Rezarel

1.8 Unflurried, 1.44 Flurried

 

◊ songster

You know, recently I've been noticing myself catching up to the other rogues in the group - they have DST and I don't. Is it possible that they stealth-hotfixed a PPM/haste nerf in between baldycenarius's tests and the other tests?

Has anyone else with DST noticed a drop in their DPS recently? Maybe some WWS logs from the last couple of weeks would shed light.

 

Kaber

I highly doubt it. We have not downloaded or patched anything (that I am aware of) for weeks. I honestly believe that everyone simply took an old rumor with no testing to back it as fact for a very long time.

 

hannigaholic

But that's the difference between a patch and a hotfix - a hotfix requires no client-side changes

I certianly think it's possible that Proc Per Minute mechanics have been hotfixed, after all we've seen proc mechanics being hotfixed before (the recent Windfury Totem proc change for instance).

That's not to say you're wrong, in fact I think you're probably right that we've just assumed mechanics worked a certain way without testing. Unfortunately it's not something we can test, since there are no old-version servers (nor would we expect there to be).

 

◊ songster

Well, we could rerun the fiery weapon test on live and see whether it's seignificantly different from the previous test. And look at WWS parses for rogues with the DST to see if there's beed an unexplained drop in DPS recently. Unlikely, of course, but formally possible.

 

• Aldriana

It just occurred to me that there's another thing that needs to be tested now.

There are a number weapons that proc at a rate of about 1 PPM - Rod of the Sun King, the Blacksmithing maces, The Bladefist, etc. Since the weapon is of fixed speed, it could not be determined whether it was a PPM or a flat% proc (i.e., is Rod of the Sun King a 1 PPM proc, or is it a 4.5% chance proc?) - and it never mattered. However, if these new finding are correct, we will need to go through and test these weapons to see which it is, because now it *does* matter. It will be interesting to see how these items are implemented.

 

The Grog

If this is true, where did the idea that PPM procs were improved by haste come from? I've seen it in several places before, enough that it seemed to be common knowledge, and nobody ever challenged it before.

 

Kaber

The closest explanation I have seen was posted either in this thread or the DST one: the rogue FAQ on the blizzard boards (which has long since disappeared over the last 2 and a half years) had hastes listed as increasing PPM. The only haste that could have been reliably tested that long ago probably came from Slice'n'Dice. Given that the fiery test was done with Slice'n'Dice, it may be a function specific to that ability since the counter-evidence is coming from a Warrior and me (Hunter).

I am curious to see that fiery test run on the PTR to see if Slice'n'Dice functions in a unique way. Perhaps a test can be run with other, newer procs to rule out whether it is fiery or slice'n'dice itself.

 

Mazz

Actually, I ran a minor test on live with old Crusader enchanted MSA. I can only upload the result screenshot later tonight, but I got exactly 1.4 ppm with or without SnD running. So I sorta doubt it is SnD.

It isn't that huge a testset either... almost 30 minutes on both tests, both on the blasted lands mobs.

 

• Disquette

I wonder if the "old" tests adjusted for the yellow attacks proc'ing crusader etc, or if it was just compared to auto attack. That would explain a lot.

 

◊ Anias

Pre-TBC there was some testing done on Unbridled wrath that showed that you gained more rage per minute from a fast offhand, as well as from flurry, indicating that at that time UW behaved as a percentage per swing proc, instead of adjusting to the weapon speed. Since then it seems to have changed.

I don't have the link (another loss of archiving my guild forums) but that's where some of the "faster = more procs" stuff likely springs from.

I'm not sure if UW is caught by wws, but if it is that's a source of some data for analysis.

What's the simplest explanation for procs now?

X Procs are PPM, calced per swing based on weapon speed at swing time (hastes included)
Y Procs are flat PPS, calced per swing based on a flat chance to proc

Does that cover everything, or is there a third proc that is PPM but not adjusted by haste?

 

◊ songster

We need better evidence on which hastes are adjusted for. From baldycenarius' test, there's a chance that S'n'D is not adjusted for. Can this be confirmed? If so, which other hastes are not adjusted for? It may be that S'n'D is a special case, since rogues are balanced for ~= 100% S'n'D uptime.

 

Rezarel

I tested the Lifesteal enchant on the PTR. It looks like Baldycenarius's results: the proc chance doesn't drop while flurried.

