U33 Updated Throwing Speed Samples

[jakeelala]

So I spent a few hours testing throwing speed a couple weeks and just remembered to throw it up here.

TLDR;
-Echoes of the Ancestors: Shiradi twist for 10% thrown attack speed actually seems to SLOW DOWN throwing animations based on these tests. This is bizarre because this twist even raises your Dexterity by 1, which should slightly improve thrown speed through extra procs in this test. If anything, Echoes slowdown is slightly more than shown in this test do to it's slightly increased Dexterity bonus.
-At end game with the proper feats, Whirling Wrists and Divine Power (for BaB30) provides a net ~10% increase in the speed of the throwing animation versus having neither.
-Having just Whirling Wrists the difference with BaB30 active is about 6% versus BaB25.
-Twist Whirling Wrists, keep Tensers/Divine Power up, and forget about Echoes of the Ancestors

Details:
I used my capped level 30 thrower character so I would have the option of throwing at both 25 BaB and 30 BaB (Divine Power clicky) for the tests. I took off all Dexterity and Doubleshot gear. I had 13% Doubleshot (10% epic feat and 3% from Kensei cores I think), and 41 Dexterity. Trained feats known to affect throwing speed were: Rapid Shot, Quickdraw, and Blinding Speed. I also untrained Advanced Ninja Training to reduce extra shuriken procs.

The main point of the test was to use stacks of 100 non-returning shuriken to ascertain how quickly rate of fire speed would change with and without Echoes of the Ancestors: Shiradi, Whirling Wrists, and BaB. I wanted to minimize extra projectiles because they introduce a random element into the time to throw 100 stars that is independent/confounding with throwing animation time, which I was actually trying to figure out. Ideally this test would be done with zero extra projectile chance, but frankly, that in and of itself is an unrealistic test. Playing a thrower you will always have extra shot chances. In any case, I think my sample size is big enough to make the tests informative.

I threw a total of 800 non-returning shuriken in increments of 100 (a single stack)using auto attack. So I have 8 trials for each different layout, and 4 different layouts, all tested at either 25 BaB or 30 BaB (that's 6400 hundred non- returning shuriken's thrown). Destiny was Fury of the Wild. I stop watched the stack of 100, and just watched my character throw 100 stars, 8 times, for each layout. It was mad boring. Layouts were the following:

Layout 1: Whirling Wrists, Echoes, and BaB25
Layout 2: Whirling Wrists, BaB25
Layout 3: Echoes, BaB25
Layout 4: Neither Whirling Wrists or Echoes, BaB25
Layout 5:Neither Whirling Wrists or Echoes, BaB30
Layout 6: Echoes, BaB30
Layout 7: Whirling Wrists, BaB30
Layout 8: Whirling Wrists, Echoes, and BaB30

Obviously I expected Layout 8 and Layout 7 to show a large speed advantage. However, based on my results posted below, the differences over 40 seconds of straight throwing show a pretty subtle difference.



The absolute slowest setup was Layout 6, which has Echoes and BaB30, followed by Layout 3 which also just had Echoes but BaB25. They were 338.5 seconds and 336.48 seconds, respectively. They were so close in fact (pretty much within the ~2 seconds standard deviation) that they were effectively identical times. That tells us that BaB bonus, when you have all the throwing speed feats trained, just isn't a factor after BaB25.

What's interesting is the next two are Layouts 4 and 5, which both had neither Whirling Wrists nor Echoes, and just differed on BaB25 or BaB30. These were also within 1 standard deviation of one another at 334.6 and 335.39 respectively. That tells me that at best Echoes does nothing appreciable, and at worst it might actually slow the animation down somehow. But given the sensitivity of the analysis (not ideal) I feel secure saying that it simply does nothing in the presence of Blinding Speed, BaB25+, Quickdraw and Rapid shot.

The next two fastest Layouts were 8 and 2. 8 had both feats and BaB30 (the expected winner), and 2 had just Whirling Wrists and BaB25. Not having Echoes in this case made this layout faster than having Echoes and BaB30. This is another strange case where it looks like Echoes actually slows down the animation speed. In this case, the Layout with Echoes and a higher BaB (Layout 8) was two standard deviations slower than the Layout 2 with just BaB25 and Whirling Wrists, to the tune of a full 2% slower throwing animation speed. Very strange.

Finally, the fastest layouts were Layout 1 and Layout 7. Layout 1 had both feats and BaB25, and Layout 7 had only Whirling Wrists and Bab30. If Echoes really does nothing or even slows the animation, than Layout 7 becomes the predicted winner (highest BaB, Whirling Wrists and nothing slowing it down), which it does handily with a total time of 305.33 versus Layout 1 at a time of 317.66.

I seriously doubt anyone ever would, but if someone wanted to repeat these tests it would be awesome to see if the results were similar.

[Sam-u-r-eye]

you should totally upload that pic on IMGUR so I don't have to think that "beautiful Ukrainian women are looking for an american man"

[mikarddo]

Thanks for sharing.

The results are indeed very puzzling. That 4 and 5 are so close while 2 and 7 are not makes absolutely no sense - ignoring the issue of Echoes. If anything I would have expected the reserve due to some sort of cap. It makes no reasonable sense to me.

[jakeelala]

Thanks for sharing.

The results are indeed very puzzling. That 4 and 5 are so close while 2 and 7 are not makes absolutely no sense - ignoring the issue of Echoes. If anything I would have expected the reserve due to some sort of cap. It makes no reasonable sense to me.

Truly. I suspect it has something to do with BaB affecting thrown attack speed with Whirling Wrists more than it does without Whirling Wrists.

Perhaps the most important thing this shows is that at a certain thrown attack speed threshold, there's a real diminishing return where investing in higher base damage, additional proc damage, or more missile procs would yield better marginal returns.

[S1375]

Thank you for sharing your research!