How SS is calculated and works

[Canadane]

I thought that it was already known it was tiered. Or at least, I have had that assumption for a while.
Good work either way!

[Krindor]

I thought that it was already known it was tiered. Or at least, I have had that assumption for a while.
Good work either way!

From the people I spoke with before this was done the common thoery was that while tooltip didn't show it, each point of SS still increased the GCD. There wasn't any data either. It is also quite illogical for it to only be useful only every 26.4 points.

[CookieMonsta]

From the people I spoke with before this was done the common thoery was that while tooltip didn't show it, each point of SS still increased the GCD. There wasn't any data either. It is also quite illogical for it to only be useful only every 26.4 points.

I actually thought it made a lot of sense to be tiered even before Dervy's work. You can save a lot of number space and processing power server side if you implement numbers and math with less precision. FP32 and FP64 hardware isn't cheap so I'm sure YoshiP made this decision to help optimize the maximum possible players per server in combat. The tier-ing gives the illusion of continuous granularity so most players won't notice.

[Krindor]

I actually thought it made a lot of sense to be tiered even before Dervy's work. You can save a lot of number space and processing power server side if you implement numbers and math with less precision. FP32 and FP64 hardware isn't cheap so I'm sure YoshiP made this decision to help optimize the maximum possible players per server in combat. The tier-ing gives the illusion of continuous granularity so most players won't notice.

This was my thought as well, from a producers perspective it's quite resonable, but from a player perspective it's going to be quite the severe effect on gear planning, not so much now compared to ARR(Due to being worth completely 0) and early HW due to low amount of secondary stats. Of course the amount of player that will probably even think about this is low, probably even lower than the clear rate of savage content.

But from what we've gathered, it isn't only due to servers, but from sloppy programming which causes performance issues. While I'm not in any way experienced in servers, I do have a self programmed simulator which is able to perform 1,500,000 of 1ms precision checks per second, so if it checks every 1ms that would be enough for 1500 players and I'm still a basic programmer so it would be a lot more efficient with some tricks. The reason I think it's caused by sloppy programming is that the formula used to calculate the GCD is inconsistant across different cast times/GCDs. If it's inconsistent in a way it means that something is not being done right or affecting the formula for some odd reason. An example of this can be seen in the 2nd google docs link comparing the formula and tooltip while under Huton.