PDA

View Full Version : Flying Kick & Crits



Nedrom
11-11-2013, 03:13 PM
Anyone who can run some lengthy parses want to see if we're experiencing the same crit issue with flying kick that rogues have with backstab?

Details of the issue here:

https://forums.station.sony.com/eq/index.php?threads/rogue-backstab-crits.204966/#post-3012328

Nedrom
11-13-2013, 02:57 PM
Wow, no one interested in testing this?

rapitiss
11-13-2013, 10:41 PM
So for us to test this you'd run a long parse with and without something that boosts FK and then look at the min/max damage as well as the crit rate.

Are we seeing a problem where our FK max hits are capped like backstabs ? ?

Gorkeyah
11-13-2013, 10:45 PM
I'm not sure I understand the issue. In general I haven't noticed anything weird with my dps. Is this something recently broken that nerfed us potentially, or just something that doesn't seem right?

Nedrom
11-13-2013, 10:59 PM
Would like to know if were affected. Sadly my parsing days are over. But for the benefit of our class would be good to know for sure.

Sometimes these types of bugs are not solely on one skill.

Steps to reproduce:

Parse flying kick by itself unmodded by spell disc or as buffs
Hit iron fist and parse again

Compare crit rates they should match. If not then we have the same bug

Bixxby82
11-14-2013, 11:48 AM
@Gorkyeh :: it sounds like an issue with crit computation, from what I could deduce.
Basically, if the max hit of a FK that’s going to crit is above a ‘numeric threshold’ of sorts (as of yet unknown/broken), it either won’t crit, or it’ll crit but not for what it’s supposed to (i.e. half the value, etc)

Bixxby82
11-14-2013, 02:41 PM
to help clarify, here's a couple of the parses (and an explanation of what you're looking at) that Brogett posted on the EQ Forums:

Here are a few short parses. They're not as long as I'd like, but I'm busy and can't spend all day on this (and my auto-backstab method for parse gathering seems to randomly flake out on me now, bah). Anyway it's enough to see the problem IMO.

Note that Backstab is a tricky one as we have 3 types of backstab all labelled the same thing in the logs. I have a tool to split them, but it labels Incursion as Battery as it was written several years back. Be assured it is infact the incursion skill I'm using. So without further ado.

Without Rogue's Fury multiplier:

Code:
$ tail -f eqlog_Brogett_antonius.txt > eqlog_Brogett_without_RF.txt
^C

$ /c/jkb/eq/bs_split2.pl eqlog_Brogett_without_RF.txt
All BS = 12944507

=== backstab 7743485 59.8206250728591 ===
Normal: 703 hits, average 6095.92460881935 min 2127, max 15262
Critical: 254 hits, average 13614.3700787402 min 3832, max 18723
Hit rate of 91.4040113740172%
Crit rate of 26.5412748171369%
346 rounds
1 per round = 0 0%
2 per round = 0 0%
3 per round = 346 99.99%
4 per round = 0 0%

=== massive strike 4375164 33.799386875066 ===
Normal: 201 hits, average 11359.4527363184 min 3359, max 15864
Critical: 76 hits, average 27525.1842105263 min 6851, max 37350

=== battery 825858 6.37998805207491 ===
Normal: 30 hits, average 21331.3 min 6467, max 35038
Critical: 7 hits, average 26559.8571428571 min 20132, max 38147

Total hits = 1271
Total miss = 90
=> total hit rate of 93.3872152828802%
With Rogue's Fury running:

Code:
$ tail -f eqlog_Brogett_antonius.txt > eqlog_Brogett_with_RF.txt
^C

$ /c/jkb/eq/bs_split2.pl eqlog_Brogett_with_RF.txt
All BS = 16259495

=== backstab 10314707 63.438052657847 ===
Normal: 474 hits, average 11990.9831223629 min 3353, max 16788
Critical: 161 hits, average 28763.8571428571 min 6836, max 46516
Hit rate of 92.4308586718619%
Crit rate of 25.3543307086614%
227 rounds
1 per round = 1 0.44%
2 per round = 0 0%
3 per round = 226 99.55%
4 per round = 0 0%

=== massive strike 4732350 29.1051474845928 ===
Normal: 156 hits, average 26038.6217948718 min 5879, max 33385
Critical: 24 hits, average 27930.2083333333 min 14327, max 54008

=== battery 1212438 7.45679985756015 ===
Normal: 20 hits, average 53773.6 min 13475, max 75245
Critical: 3 hits, average 45655.3333333333 min 42802, max 47082

Total hits = 838
Total miss = 52
=> total hit rate of 94.1573033707865%

Notice the final "battery" bits. With RF my highest Incursion critical hit is 47k whereas my highest non-crit is 75k. The multiplier for massive strike crit vs non-crit is suspect too, compared to the non-RF one.

Now I know some will look at this and say "but it's only 3 incursion crits". Too small a sample size, yada yada. However the critical chance is also down and also the same issue can be seen with the lower but more frequent massive strikes. Consider massive strikes crits: without RF it was 27%, with RF it's 13%. Then there's the multiplier of crit to non-crit BS: without RF it's 2.4x, with RF it's 1.6x. That large a shift on both crit rates and multipliers isn't just down to small sample sizes, and I've seen it on much larger parses. It's simply that the large critical hit values have been gone.

It's not noticable on normal backstab, somewhat noticable on massive strikes, and very noticable on incursion line. Basically the stronger the base backstab attack the more prevalent the issue is with high crits vanishing. This correlates well with the claim that high damage values are being forced into non-crit due to an artificial cap.

---
---

nother parse, this time on a 750Mb log from soloing, grouping, raiding, with and without discs and with and without support. Basically a mish-mash of everything.