If the Lifesteal proc worked the same way as the other procs I've tested, I'd expect the flurried proc chance to be ~23.2% / 1.25 = 18.6%, well below the observed rate. Instead, I observed an increase in real procs per minute of about 18% (you'd expect 25%, but that's probably due to errors in tracking haste gains/losses in the combat log).

It looks like we're dealing with a flag that's set for each proc. Some of the oldest procs keep the same proc chance when hasted, but everything relevant these days lowers its proc chance to maintain a constant real PPM.

Again, I should note that this is all ignoring instant attacks. I tested with autoattack only.

 

◊ Anias

Thank you rez.

Time to make a chart I suppose.

 

Mazz

Hmmm

Crusader was adjusted to lower in power when your level increases right? Did the same thing happen to lifestealing and fiery? If not then this might have been incorparated in these pre TBC changes

the test I mentioned earlier. Basically, I used a flower in my offhand and used only autoattacks during the non SnD test. I had to SS 38 times to keep SnD up during the second test, however I timed it so I could distinguish procs from autoattack from those from SS. Out of those 38 SS's only 3 procced crusader, which is incidently in line with the autoattack swings.

[img=http://img212.imageshack.us/img212/8509/ppmresultsfb9.th.jpg]

 

Rezarel

Lifestealing only heals/damages 18 points at level 70, instead of 30 at level 60. I don't know if Fiery was changed.

 

◊ Kalman

Although I wasn't the original source of the notion, I tested it back in MC/BWL with Lifestealing after they converted it to shadow damage. Results were consistent with PPM procs being improved by haste. Based on that, there was some other testing done with Fiery. Results consistent with PPM procs being improved by haste.

From there, the assumption was "Well, it's true for 2, probably true for all.

 

◊ songster

Aren't fiery and lifestealing both considerably higher than 1ppm? Maybe that's the diffference.

 

Rezarel

Yah, but Unbridled Wrath is up around 15 PPM... but Fiery and Lifestealing do have about the same proc rate, I think. Who knows, maybe that does have something to do with it.

 

Sarlunas

Done some testing on Live with the help of a trusty Servant of Razelikh. 2.8 speed maces with fiery enchant and both romulo and dragonspine equipped. Testing with and without SnD since I don't have access to enough haste gear to do any useful tests on that. I also used the buff cancelling macro to remove the haste buff. In the first test I used only SS, standing sideways to avoid auto-attacks.

Yellow test without haste:
SS 972
Fiery 286 - 29.4%
Romulo 38 - 3.91%
Haste 50 - 5.14%

Yellow test with SnD:
SS 1143
Fiery 322 - 28.2%
Romulo 47 - 4.11%
Haste 58 - 5.07%

Expected proc rate for 1 ppm at 2.8 speed is 4.67%. Smallish sample size but still no indication of lower proc rates with SnD. Resisted fiery/romulo procs are included btw. Next I did a couple tests on white hits, specced setup to get combo points for SnD.

White test without haste:
Hits 2810
Fiery 795 - 28.3%
Romulo 120 - 4.27%
Haste 155 - 5.52%

White test with SnD:
Hits 3838
Fiery 1076 - 28.0%
Romulo 148 - 3.86%
Haste 145 - 3.78%

Got a bit of a streak of haste procs at the start of that first test, was up well above 6% after 1k hits. It kept going down but it's still quite a bit above the expected value. If SnD doesn't increase ppm then the expected proc rate at 30% haste would be 3.59% unless I'm mistaken, so it's not too far off. My guess is that old stuffy procs like fiery simply doesn't work quite like our newfangled modern procs.

 

Derex

First, just found you guys and this thread, GREAT board.

Ok, so I'm going to try to recap what I've learned in this thread, please correct me where I've gone astray.

* PPM is calculated as a percentage chance to proc by the game code. Current feeling is that for AT LEAST BC content, this PPM is modified by your current hasted weapon speed. Thus the PPM Rate for BC content goes down as weapon speed goes down. (white hits)

* We feel this is the same on Live as it is on PTR, there is no "stealth nurf" to proc rate coming from the 2.2 changes incoming in a few weeks. Other than items like the DST and the known haste rating change haste and procs will work the same as it does now when 2.2 is released.

