Yes I believe someone did mention that it's different for all TOA cast speed %s, but I'm sure some will share the same value out of coincidence.

 

-----signature-----

Yes, the breakpoints changes with cast speed.

 

-----signature-----

1) its quite possible that its the result of a truncation that is occuring in the code at several points in the calculation. when we model the game data we like to think that one equation solves everything, but programers have a way of doing things bass akwards and using 5 steps rounding 3 times and missing a negative sign and few parenteses to do the same thing.


2) Its also possible that it is a result of truncation on the part of the tester. 100 casts is a very small sample size, and those with only 25 casts are laughable. the game delves will only read to +/- 1 second, so the error can be huge.


3) The last possibility id throw out there is a testers phenomenon often referred to as "double casting". this is where two consecutive spells cast within 1 second. this is can be both visual (watchign your character casting animations break and restart, casting btoh very quickly, and noticible in the logs.


in the logs again note that each spell cast actually is recorded in whole secondsthere is no such thing as a 1.2 second cast, only a log with 80% occuring at 1 second intervals and 20% occuring at 2 second intervals, producing an average of 1.2. this is one reason why large sample sizes are so important. they help reduce this rounding error. but back to double casting... double casting occurs in logs when one spell seems to cast in ZERO seconds in the log. now ask yourself, even if te game is rounding the time stamp, how does a 2.8 second spell casting at around 1.2 seconds come out to zero seconds? lag? well yes, but the strange thing is that certain dex breaks increase the frequency of double casts. this could be because of lag in calcualtions or soe other error, but the end result is noticible breaks where "double casts" occur more often. and in actual gameplay this is mroe useful. in the few seconds you get to free cast on someone, being able to cast 2% faster wont help, but if you get more frequent double casts it can increase the casts by 10-20%.


-----------------------------

In my personal testing id done in the past i notices some break points (nto as dramatic as the OP), and my tests were on the order of 5,000-20,000 casts.


Thre truely difficult step would be to take the data gathered and actually interpret it with mroe than a chart. reverse engineer the formula or an equivolent fit that will produce the same data INCLUDING the break points. and im not talking a forced cap or using 6 different equations, but rather a system with truncations that would cause the same errors that the game seems to produce.


PS, use lvl 1ish spells on list casters to chain cast for hours on end without running oop, you get better results.

 

-----signature-----

Firstly, I didn't use logs to calculate the cast speeds, it was all done manually.

Secondly, a lot of people have replicated and confirmed the results. My posted times might not be 100% accurate at all times because of lag etc, but they give good pointers.
Look back at the first page and you'll see that i did a test with 3x 100 casts per dex value for 2.8 sec spells and got very little variation (a couple of tenths of a second over 100 casts).

The tests are very easy to replicate, anyone can do it. So far every person who have actually tested it seem to find the same breakpoints.

Koee ran an almost as big test as i did using other spells and classes (posted in the ywain thread) and found some disagreeing cast times for some spells. Those results could also be replicated though, and we quickly came to the conclusion that all spells with the same delves seem to share the breakpoints but not necessarily the actual cast speed (see mentalist vs eld/cab nearsight discussion). More testing is needed there.


If someone wants to run larger tests, go for it! I'd be happy to be proved wrong if it meant a real formula or w/e was discovered. Until then though, I'll just go with this. I'd be crazy if i didn't belive my own testing, especially when both me and others can, and have, replicated it so easily.

 

-----signature-----

Yeah it seems to be a sport on VN to try to dismiss or question other peoples work with nothing but rather inane statements like "bigger sample size!" and such without actually attacking the numbers with facts of your own.

While the basis of that argument is certainly true, the fact is you're being completely unscientific by challenging data with no data or mathematical proof of your own. If you did either some actual data collection of your own, or at least would analyze the provided data statistically, you'd see that 100 casts is plenty to get a quite accurate idea of a spells cast-time; certainly down to .1s accuracy with minimal error. (.01s accuracy would require an equally greater order of magnitude in the sample size, but few people will care about that kinda accuracy anyway.)