$ /c/jkb/eq/bs_split2.pl Backup/eqlog_Brogett_antonius_20131027.txt
All BS = 807180845

=== backstab 388075952 48.077943673205 ===
Normal: 28901 hits, average 6664.98927372755 min 144, max 60570
Critical: 11442 hits, average 17081.8997552875 min 571, max 81571
Hit rate of 92.0274647545046%
Crit rate of 28.3617975857026%
17865 rounds
1 per round = 4000 22.39%
2 per round = 2073 11.6%
3 per round = 11773 65.89%
4 per round = 19 0.1%

=== massive strike 324967722 40.2595928797096 ===
Normal: 14866 hits, average 14137.1177855509 min 427, max 49121
Critical: 4658 hits, average 24646.9147702877 min 778, max 115195

=== battery 94137171 11.6624634470854 ===
Normal: 2924 hits, average 28550.6744186047 min 2747, max 100570
Critical: 344 hits, average 30973.8343023256 min 6865, max 82809

Total hits = 63135
Total miss = 3495
=> total hit rate of 94.7546150382711%

Again Incursion (battery) has weaker crit than normal and even the average crit is only barely higher than the average non-crit. The crit rate is way down vs other forms of attack too, demonstrating that issue again. Given that this is averaged across all forms of my game play it shows the effect is significant and noticable on average and not just in tightly controlled circumstances.




pretty weird stuff.

Nedrom
11-27-2013, 10:14 AM
Anyone get a chance to try this out?

Nedrom
12-10-2013, 04:41 PM
No one parsing anymore?

Ishtass
12-10-2013, 08:15 PM
I don't care enough to parse :)

You want it, you get it done :)

tanecho
12-20-2013, 06:10 PM
Looks like it most certainly does. Kick test run firing synergy + stunning kick + flying kick as much as possible. Either there is some strange partial crit suppression on thunderfoot, or we have the same issue as rogues with overflow crits.

Rapid Kick + Zan Fi active (15% damage modifier): 125499 total kicks fired. Difference between highest crit and highest non-crit was 9.2%, and the only time in these 2 runs that the higher number is the crit. Difference between averages was 14.8%, with the average normal being higher.

Heel of Zagali, Zan Fi, Infusion of thunder (85% damage modifier): 110730 kicks fired. Difference between highest crit and highest non-crit was 74.8%.... with the non-crit being higher. Difference between average crit and average non-crit was 30.3%... again favoring the non-crit.

Not posting numbers as someone would just use them as nerf-bait, and I used a disc combination that is rarely used (at least in my case) to make the issue more apparent.

Other fun findings from my parsing run:

EOTS does not appear to be a flat damage bonus of the number in spell files. It added 277 to both my max crit and non-crit hits with my h2h weapons, and added 251 to both of those with my 2hb. Minimums went up more, but likely more as a function of the +minimum damage portion of the spell. I ran it with infusion of thunder in all cases to override the max portion to see the flat damage boost. I also tested Zan Fi, it was 85 to max/min crit/non in all cases for h2h, and 185 in all cases for 2hb, as expected.

Full raid buffs, including bard songs, I tested the T3 h2h and T4 class weapon against the Eye of Lxanvom. Parses ran a minimum of 8 hours, anywhere from 1.75-2.00 billion damage apiece to get a reasonable confidence in the accuracy of the numbers

Terrorpalm: Eye of Lxanvom won by a margin of 5.6%
Speed Focus + Infusion: Eye of Lxanvom won by a margin of 4.1%

Obviously, better weapons will make up parts of the margin, but I think we will reach a point where we have to weapon swap for optimal DPS depending on discs. The terrorpalm scenario did not include Destructive force as it would in a normal combat setting, and DForce will favor a 2hb weapon.

I didn't have time in this run to do DPS of EOTS h2h vs. 2hb, but my findings on the damage bonus lead me to the conclusion the h2h will likely win that one.

I did these for my own reference, just sharing the findings with everyone else. If anyone wants more information, let me know.

Ishtass
12-21-2013, 12:34 AM
Nice Tan, thanks

Nedrom
12-21-2013, 05:39 AM
Sojuu can you pass this on so when rogue BS crits are looked at our special attacks are also looked at? I had a feeling this was not rogue specific and probable that it affects most or not all special attacks for all melee.

brogett
12-22-2013, 07:09 PM
Looks like it most certainly does. Kick test run firing synergy + stunning kick + flying kick as much as possible. Either there is some strange partial crit suppression on thunderfoot, or we have the same issue as rogues with overflow crits.


Do you also see the crit percentage dropping? When I did my backstab tests I found that without my Rogue's Fury AA (large damage multiplier) I'd be getting around 25-30% crit rate on Incursion (a high backstab skill attack), but with Rogue's Fury activated it'd drop to 2-3% instead and all those crits were low ones.

My hypothesis was that the high crits were simply being eaten, but looking further it appeared that it's more likely the high crits were being turned into non-crits instead, so in addition to the crit roll check it seems that maybe there is another check for whether it is too high. That is after various damage mods get applied.

Also what value are you seeing this? With backstab it's pretty low. My max crit for incursion was something only around 48k, which is rather pathetic. I get the feeling it can't be that low for the kicks, so maybe there is something else I haven't taken into account. Aristo asked for as much data as we can get, specifically whether it happens that certain values can crit while other values cannot, or for the backstab example whether it's visible with one weapon but not with another. (It's not that clear alas as the biggest backstab values aren't weapon based by instant-use disc-attack based).

tanecho
12-22-2013, 08:23 PM
87687 was the highest crit. Keep in mind though that I was using synergy, so the mob had a 122% incoming kick damage increase, so that may play a role as well.

Crit rates:
Without infusion: 21%
With Infusion: 18.9%

And yes, I think it is likely that some are simply being called regular attacks. In fact, on one of the parses I ran for eye of the storm I had a crit on my 2h weapon converted to normal and threw my max hit numbers off.

I wish I knew the exact damage bonus our kicks got from AA, so I could better construct some estimates to compare. I think it was 1440 before this round of AAs, but that number is likely FMA and wouldn't be accurate after the COTF AAs anyway.

brogett
12-23-2013, 08:41 AM
87687 was the highest crit. Keep in mind though that I was using synergy, so the mob had a 122% incoming kick damage increase, so that may play a role as well.

Crit rates:
Without infusion: 21%
With Infusion: 18.9%

And yes, I think it is likely that some are simply being called regular attacks. In fact, on one of the parses I ran for eye of the storm I had a crit on my 2h weapon converted to normal and threw my max hit numbers off.

I wish I knew the exact damage bonus our kicks got from AA, so I could better construct some estimates to compare. I think it was 1440 before this round of AAs, but that number is likely FMA and wouldn't be accurate after the COTF AAs anyway.

21 to 19% drop doesn't sound very significant unless it's a huge sample size. You'll need approaching 2000 in order to have around 95% confidence that the drop isn't just a fluke.