* We feel we have run into a case where it seems at least some old world procs like Fiery and Lifesteal support a different understanding: PPM Rate for these procs are not re-calculated as haste is applied. So assuming fiery was 2ppm a 2.0 speed weapon will always proc fiery at say a 6.66% chance regardless of whether that weapon, via haste, has decreased to 1.5 speed. Thus, for some old world enchants, we feel that haste ups proc rate. This would correlate with some old sources (the rogue FAQ for instance) where testing was done a long time ago, likely well before the "changes" were made.

* We speculate that yellow damage (sinister strike for instance) is NOT affected by the decrease in PPM that would be calculated if haste were applied. i.e. With fiery above the applications of fiery would remain 6.66% when applied with sinister strike regardless of whether you were at 0 haste or 100% haste.

*Postulate: Could the PPM rate for a proc with something like Sinister Strike be based upon the "default weapon speed" that is used for calculating Attack Power additions to weapon damage (i.e. Attack Power generated DPS is applied to a Dagger at 1.7 speed regardless of the actual weapon speed of the dagger, 2.7 for swords, etc.)

Conclusion:

It looks like the changes actually went in with modifications to how BC procs worked at the time BC was pushed out. Our prior calculations and parses may be correct since most of the data we have on this subject is pre-BC. What we do seem to have found out is that:

1) BC Changed the way Procs work to be more what the standard definition of "Proc Per Minute" would mean and...

2) Blizzard didn't go back and change the old stuff because it's mostly deprecated anyway due to the advances in enchants and procs on the BC weapons...not many people use fiery, lifesteal, etc. at 70 anymore.

In essence we have no data that supports anything BC related working any differently than the first couple assumptions I mentioned here.

Again, this is just my reading of the thread "so far", anyone care to correct any of the information I have above or is this pretty much a valid summary of the data we've seen up to this point?

* By "our" and "we" I mean the hard working people on the board, I exclude myself from this because 1) I'm new, 2) I haven't done any testing, at least for this thread, and 3) I deserve no credit nor am I trying to take any for any of the hard work in this thread.

 

• Vulajin

No, you haven't done any testing, Derex, but you've put together what looks like a perfectly accurate synopsis of the thread so far. Perhaps Blizzard really did foresee the feedback loop items like Dragonspine Trophy and enchants like Mongoose (basically, haste procs without internal cooldowns) could create with themselves, and thought to change the mechanics for those procs to eliminate that feedback loop. If that's the case, I sure feel stupid for all the crying I've done over the past several months about Blizzard's folly.

In response to Sarlunas, I commend your efforts. Do you think you could be troubled to pick up a couple faster weapons and repeat your test of yellow attacks? I'm wondering whether the proc rates of PPM procs from special attacks are calculated based on the speed of your main hand weapon, or some normalized speed (for example, the speed to which attack power contribution is normalized; 1.7 for daggers, 2.4 for other one-handers). Or if anyone else could attempt such a test, it would be greatly appreciated. I'd do so myself if I had a Dragonspine Trophy or Romulo's Poison Vial, but alas, drops have not been kind to me.

 

koaschten

I really need to get that feral druid guildmate to give me his account data, i could run endless test-rows with iLotP heal on Blasted Lands Servants... :/

 

◊ songster

Lovely summary, Derex.

As I see it, there are two possible distinctions between the procs which are normalised to current weapon speed, and those which are normalised to base (or default) weapon speed. Recall that most proc effects seem to be 1ppm, while only a few such as Fiery and Lifestealing have higher proc rates. The options are:

1) Old world procs normalise to base/default weapon speed, newer procs normalise to current attack speed. This is the one you suggested.

2) High proc rates normalise to base/default weapon speed, while the standard 1ppm procs normalise to current attack speed.

One way of testing this would be with Crusader enchants. It's an old-world enchant, but it looks to follow the standard 1ppm proc rate. The converse test would be a TBC item or enchant with a high proc rate (i.e. greater than 1ppm). Is there any example of such an item/enchant?

 

Bloodtear

A bit offtopic, but all this talk about Lifestealing enchant made me wonder about something.
Does Lifestealing still scale with +spell damage?
If it does and assuming enhancement shamans with at least 2K+ AP, is the new shaman talent converting AP into spell damage enough to make it a viable enchant?