I know this thread has been around awhile, but I just got around to actually comparing the results to some of my own data,
and let me just say, "wow!"

Thanks for sharing this, and wanted to tag it for future reference. Just for the record, did my own testing to confirm your
results that are relevant to me, and all I can say is thank you! This info is highly useful.

 

-----signature-----

bump

 

-----signature-----

bump #2

 

-----signature-----

I realize the figures for the break points are meant for level 50s with +10% cast speed bonus, but

1) If a L50 does NOT have +10% cast speed, do the break points still remain at the same point? and

2) do the breakpoints also remain the same with L39 toons, who cannot get the +10% cast speed and have lower Dexterity?

 

-----signature-----

Both dex and cast speed seem to be a factor, as you can sometimes drop cast speed % without actually casting any slower.

So I assume the formula looks something like this:
Actual cast speed = (Delve (ToA bonus factor * Dex factor))

... and these somehow gets rounded to produce the tiers we see.


I have not tested this for molvik toons, but assuming the above is correct there would be breakpoints for molvik toons too, although not the same as you'd have with 10% cast speed.


I'm planning on opening my accounts again once 1.108 goes live. I might do some more testing if i get bored enough

 

-----signature-----

I really dont know anything about the formula for determining cast speed, but in my mind it looks like this:
Actual cast speed = (Delve speed * Dexterity) * ToA cast speed bonus.

It seems to me that cast speed bonus is only used to increase the number of times any particular spell is cast within any given time frame. They only increase a percentage of the base cast speed, which is affected first by Dexterity, then the cast speed bonus.

Therefore, the break points should remain the same no matter the % to cast speed, right?

If the actual cast speed is affected by the % bonus BEFORE bringing Dexterity into the formula, then the bonus is more important than the Dexterity, for templating reasons.

At what point in the cast speed calculation are Dexterity and % bonus to cast speed determined?

 

-----signature-----

You get tiers when chaning nothing but cast speed too though.

From the ywain thread:




Vanesyra got 3 tiers here, but if you compare the results with my tests, both the 1.41 and 1.26 sec casts almost exactly matches my results for 2.8s spells.
So with 371 dex and 6% ToA bonus you get the same "stepup" as you do with only 356 dex and 10% ToA bonus. Testing is needed to determine more exactly where this stepup happens, but i guess it would be safe to assume it's in the 360-370 dex range.

Although another poster claims to have tested the 2.4s breakpoints on his thane and while he found the same breakpoint at 350 dex he found it with only 9% ToA cast speed, which is very strange (the dex required should go up with less cast speed). My best guess is that at this exact point 1% cast speed < 1 dex. This might be a rare artifact or very common, i don't know.

Again, I guess it all boils down to "more testing needed".

 

-----signature-----

I can't believe how many people were entertaining the idea that your data gathering methods needed to be more professional and detailed to prove what you have surely proved...

This was barely a case of statistics or sample space, with the exception that you needed a couple sets of 100 casts in each dex setup so that we could see your data was consistent enough. I'm pretty sure a 12 year old could do your test and clearly see these huge gaps in cast time that don't really require any explanation to know that you should change your RA/Template setup.

Coming from Mordred ... I made my Wizard around 2004 and within a year I noticed the possibility for a tier system in my casting speed because on Mordred we didn't use buff bots in 8v8 and when a Shamman did my dex/qui the cast speed on my 2.8 nuke was slower because the shamman dex/qui is actually a few points lower than the cleric/druid (or something like that)..

I'm not positive but I think its possible my Wizzard was fluctuating around 389 and 392 dex, but even with this specific example aside.. I can remember various times on lots of casters and discussions in various Ventrilo channels with other casters about the possibility of a tier system, because lots of people were noticing big differences when reaching specific values of dex (even when the difference is just 1 point of dex)

 

-----signature-----

Below are test results with 2.6 speed baseline Sup DD on my SM.

All tests were done the Slazjer method - 100 casts using a manual timer. All results for each tier are within .5 seconds of each so user error shouldn't be an issue.

__9%__

383 Dex - 110.60
382 Dex - 111.03
381 Dex - 110.69
380 Dex - 110.62
~~~~~Breakpoint~~~~~
379 Dex - 113.54
377 Dex - 113.56
375 Dex - 113.63
374 Dex - 113.22
~~~~~Breakpoint~~~~~
373 Dex - 126.23

Interesting that there is still a breakpoint at 374, however, 380 Dex/9% spd is needed to meet the 374/10% spd breakpoint.

__8%__

381 Dex - 110.50
380 Dex - 110.81
~~~~~Breakpoint~~~~~
379 Dex - 123.54
378 Dex - 123.75
377 Dex - 123.54

Breakpoint is the same as with 9% spd at 380 Dex.

__7%__

387 Dex - 110.68
386 Dex - 110.86
~~~~~Breakpoint~~~~~
385 Dex - 113.45
384 Dex - 113.95
383 Dex - 113.93
382 Dex - 113.54

379 Dex - 126.29

I did not find the 113/126 breakpoint here.

Summary: It appears there is a mini tier where the ToA cast spd is brought into the calculation. This would explain some of the mini tiers apparent in Slazjer's original data that wasn't presented as a breakpoint.

I would recommend hybrids that do not have 10% cast speed and enough dex to make it into a higher tier in Slazjers data do testing on whatever stats they may have in a temp. Fortunately, casters have no issue getting 10% ToA spd and especially with the new 7 tier RA system few dex points will be wasted.

 

-----signature-----

Bump cause I need someone to comment on my data to make me feel that time was not wasted!

 

-----signature-----