Unless synergy applies to a different portion of the attack function from rogue's fury (anything is possible in EQ code!) then I don't think that impacts things much. Although something is curious as clearly there are different thresholds and not all cutting out at one value, which is unlikely. So somewhere earlier on in the formula there looks to be a limit, but it's not quite clear where. If it was before Synergy then it'd make zero difference to your crit rates regardless. For RF I certainly see crit rates changing with and without it, which implies the bug is after RF has been applied.

Also when I said hits get turned into non-crit, I don't believe they are still hitting for crit damage values but reported as non-crit. I believe they're having the crit multiplier removed too. So I don't see how it would change your max hit value. I ought to do some proper testing on the test server I guess to get more data.

tanecho
12-24-2013, 02:04 AM
Sample size was over 100k. In the first test, only 7599 of 125499 were thunderfoot procs, or about 6% (counting twincasts as 2). It's unlikely that too many attacks would cause overflow for a monk, other than thunderfoot. If a large portion of them were converted to non-crit, it would be nearly 2% of our attacks. Monks have abysmal crit rates, 2% is a big chunk to us.


Synergy is a debuff on the mob, that increases incoming flying kick damage, much like your pinpoint line. This calculation is applied after your damage calculation, so in display a critical thunderfoot looks like this.

[Fri Dec 20 07:48:59 2013] You begin casting Thunderfoot.
[Fri Dec 20 07:48:59 2013] Tanecho scores a critical hit! (39462)
[Fri Dec 20 07:48:59 2013] You kick Test One Hundred Three for 87687 points of damage.

I have never seen that 2nd line reach 40k.

I ran 2 tests, one with +15% outgoing, +122% incoming, and the other with +85% outgoing, +122% incoming. The maximum crit in both cases was within about 800 damage, but there was a wild swing between the max non-crits (+94%), larger than could possibly be accounted for by the 70% increase damage bonus alone.

brogett
12-24-2013, 07:16 AM
Ah I see, it's a compound distribution, hence why it appears only a small percentage drop. When I see this bug with backstab the crit drop is HUGE. I think it would be informative to split the distribution into components - the various different types of kick basically. Gamparse can't do it, but I have a hacky perl script to do this (3 ways) for the various types of backstab hit.

I haven't been able to parse boosts to incoming damage on rogue as they're up for such a short time with a relatively long reuse (and a resisty one at that). So my boosts have mainly been to my outgoing damage, and that appears to be applied before the crit calculation stage.

The hit distribution can be informative too. I copied to test server and did a /resetaa to obliterate most, then bought a few essentials back (combat fury for crit chance, triple backstab to boost rate of gathering parse data), Rogue's Fury so I had a non-disc way to boost outgoing dps. I used Twisted Chance discipline which is a crit rate modifier (500%); not the 100% that would make things trivial to demonstrate, but rogues don't have that. Maybe berserkers can parse this with their 100% crit rate disc to demonstrate the bug in a clearer fashion.

Anyway the distribution then shows lots of useful data. Eg:

http://i39.tinypic.com/necj9w.png

The above isn't a long parse, but is long enough to witness the basic components at play. Massive Strike is coded at a specific damage value, so the only variables are the DI roll (1 to 20) and the PCDamageTable roll (1.00 to 3.95 for rogues, in 0.01 increments).

This was after testing Assassin's Wrath rank 1 (crit multiplier) and annoyingly I can only resetaa once a week. Gah! The min/max non-crit were 5099/29570 with +210% hit (and min hit) mod from Rogue's Fury. Massive Strike is a backstab skill attack for 1200. Damage bonus looks like 179 (varies by AA, but I ditched most).

Min=179 + 4.1*1200 (some combination of ferocity+min from RF I guess)
"2x"=179 + 3.1*2*1200
Max=179 + 3.1*(1+2*1200*3.95)

The 2x spike is the second high spike from left, corresponding to the max DI roll (frequent) but minimum DamageTable roll (1.00). We also see lots of other significant spikes which come from DI 1 to 19 combined with the max DamageTable roll (with DI 20 being the max hit on right). Eg DI 19 would be 179+3.1*(1+19/10*1200*3.95) = 28100, exactly where I see it (ok with a +1 fudge in there ;-)).

Anyway, all the components of the formula are present and correct, as expected. Now look at the crits instead:

http://i39.tinypic.com/fc2sn.png

The 2x spike is 13580, explained by 179 + 1.8*(5 + 3.1*2*1200)
The min is again 179 + 1.8*(5+4.1*1200) = 9044. As observed.
The max *should* be 179 + 1.8*(5+3.1*(1+2*1200*3.95)) = 53091, but we don't observe it. Indeed we don't even observe the DI roll 14 (37222). The last is 13 (=> 34577).

That implies for us it's capping somewhere between the max I observed (36083) and 37222. I need a lot longer parse to get the small blips between the main DI rolls. (They correspond to DamageTable rolls that aren't 1.00 nor 3.95, but somewhere in between - much rarer.) There's nothing odd about that number. I was expecting 32768 initially, but it didn't pan out.

Why it differs to the ~40k you see I do not know. I should try other ranks of Assassin's Wrath to test up to 2.3x crit multiplier. Does the max figure change? If so it's a cap pre-AA, maybe hard coded with the original 1.7 EQ crit-mod.

I can also observe this using Incursion without needing Rogue's Fury, but it's HARD to obtain data. I run out of endurance quickly as the auto-regen seems broken on Test server and the crit rate is so weak it's hard to obtain 1000s of samples. I need a way to skew the DamageTable roll, but it's likely fixed. (I assume it's attack vs AC which is skewing the DI roll, as that will be common to NPCs.)