And obviously the "breaking points" are so large that you'll actually notice them yourself when you reproduce his tests in a controlled environment. I could actually tell I hit a break point just by the amount of extra healing my target suddenly required to stay alive, the speed with which the sounds & graphics were coming, and the spam text even. This is obviously just a subjective feeling, but again if you do it, I bet you you will see it too. It's not hard to notice a ~9% increase in casting speed that happens literally with 1 extra DEX being added to hit a break point.

I think that's the fundamental message Slajzer was trying to convey, but naturally every self-respecting DAoC power gamer will go and test an exact breaking point for his class/template/setup by themselves rather then blindly rely on someone else's hard work.


The fact is for my NS I can get AugDEX5 & a +6 DEX Myth and hit 445 DEX and my casting speed doesn't change one bit from 374 DEX.

I dunno, but that's a pretty important thing to know if you ask me.


I mighta guessed/known that above 400 DEX is not really worthwhile, but even to get 400 DEX I need at least AugDEX3. I would be spending 10 RPs and not gain any casting speed at all, but only like 11 damage per DD. (Far worse then even MoM3, in fact barely more then half as much damage increase as MoM3. Very poorly spent RPs!)


Due to Slazjer's tests, gone are the days where 9% Cast Speed in a temp, or lacking a couple of points of DEX off your cap was acceptable or "no big deal." Not if you're missing a break-point and possibly giving up ~9% casting speed rather then ~1% as you thought you were losing!

And conversely, if you have hit your last break point very easily, and are unable to reach the next one anyway, you can possibly slightly improve your template by dropping DEX down a tad here and there without losing any casting speed what so ever. (You know how often that a 28DEX gem in an SC slot would be so nice to drop down to 25 and use that utility elsewhere to raise 2% more resists or something else.)

 

-----signature-----

looking at all the data posted, the break points follow an interesting trend that suggests a simple formula.


the break points for casting times fall at roughly:

0.95, 1.00, 1.10, 1.25, 1.40, 1.55, 1.70, and 1.85. now if you were to say that given the sample size the difference between 0.95 and 1.00 is difficult to distinguish, then you end up with break points at 0.95+0.15n for integer values of n.


which is somewhat interesting. if it was a flat truncation at the end of the calculation we would expect to see break points at every 0.10 seconds, not every 0.15 as shown. why would a programmer round like that? so the truncation might happen earlier and then have the final answer multiplied by a scaling factor that produces the .15 difference, meanign that the programmer seemingly wanted to create larger break point. interesting really.


the trick is to fit a casting speed formula that when truncated down at appropriate places fits those break points for all spells. or just force the fit and round the final numbers (also quite likely). and since its friday and a slow day at work, i just might do that... modeling and paper daoc are always fun.

 

-----signature-----

It makes me a sad panda that my Lurikeen Mentalist (w/ +10 dex at creation) can only reach 391 dex with AugDex3:

80 (base) + 10 (creation) + 23 (lvls) + 101 (items) + 155 (buffs) + 22 (AugDex3) = 391


If the numbers posted are true, I'd have to spend another 10 points on AugDex4 to get to the next "tier" of speed for the 2.8s spec light nuke and the 2.0s nearsight

It would also mean that the points spent on AugDex3 are nearly worthless!

 

-----signature-----

Abmfett obviously you have some knowledge of coding and/or programming, am interested if you came up with anything such as a model? Since none of the original devs or coders are with the game anymore, we rely on independent testing, mostly by the players themselves, and thankfully we have some such as the OP here that wish to share their info...we need all the help we can get.

 

-----signature-----

You can respec creation points any time at no cost at all

Just click Customize and then Adjust Atributes in the char selection screen.

 

-----signature-----

Dunno how much you're short as I'm too lazy to look it up, but I figured people should know that there are Infernal Sleeves available that have mythical stat + caps bonuses, like +3 or +5 I forget. They're not the 102AF ones but the blue con ones, but for a caster does it matter really?

 

-----signature-----