So, I gave it some thought, and I think the best way to test for the “fall-off” effect is this:
You use a 2H mace with 1.00 base APS. Ambo’s in the cube, no Lamentation.
No AS anywhere, including paragon. This gives you a 30 frame DD and a 40 frame WW tick.
0.0-Rend applied… Though, now that I think of it, what point is this, exactly? I guess the DD is spawning before the actual WW hit, and it’s this that applies Rend. Which means that the WW tick at “0.4” is actually occurring at “0.1”…
So:
0.0 - Rend applied (by DD)
0.1 (WW), 0.3 (DD), 0.5 (WW), 1.0 (DD)* - hits that could apply Rend
1.0 - Rend drops. The trial here splits here into “Case A” where Rend is reapplied by the DD hit with the *, above, at 1.0, and “Case B” where it can’t reapply at 1.0, and instead is reapplied at 1.3, by WW or DD (both should take place at 1.3)
“A”: continues to refresh Rend exactly every 1.0 second
“B”:
1.3 - Rend reapplied (by WW or DD) 30 lost frames
2.0 (DD), 2.1 (WW), 2.3 (DD) - hits that could apply Rend
2.3- Rend drops. Note that the DD at 2.3 can’t reset Rend in this scenario, since if it could, we’d be in “Case A”, where Rend can be reset at an even count.
2.5 - Rend reapplied (by WW) 20 lost frames
3.0 (DD), 3.3 (WW), 3.3 (DD) - hits that could apply Rend
3.5 - Rend drops
4.0 - Rend reapplied (by DD) 10 lost frames
4.1 (WW), 4.3 (DD), 4.5 (WW), 5.0 (DD) -hits that could apply Rend
5.0- Rend drops
5.3- Rend reapplied (by WW or DD) 30 lost frames
I believe that the cycle of “B” is essentially repeating, with 60 lost frames every 4 seconds. This should cost you a full 25% of your damage.
“A”, in comparison, should lose no frames at all, and suffer no damage loss. Because Rend deals the same damage on every cast, you can calculate out how much damage it should be doing over a certain period of time, if fall-off is not a factor.
For instance, you might calculate, based on your weapon damage, crit chance, crit damage, and strength, that Rend should deal 1 Billion damage per minute. Then you Whirlwind an enemy (I always use Manglemaw for testing) for a minute, and see how much damage you’ve done. Assuming you’ve got no buffs to WW going, the vast majority of your damage should be coming from Rend (since it is getting a 15x multiplier from Ambo’s)
So, in this example, if “A” is true, you should do slightly more than 1 Billion damage in one minute (the Rend damage, plus some minor WW damage).
If, on the other hand, “B” is true, you should only deal a little over 750 Million damage in that minute.
If you get “B”, you know fall-off is real. If you get “A”, there’s still more testing to do, since Rend getting refreshed exactly every second, and being continually refreshed by WW (aka, “fall-off” doesn’t exist) should produce the same results.
So, for “Case C”, you repeat as in the first set of trials, except now you put 1 point in Attack Speed in your paragon. This will give you 1.0001 aps, dropping your WW from 40 frames to 39, and your DD from 30 frames to 29.
So:
0.0- Rend applied, by DD… the WW tick at “0.39” is actually occurring at “0.09”.
0.09 (WW), 0.29 (DD), 0.48 (WW), 0.58 (DD) - hits that could apply Rend
1.0 - Rend drops
1.27- Rend reapplied (by either WW or DD, both of which happen at 1.27) 27 lost frames
1.56 (DD), 2.06 (WW), 2.25 (DD) - hits that could apply Rend
2.27 - Rend drops
2.45 - Rend reapplied (by WW) 18 lost frames
etc, etc… the point is not the exact amount of damage lost in this case, but rather the fact that by increasing our attack speed, even by an incredibly small degree, we have de-synced WW and Rend, and, if “fall-off” is real, we will lose a non-trivial amount of damage (considering that in less than 3 seconds we have already lost 45 frames, I would look for at least 15% damage loss compared to the calculated maximum).
So, do another test on your chosen enemy, and note the damage after a particular length of time. If the damage you deal in this case, “C”, is less than was dealt in “A”, then you know that fall-off is real. If the damage is nearly identical to “A”, then WW is probably refreshing Rend stacks, resetting the duration of each instance of Rend to the maximum (1 second) each time it hits.
Simple, right?
Edit: I ran a much quicker and less painful version of this trial. Since in “Case B” we should be losing as much as 30 frames (half a second) at a time, and in “Case C”, as much as 27, this should be evident from a simple visual inspection. I WWed a skeleton with a 1.00 aps mace with no bonus to aps, and then with 1 point of paragon in AS, and in neither case did I see the Rend icon drop at any point.
So, that’s not 100% conclusive, but I’m guessing that the duration of Rend instances is actually being refreshed with each WW/DD hit.
Now, what that means, will involve further testing. For instance, hard-cast Rend can proc AD… so if you hard-cast Rend in density, then start WWing, is the AD-dealing ability of those Rend instances preserved, and refreshed repeatedly? (Edit: no, it’s not. When you start WWing, your HC Rend is overwritten).