Edit: One useful thing to analyse though is the delay between incursion attacks. It's another triggered AA on 21sec reuse (with appropriate AA) and is only ever one backstab hit (although it can proc Massive Strike just like normal stabs). The delay is always 21 sec in the logs, regardless of whether or not it crit. Yet we have far fewer crit messages than we should due to the bug. This is clear demonstration that the critical hits aren't being thrown away, but that the crit is being cancelled and it's landing as a normal hit instead. In theory we should also see an enrichment at the higher end for normal hits than the lower end, as that'll be the 20-25% critical hits being transferred over. (The data is too weak to observe that though.)

brogett
12-24-2013, 09:55 AM
With Assassin's Wrath 3 (2.0 multiplier for crits) and a longer parse my max Massive Strike crit was 40073:


$ /c/jkb/eq/bs_split2.pl eqlog_Brogett_test.txt
All BS = 143471369

=== backstab 82658349 57.6131318576879 ===
Normal: 6720 hits, average 8294.840625 min 2429, max 13626
Critical: 1656 hits, average 16254.2391304348 min 4689, max 27083
Hit rate of 91.3214129861185%
Crit rate of 19.7707736389685%
3300 rounds
1 per round = 0 0%
2 per round = 730 22.12%
3 per round = 2570 77.87%
4 per round = 0 0%

=== massive strike 60813020 42.3868681423121 ===
Normal: 2760 hits, average 19109.018115942 min 5099, max 29570
Critical: 360 hits, average 22422.5833333333 min 10029, max 40073

=== battery 0 0 ===

Total hits = 11496
Total miss = 796
=> total hit rate of 93.5242434103482%


This is below the expected amount still (58971).

It's starting to fit a pattern of 20k being the upper cap (before damage bonus is applied), *before* the crit multiplier takes place. So with 1.7x crit it's 34k cap, with 1.8x it's 36k, with 2.0x it's 40k. This is what I observe.

Do monks have any AAs to boost the critical multiplier for kicks? If so how many ranks? If it follows Veteran's Wrath and Assassin's Wrath it'll be +0.1 per rank. If you have none, do kicks get boosted by Veteran's Wrath? It's easy enough to work out the multiplier if you look at the distributions.

tanecho
12-24-2013, 10:16 AM
One other thing, ran one more kick test, same parameters as 2nd test, but added First spire.

Thunderfoot proc rate:
Test 1: 6.0318%
Test 2: 6.0550%
Test 3: 6.2036%

sojuu
12-24-2013, 12:56 PM
Sojuu can you pass this on so when rogue BS crits are looked at our special attacks are also looked at? I had a feeling this was not rogue specific and probable that it affects most or not all special attacks for all melee.

Believe they are on xmas vacation atm, not much being said in channels.

tanecho
12-24-2013, 01:44 PM
Monks do not have anything other than Veteran's Fury for crit damage, to my knowledge.

The damage interval for a normal flying kick is *extremely* tight. Our special damage does not work like backstab, in that we still have our POP era DI tables, with a huge number (somewhere in the realm of 2000) just tacked on. That huge number does not get modified by disciplines. Easy enough to see if you tiger claw a few things, hit ironfist (280% special damage mod) and tiger claw some more. Your damage doesn't go up 280%, it goes up about 20.

Thunderfoot is a skill attack for 3000.

tanecho
12-26-2013, 12:40 AM
Using a methodology more similar to yours, I removed all specials other than baseline flying kick to restrict the damage types to FK and thunderfoot, and remove the effects of synergy (which doesn't have 100% uptime.)

I ran a much smaller parse (21 minutes) with 280% damage mod on, during which I fired off 42 thunderfoot procs. Only one of them crit, for 29204. The most frequent value was the highest, 26@98700, and 7 of them hit for 26595.

Removing the damage mod and repeating for 30 minutes, I had 7 crits, 5@21362, a 28283, and a 32225. For noncrits, the highest value was 34658, occuring 52 times, the lowest 12505 occurring 7 times.

It's hard to break down where my spikes are occuring with the tools I have available. I can't find a decent parser for damage intervales. Gamparse just breaks every hit number into its own column, the distance between a hit for 5k and 30k looks the same as every other on my big parses, and there are so many unique values that the numbers on the axis are unreadable. I have a 90,000 kick parse, but it's not of much use other than to note that my max normal was 98k, and my max crit was in the 30s.

brogett
12-27-2013, 06:47 AM
Using a methodology more similar to yours, I removed all specials other than baseline flying kick to restrict the damage types to FK and thunderfoot, and remove the effects of synergy (which doesn't have 100% uptime.)

It'll help further if you can break apart normal kick from thunderfoot procs. You may be able to do it later, but it's easier if the log file has messages than indicate this (for massive strike it's annoyingly the line *after* the appropriate backstab).


I ran a much smaller parse (21 minutes) with 280% damage mod on, during which I fired off 42 thunderfoot procs. Only one of them crit, for 29204. The most frequent value was the highest, 26@98700, and 7 of them hit for 26595.

Removing the damage mod and repeating for 30 minutes, I had 7 crits, 5@21362, a 28283, and a 32225. For noncrits, the highest value was 34658, occuring 52 times, the lowest 12505 occurring 7 times.

I tried figuring out the maths from these numbers, but can't get consistent answers. I think I'd need to see the actual log file to work it out, or at least the distribution of hit values (and whether or not they crit).



It's hard to break down where my spikes are occuring with the tools I have available. I can't find a decent parser for damage intervales. Gamparse just breaks every hit number into its own column, the distance between a hit for 5k and 30k looks the same as every other on my big parses, and there are so many unique values that the numbers on the axis are unreadable. I have a 90,000 kick parse, but it's not of much use other than to note that my max normal was 98k, and my max crit was in the 30s.

I tend to just do it via observation of min,2x and max peaks. I say "2x" (no idea what to call it) as it's basically the damage bonus plus 2x weapon damage peak, coming from DI roll of 20 and Damage Table roll of 1.00. Observing the 2x and max value spikes are enough to work out the damage table. Minimum value is harder to use as it'll vary depending on worn ferocity and other buffs affecting minimum hit. So basically look for 3 key spikes, with the 2nd being closer to left edge than right.

Without any minimum hit modifier or worn ferocity, you may be able to work out all the DI rolls, but it's generally not needed.

Gamparse doesn't have an option to dump out the histogram as raw value,count tuples, and its axis tend to be a bit elastic, adjusting them to get a nice looking plot. If it did allow a dump of raw data it would be useful data to cut and paste here. (I normally do that sort of thing on the command line through awk or perl.)

Lyndar
12-27-2013, 04:28 PM
What would be the ideal tool? number of repetitions for each kick value?

brogett
12-27-2013, 08:17 PM
Broken down by type of attack too, and breaking apart crit vs non-crit. I have a backstab equivalent which does the following:

bs_split bs_type crit_type < eqlog_blah.txt

where bs_type is 0 for normal backstab, 1 for massive strike, 2 for incursion and crit_type is 0 for normal and 1 for critical hit. It produces a text file of "value frequency" per line, so I can just load it into gnuplot.

