differences were. I figured it was probably load times, and I could rule out within-level framerate by counting how long it took for 100 ticks on the ingame timer to elapse. Doing some quick and dirty timing by hand with livesplit, I found that, looking at Ohon's 15:43 it takes about 48.X seconds for 100 ticks, while in PerroVegano's 12:02 it takes in the high 49-low 50 second range (and got low 50 on a live timer on my machine). More careful testing is needed, but I was unpleasantly surprised to see up to a 2% difference. (I'm also a little disappointed that it appears to run a little faster than a nice 0.5s/tick on Win7).
If this was just a load time issue we could probably find a way to retime that takes those into account and reunify the leaderboard, but since it actually appears to be a framerate issue that can swing things by potentially up to 2% (even before looking at loads), this is much more uncomfortable.
