So I finally got some time to throw a quick video together with crap quality to showcase the differences in spender gcd. Gcd stands for global cooldown and I define it as the "lockout"-time between abilities.
Note that in most of the quick demonstrations, I immediately start autoattacking/bashing straight afterwards to show you just how big the difference is between the spender gcd and autoswing/bash gcd.
With this demonstration I hope to start a discussion as to how this mechanic works. I can be certain to say that %ias increasing effects does work, as you all just saw in the video. I did equip 20% casting speed gear but that appears to have no effect, so it appears casting speed does not decrease the gcd. In all the clips I have 0% casting speed.
The affix +aps might or might not work as well, this specific thread is not about the mechanics of +aps on gear, since it would mostly turn out to be guesses anyway.
My best guess so far is that the spender gcd has a hidden ias bonus. Something like this:
I agree w your conclusion that ability APS (attacks per second) seems to be based on weapon APS. We don't know if its multiplicative or not. However, please stop calling it a GCD. It is not a GCD (global cooldown), its an attack time. Saying that a d3 attack takes 0.9 seconds or 0.5 seconds or 0.3 seconds is mechanistically the same as saying that a D2 attack was a 13-frame, 7-frame, or 4-frame attack. There's no cooldown. You don't even see a cooldown timer when you use the ability.
If you'll remember, the calculation for frames-per-Hunger, frames-per-Zeal, etc were pretty complicated (although they eventually were just tables of breakpoints).
As for the semantics issue: GCD is more descriptive than attack timer because the latter doesn't convey the fact that other skills are not useable for the duration of the attack, it just conveys that the attack takes time. Also GCD is a familiar term for many.
By the way if you're serious about figuring this out, you could take a look at the animation in a video editing program and just count the time it takes to complete with a very fast and a very slow weapon. See what numbers you get. My guess is a multiplicative bonus.
I've been a bit busy with work lately but now it's finally weekend and as such I got time to take a closer look at this once again. Counting the frames is not going to be a hassle, I think the most accurate way of measuring the frames per ability would be to do count the frames for all 5 abilities and average it. Then I would do a comparison of the frame count with the same ias effects but different weapon speeds. Going to edit this post later today.
However, please stop calling it a GCD. It is not a GCD (global cooldown), its an attack time.
To me the term comes naturally since it behaves like a global cooldown in the sense that during the time (cooldown) you're locked out, you are locked out from every ability (hence global).
The only real argument against it as I see it is, as you say, that there's nothing in the ui showing you it's on cooldown (as in WoW). In the end I don't think discussing what the term should be is very interesting; I'm not going to dictate what we should use, but until the community decides to stick with one term, gcd works fine for me.
edit: all right, so I've spent some time with different weapon speeds and so on and I first want to put a small disclaimer: frame-counting is not very exact, although it might sound as such. The reason for this can be slight lags in computer hardware and so forth. Just one frame off is quite the margin of error, since recording at 30 fps, 1 fram equals to 3,333.. hundreds of a second.
With that said, preliminary results appear to be that spenders has a hidden 10 % ias bonus. This bonus appears to be additive (just like the other ias bonuses), but I can't be quite certain. This is due to how even they are (at different ranges they are within thousands of a second apart, which with the error of margin in counting frames is impossible to give a true answer).
As a last note: it appears the weapon speed of your offhand has either zero or very low influence on the spender gcd. It appears to derive its weapon speed on the basis of the mainhand weapon aps. This was tested with 1,2 + 1,5 mh oh and 1,5 + 1,2 mh oh. The results were as such that they seemed to matched each respective mainhand speed calculation. They definitely differed from eachother when switching places.
This has a huge impact on where you should put your weapons; you should have your fastest weapon in the mainhand, since there is no offhand penalty. The only function mainhand / offhand has in D3 (as far as we know of) is to decide which weapon hits the very first attack upon entering a new game. I'm going to find out how damaging spenders are calculated in terms of mh vs oh, which might reduce the impact of (or completely nullify) these results.
The hidden bonus is perhaps like any other +aps from gloves, rings etc.
It is possible but to me the logic of it sounds weird! Time will tell later once we'll be able to stack some +aps.
I just found out a slightly related mechanic to this (it is new to me at least). When dual-wielding, spenders only use one weapon damage range to calculate the damage, not both. It uses both mainhand and offhand, in favor of the one which hit latest. I tried this by using one 25 dps 8-16 + elemental proc mainhand with a regular 2-3 damage range offhand. What I would see was that when I had hit with my very weak offhand, the spender dmg would match the calculations as if using the base weapon range of the offhand. Vice versa; if I had just done a high-hitting cleave, about 10 times more damage than the offhand (so pretty noticeable), the spender damage would end up doing about 10 times more damage than the offhand spender damage in this case.
I'm going to consolidate the information in the spreadsheet thread in my signature since not everything is on topic here.
Nice findings. So spenders should always be used right after the better of the 2 weapons was swung.
Yes in a spreadsheet scenario that is optimal, however one could debate the viability to do so in a gameplay scenario.
What it also means is that in scenarios where your offhand isn't up to snuff it is actually a dps increase to use a shield or not use an offhand at all. The loss is 15 % ias, the gain is that your mainhand doesn't get delayed by your offhand and is always used for the spender calculations. If your shield has offensive stats this especially is of importance to keep in mind. Basically the (huge) increase of mainhand damage and any offensive stats on your shield has to make up for a 15 % ias drop (and offhand damage).
I am certain this scenario will play out for several of us while gearing our Barbarians. We're not very likely to find 2x insane 1handers at the same time (not counting AH).