It's just a hacky perl script. For simple stuff though like pierce or crush I use a one-liner. Eg:

awk '/critical/ {c=1} /You pierce/ {if (!c) {d[$12]++};c=0} END {for (i in d) {print i,d[i]}}' eqlog_Brogett_test_no_ferocity.txt|sort -n > a

An example file "a" ends with:

1045 13
1047 1
1048 11
1049 119
1050 13
1052 12
1053 1
1055 16
1057 9
1060 13
1062 5
1064 14
1067 13
1069 13
1072 10
1074 13
1076 14
1079 8
1080 2
1081 15
1084 11
1085 1
1086 13
1087 1
1088 12
1089 1
1091 8
1092 1
1093 10
1096 140
1098 13
1100 10
1103 9
1105 7
1108 15
1110 13
1112 8
1115 5
1117 15
1120 9
1122 12
1124 12
1127 13
1129 13
1132 12
1134 8
1136 8
1139 9
1141 11
1144 2748

I can generate hits with pretty much the exact same distribution, including the bigger internal spikes and the impossible to obtain values like 1140, 1142, 1143 using a weapon damage simulator (also a hacky perl script). I plan on adding a critical hit bug in it to test the effect on dps, and also to see what it does to the damage distributions.

Happy to post perl code if anyone cares, but it's kind of hacky and has lots of comments to my earlier experiments embedded throughout.

Edit: so for kicks, you'd need to monitor the log lines so you know what type of kick is being generated and whether it's a thunderfoot proc or not. I don't know the corresponding log messages, but I'd happily amend my bs_split script if someone sent me a log file.

tanecho
12-27-2013, 09:36 PM
"You begin casting Thunderfoot." or "You twincast Thunderfoot." You could actually just key it off the word Thunderfoot to pick up the next numerical value outside of timestamps and parentheses.

The number outside of () or [] is a thunderfoot kick. Isolate next * where
" for " * " points of"

Lyndar
12-27-2013, 10:28 PM
I may have some time to play with code for this - but I don't have thunderkick yet, so my logs won't have any. Can someone post a few example thunderfoot or thunderfoot crit lines?

I'm assuming its like crits, so it'd be something like:
* You begin casting Thunderfoot.
* You kick xxxxx for yyyy points of damage.

or
* You begin casting Thunderfoot.
* Lyndar scores a critical hit! (xxxxx)
* You kick xxxxx for yyyy points of damage.

tanecho
12-27-2013, 10:42 PM
See above, remove critical line for non-crits.

[Fri Dec 20 07:48:59 2013] You begin casting Thunderfoot.
[Fri Dec 20 07:48:59 2013] Tanecho scores a critical hit! (39462)
[Fri Dec 20 07:48:59 2013] You kick Test One Hundred Three for 87687 points of damage

tanecho
12-27-2013, 10:46 PM
You need to have a parser read for "Thunderfoot." + Line break + * + "critical hit!" + * + Line break + * + "for ", then capture the number here that occurs before the string " points of damage." Potentially 4-6 digits.

That gets crits. To get non-crits you need to do the same as above, but cut out the critical hit and second line break pieces.

tanecho
12-27-2013, 11:55 PM
I used the above logic to just parse all of the data in excel, using if statements and string searches. I isolated thunderfoot, and split them into critical and non-critical values.

For the non-crits (3971 total), here are the highlights from the distribution of the damage values.


Damage Count of Hits
97840 2195
25735 433
88266 85
73905 79
93053 78
59544 77
64331 73
54757 72
69118 72
83479 72
78692 71
49969 62
34658 45
98700 26
23371 21
22189 18
19825 14
24553 14
18643 12

For the crits, there is no real distinctive spike, and only 35 crits happened for all of those non-crits. Here is the entire distribution

Damage Count of Crits
26157 6
34904 6
21362 5
28344 5
30531 3
37091 3
29204 2
32718 2
23971 1
28283 1
32225 1

brogett
12-28-2013, 07:35 AM
I used the above logic to just parse all of the data in excel, using if statements and string searches. I isolated thunderfoot, and split them into critical and non-critical values.

For the non-crits (3971 total), here are the highlights from the distribution of the damage values.


Damage Count of Hits
97840 2195
25735 433

That's the "2x spike" and max hit spike (well I see some 98700 too, but they're presumably an accidental focus or something?).

Looks like damage bonus is 2095.
Damage is 3000 (from thunderfoot rank 3).
Damage Table is 4.05
Damage multiplier is 3.94 (1 + 2.80 + 0.14). Where is the 0.14 from? Buff?

Ie:

25735 = 2095 + floor(3.94*round(2*3000))
97840 = 2095 + floor(3.94*floor(round(2*3000)*4.05+1))

Did you cut out infrequent values from your table though? I'm suprised to not see the less frequent inbetween values. Eg I'd expect to observe (given enough samples) 97604, 97368, 97131, etc from Damage Table 4.04, 4.03, 4.02 and so on.


Damage Count of Crits
26157 6
34904 6
21362 5
28344 5
30531 3
37091 3
29204 2
32718 2
23971 1
28283 1
32225 1

With critical hit multiplier 1.7, even the 2x spike would go over 20k. So it needs DI rolls less than 20 to get a chance to critical hit. I can't see any obvious values falling out of this though. It's around 2187 between the sums, but that doesn't correspond easily to any DI roll interval that I can see, unless I get nonsensical crit multipliers less than 1 (0.45685).

Baffled by those numbers. Perhaps they're coming from Damage Table rolls that aren't max (so 4.04, 4.03, etc) rather than DI rolls 1 to 19. It'd be easier to see crit tables that work (ie without the +280% damage mod). Then you could work out the proper crit multiplier. I'm assuming 1.7, but don't know for sure for kicks.

tanecho
12-28-2013, 01:24 PM
Yes, it was truncated, here is the complete distribution if you want it.

Row Labels Count of Noncrits
Grand Total 3971
97840 2195
25735 433
88266 85
73905 79
93053 78
59544 77
64331 73
54757 72
69118 72
83479 72
78692 71
49969 62
34658 45
98700 26
23371 21
22189 18
19825 14
24553 14
18643 12
13915 10
16279 10
15097 7
26595 7
12505 6
45182 6
17461 5
21007 5
63326 5
23261 4
41104 4
69472 4
81292 4
35431 3
35904 3
39213 3
40395 3
42050 3
45832 3
57416 3
60962 3
68763 3
77746 3
79165 3
90748 3
95004 3
12733 2
18377 2
25975 2
26211 2
26684 2
28145 2
28575 2
29048 2
29757 2
31402 2
32358 2
32523 2
35194 2
35667 2
38268 2
40632 2
41341 2
42759 2
43197 2
43705 2
44178 2
47014 2
48433 2
49378 2
50088 2
50560 2
51270 2
51742 2
52215 2
52452 2
52688 2
53870 2
54934 2
57180 2
58834 2
59780 2
60253 2
60489 2
62380 2
64035 2
65217 2
65926 2
66163 2
67345 2
68054 2
68527 2
69000 2
69709 2
70654 2
71127 2
71364 2
72782 2
73018 2
77510 2
79401 2
82474 2
83893 2
86966 2
87675 2
88621 2
89803 2
91221 2
91458 2
95240 2
96422 2
96658 2
16572 1
17110 1
21956 1
22192 1
22523 1
22535 1
22642 1
23398 1
23457 1
23723 1
23800 1
24125 1
24889 1
25006 1
25130 1
25301 1
25502 1
25511 1
25738 1
26067 1
26448 1
26779 1
27062 1
27157 1
27582 1
27866 1
28102 1
28150 1
28221 1
28339 1
28812 1
28906 1
29284 1
29521 1
29544 1
29774 1
29994 1
30396 1
30703 1
30821 1
30939 1
31648 1
31885 1
32121 1
32594 1
32641 1
32736 1
32984 1
33030 1
33540 1
33953 1
34012 1
34249 1
34485 1
34958 1
35112 1
35384 1
35573 1
35951 1
36376 1
37133 1
37204 1
37322 1
38031 1
38055 1
38504 1
38977 1
39331 1
39473 1
39922 1
40159 1
40868 1
41577 1
41684 1
41814 1
42192 1
42286 1
42523 1
42700 1
42996 1
43126 1
43137 1
43232 1
44095 1
45123 1
45360 1
45454 1
45596 1
46246 1
46423 1
46542 1
46920 1
47274 1
47487 1
47605 1
47771 1
48196 1
50040 1
50324 1
50501 1
50797 1
50829 1
50915 1
51979 1
53303 1
53634 1
54343 1
54579 1
54745 1
54816 1
55288 1
55525 1
55617 1
55761 1
55856 1
55998 1
56222 1
56234 1
56943 1
57156 1
57274 1
58125 1
59071 1
59366 1
60726 1
61198 1
61376 1
61494 1
61671 1
62144 1
62782 1
62853 1
63090 1
63633 1
63799 1
64272 1
64508 1
65454 1
65643 1
66399 1
66636 1
66872 1
67108 1
67581 1
67699 1
67818 1
67900 1
69945 1
70182 1
70749 1
70891 1
71836 1
72073 1
72309 1
73728 1
74189 1
74673 1
74862 1
75382 1
75619 1
76092 1
76245 1
76328 1
76564 1
76801 1
77037 1
77274 1
78928 1
79460 1
79874 1
80110 1
81056 1
81600 1
81671 1
82002 1
82947 1
83846 1
84129 1
84339 1
86091 1
86408 1
86644 1
87439 1
87912 1
88148 1
88857 1
89094 1
90512 1
90985 1
92167 1
92640 1
93112 1
93349 1
93585 1
93822 1
94058 1
94294 1
94531 1
94767 1
95713 1
95949 1
97131 1
97368 1
97604 1


Going to start a new set, stripped of all buffs (except haste) and removing ferocity to reduce variables.

Lyndar
12-28-2013, 06:57 PM
I wrote a quick console app in .NET to do the above - number of occurrences per damage number per crit/non-crit/tkick/tkick-crit. I dont have tkick yet, but if anyone wants it i can pretty it up and email it.

Sample output:
5434 kicks: [4512 normal, 922 crits, 0 thunderkicks, 0 thunderkick crits]
--------------------------Normal----------------------
1699 x 908
1698 x 1
1695 x 4
1694 x 3
1693 x 2
1691 x 1
1690 x 2
1687 x 3
1686 x 1
1684 x 6
1683 x 4
1682 x 5
1679 x 4
1676 x 1
1675 x 1
1674 x 3
1673 x 2
1671 x 43
1670 x 3
1669 x 2
1668 x 3
1667 x 4
1666 x 2
1665 x 2
1662 x 10
1661 x 4
1660 x 6
1659 x 2
1658 x 2
1654 x 2
1653 x 538
1652 x 4
1651 x 5
1650 x 3
1649 x 3
1648 x 4
1647 x 2
1646 x 4
1645 x 3
1644 x 42
1643 x 2
1642 x 3
1641 x 3
1640 x 4
1639 x 3
1638 x 1
1637 x 3
1636 x 2
1635 x 6
1634 x 4
1633 x 6
1632 x 4
1631 x 2
1630 x 1
1629 x 2
1628 x 21
1627 x 2
1626 x 4

----truncated to avoid spamming

brogett
12-28-2013, 07:42 PM
Did you cut out infrequent values from your table though? I'm suprised to not see the less frequent inbetween values. Eg I'd expect to observe (given enough samples) 97604, 97368, 97131, etc from Damage Table 4.04, 4.03, 4.02 and so on.


Yes, it was truncated, here is the complete distribution if you want it.

Row Labels Count of Noncrits
Grand Total 3971
97840 2195
25735 433
88266 85
73905 79
...snipped...
95949 1
97131 1
97368 1
97604 1


Going to start a new set, stripped of all buffs (except haste) and removing ferocity to reduce variables.

Bingo! Well at least I know my melee hit model is working :-)

That gives me more hope that my theory of 20k pre-crit pre-DB is correct too. Once the devs get back I reckon that should be enough to nail this. God I love parsing. lol

Ishtass
12-28-2013, 07:57 PM
Brogett reminds me of the Rainman.

On the weekends, he helps Tom Cruise gamble.

tanecho
12-28-2013, 08:30 PM
Here is a smaller run, no fero, no buffs except haste and proc increaser. Trying to investigate a few other things now using different multipliers.

Row Labels Count of Crits
Grand Total 161
14122 42
28897 9
33772 7
11113 6
26461 6
24023 5
9308 4
21585 4
31335 4
7502 3
12317 3
12918 2
32057 2
33140 2
7186 1
8099 1
8706 1
9909 1
10511 1
11006 1
13520 1
14243 1
14605 1
15122 1
15207 1
15905 1
16048 1
16650 1
17133 1
18552 1
18697 1
18836 1
19057 1
19443 1
19763 1
19780 1
19846 1
19900 1
20021 1
20123 1
20983 1
21345 1
21634 1
22037 1
22850 1
23030 1
23270 1
24234 1
24835 1
24859 1
24992 1
25160 1
25558 1
26279 1
26399 1
26641 1
26762 1
26881 1
28205 1
28446 1
29408 1
29529 1
30522 1
30974 1
31605 1
31695 1
32178 1
32417 1
33019 1
33742 1
33862 1
33988 1
34102 1
34223 1
34344 1
34729 1


Row Labels Count of Noncrits
Grand Total 1826
30753 817
9158 164
25018 48
27885 36
22150 32
29319 30
26452 28
23584 27
17849 26
19283 26
20717 24
16416 21
7388 11
13548 11
14982 9
6680 8
5618 7
7034 7
5972 6
8096 5
12840 5
12982 5
16026 5
20911 5
5264 4
7742 4
8450 4
8804 4
10645 4
12115 4
17584 4
26292 4
27496 4
27567 4
29053 4
30328 4
6326 3
9159 3
9371 3
11991 3
12557 3
13194 3
14398 3
16663 3
16876 3
20841 3
20982 3
21336 3
23035 3
23389 3
24097 3
24381 3
25584 3
25797 3
25938 3
27354 3
28275 3
28983 3
29549 3
30399 3
30682 3
4910 2
9513 2
9654 2
9725 2
10221 2
10575 2
10653 2
10716 2
11070 2
11212 2
11424 2
11566 2
12203 2
12699 2
12784 2
13293 2
13477 2
14256 2
14893 2
15460 2
15531 2
15601 2
15885 2
16097 2
16168 2
16451 2
16947 2
17655 2
17679 2
18504 2
18717 2
19000 2
19141 2
19708 2
19991 2
20416 2
20770 2
21265 2
21549 2
21903 2
22257 2
22327 2
22965 2
23106 2
23319 2
24664 2
26717 2
26788 2
27000 2
27637 2
27708 2
28062 2
28204 2
28770 2
29195 2
29407 2
29691 2
29761 2
29832 2
4556 1
5024 1
5028 1
5866 1
6553 1
6635 1
6773 1
7141 1
7258 1
7336 1
7527 1
7608 1
7884 1
7955 1
8114 1
8185 1
8252 1
8380 1
8387 1
8486 1
8557 1
8720 1
8819 1
9006 1
9102 1
9300 1
9612 1
9661 1
9796 1
9937 1
10008 1
10150 1
10235 1
10553 1
10610 1
10681 1
10787 1
11141 1
11382 1
11573 1
11693 1
11778 1
11814 1
11884 1
11923 1
12051 1
12132 1
12345 1
12359 1
12415 1
12529 1
12709 1
12769 1
12805 1
12865 1
12911 1
12985 1
13088 1
13092 1
13123 1
13177 1
13230 1
13254 1
13265 1
13336 1
13393 1
13407 1
13442 1
13520 1
13619 1
13690 1
13814 1
13831 1
13902 1
13973 1
14115 1
14185 1
14306 1
14327 1
14423 1
14539 1
14610 1
14681 1
14695 1
14752 1
14897 1
14950 1
14993 1
15078 1
15106 1
15318 1
15354 1
15389 1
15651 1
15708 1
15732 1
15743 1
15814 1
15860 1
16239 1
16341 1
16352 1
16522 1
16593 1
16805 1
17088 1
17180 1
17230 1
17244 1
17265 1
17357 1
17371 1
17513 1
17605 1
17690 1
17867 1
17881 1
17938 1
18079 1
18292 1
18363 1
18391 1
18554 1
18787 1
18964 1
19071 1
19106 1
19177 1
19241 1
19276 1
19297 1
19411 1
19425 1
19432 1
19704 1
19779 1
19793 1
19849 1
20012 1
20133 1
20274 1
20345 1
20395 1
20430 1
20557 1
20628 1
20699 1
20777 1
20812 1
20823 1
20876 1
20961 1
21053 1
21124 1
21131 1
21195 1
21407 1
21460 1
21478 1
21506 1
21619 1
21676 1
21690 1
21733 1
21832 1
21973 1
22023 1
22097 1
22115 1
22186 1
22189 1
22204 1
22469 1
22526 1
22681 1
22752 1
22795 1
22929 1
23248 1
23460 1
23489 1
23673 1
23715 1
23814 1
24027 1
24168 1
24239 1
24310 1
24522 1
24593 1
24621 1
24791 1
25159 1
25230 1
25301 1
25372 1
25513 1
25655 1
25726 1
25729 1
26080 1
26331 1
26646 1
27071 1
27213 1
27283 1
27425 1
27779 1
27850 1
27921 1
27974 1
28345 1
28558 1
28629 1
28912 1
29117 1
29266 1
29337 1
29620 1
29903 1
29974 1
30045 1

tanecho
12-29-2013, 02:31 AM
And again, with intensity of the resolute running to give a 55% damage mod (but no minimum)

The strangest part of all of this to me, is that I had a sample with nearly 10x the kicks, and in these smaller samples I am getting crits WAY higher than anything in there. 34486 max crit in this run, but till the max hit is higher. The max hit of 44118 occurs 607 times, no hit was higher so it's pretty easy to see that this is the max hit spike. With 1576 hits I would have expected about 320 crits instead of the 101 I got, and I would have expected one of them at least to beat that 44118, so I think it's pretty plain to see the system is just ignoring crits at some point.

The highest crit in this run was lower than the highest in the unmodified run, 34729.

Row Labels Count of Crits
Grand Total 101
19732 48
14438 3
10027 2
13556 2
15321 2
18850 2
27323 2
34248 2
12473 1
12674 1
17085 1
17967 1
18763 1
19064 1
19822 1
20264 1
20616 1
20970 1
21498 1
21675 1
21852 1
22557 1
22734 1
22875 1
22911 1
24410 1
24914 1
25028 1
25946 1
26615 1
27674 1
28205 1
29334 1
29748 1
29959 1
30462 1
30498 1
30675 1
31380 1
31928 1
32440 1
33057 1
33851 1
34204 1
34486 1
9144 1


Row Labels Count of Noncrits
Grand Total 1475
44118 607
12458 128
29405 31
25201 28
23099 27
31507 27
35710 26
37812 26
39914 23
42016 21
27303 18
20997 17
33608 16
7787 9
10382 7
10901 7
11420 6
16793 5
17961 5
18895 5
29690 5
42042 5
7268 5
14224 4
14639 4
15781 4
16819 4
17338 4
35918 4
39759 4
41316 4
11939 3
13912 3
14950 3
15988 3
19829 3
22009 3
23670 3
26991 3
29171 3
30520 3
30728 3
35399 3
37786 3
38721 3
43392 3
6749 3
8825 3
12874 2
12978 2
13082 2
13186 2
13373 2
13601 2
13705 2
14120 2
14535 2
14743 2
15262 2
15885 2
17753 2
18064 2
18480 2
19102 2
19414 2
19518 2
20140 2
20348 2
21178 2
21801 2
22113 2
22320 2
22735 2
22839 2
22943 2
23462 2
24189 2
24500 2
24811 2
25227 2
25746 2
26161 2
26265 2
26576 2
26680 2
27614 2
27925 2
28133 2
28237 2
28548 2
28652 2
28756 2
29275 2
29794 2
30105 2
30313 2
31039 2
31455 2
31766 2
31974 2
32596 2
33012 2
34257 2
34672 2
35814 2
36022 2
36126 2
37267 2
38098 2
38513 2
38824 2
39240 2
40174 2
40693 2
41004 2
41419 2
42354 2
42665 2
44014 2
8306 2
9863 2
10331 1
10446 1
10534 1
10757 1
10814 1
10923 1
11053 1
11089 1
11157 1
11380 1
11546 1
12200 1
12210 1
12397 1
12449 1
12667 1
12750 1
12760 1
12771 1
13098 1
13196 1
13637 1
13944 1
13990 1
14016 1
14431 1
14613 1
14764 1
14784 1
15345 1
15469 1
15776 1
15838 1
16092 1
16196 1
16300 1
16487 1
16611 1
17026 1
17234 1
17322 1
17665 1
17805 1
17846 1
17867 1
17955 1
18137 1
18272 1
18314 1
18687 1
18708 1
18791 1
18807 1
18978 1
18999 1
19217 1
19549 1
19663 1
19689 1
19834 1
19933 1
20037 1
20244 1
20296 1
20452 1
20530 1
20556 1
20576 1
20763 1
21075 1
21241 1
21282 1
21308 1
21433 1
21697 1
21755 1
21843 1
22165 1
22216 1
22268 1
22403 1
22528 1
22554 1
22632 1
22818 1
22995 1
23005 1
23047 1
23203 1
23358 1
23566 1
23602 1
23799 1
23846 1
23872 1
23877 1
24292 1
24334 1
24396 1
24884 1
25061 1
25112 1
25123 1
25434 1
25450 1
25844 1
25849 1
25901 1
25912 1
25953 1
26057 1
26213 1
26233 1
26649 1
26784 1
26887 1
27043 1
27157 1
27199 1
27406 1
27718 1
27785 1
27822 1
28055 1
28860 1
28963 1
29166 1
29327 1
29379 1
29482 1
29586 1
30209 1
30479 1
30624 1
30946 1
31117 1
31133 1
31143 1
31558 1
31989 1
32908 1
33115 1
33136 1
33219 1
33224 1
33313 1
33717 1
33842 1
33946 1
34050 1
34133 1
34548 1
34569 1
34776 1
35150 1
35191 1
36333 1
36437 1
36748 1
37475 1
37766 1
37994 1
38202 1
38617 1
38700 1
38762 1
39032 1
39136 1
39551 1
39748 1
39862 1
40070 1
40278 1
40485 1
40797 1
40900 1
41108 1
41212 1
41627 1
41731 1
41835 1
42250 1
42769 1
42873 1
43080 1
43288 1
43495 1
43703 1
43807 1
6230 1
7321 1
7710 1

tanecho
12-29-2013, 11:04 AM
Added fero to the above test (no buffs but haste and intensity), to see if it was the cause for the difference between what was previously observed as a cap around 20k, it doesn't appear to be related.

All together, I'm just coming up with more questions than answers about the point at which it stops counting crits. What does a minimum damage modifier do exactly? Are crits being cut at some specific numeric value, or is it being cut at some specific multiplier value?

Side note: I would find it very odd if they coded some cap in for melee attacks, but did not do so for spell attacks, when those have traditionally been for much higher numbers. Or maybe they did, and wizards haven't hit it yet at half a million. :P

Next test will keep same parameters, and add in a crit glyph.


Row Labels Count of Crits
Grand Total 250
19734 79
12288 13
30677 13
13558 10
34250 8
12676 6
15323 6
17087 5
16205 4
18852 4
32442 4
14440 3
17969 3
22206 3
25206 3
35265 3
14027 2
14707 2
26265 2
26794 2
28382 2
29618 2
30323 2
32970 2
33501 2
33676 2
35088 2
35794 2
12766 1
13252 1
14240 1
14619 1
15272 1
15360 1
16596 1
18148 1
19472 1
19913 1
19940 1
20831 1
21324 1
21368 1
21500 1
21536 1
21766 1
21934 1
22277 1
22559 1
22736 1
22894 1
22966 1
23265 1
23406 1
23583 1
24501 1
25030 1
25189 1
25383 1
25868 1
26044 1
27068 1
27104 1
27668 1
27906 1
28030 1
28127 1
28207 1
29750 1
29944 1
29971 1
30527 1
31073 1
31471 1
31559 1
31911 1
32273 1
32467 1
33747 1
33853 1
34329 1
34383 1
34558 1
34699 1
34735 1
35617 1
35706 1
35971 1


Row Labels Count of Noncrits
Grand Total 3766
44120 1627
12460 324
39916 75
29407 70
35712 68
42018 65
27305 61
31509 61
23101 59
37814 57
25203 56
33610 56
8080 42
20999 37
10903 20
18897 19
9346 19
10384 17
8308 17
8827 13
11941 12
9865 11
11422 10
29381 7
16795 6
23568 6
26370 6
34259 6
37373 6
40799 6
43705 6
12565 5
12773 5
13707 5
14226 5
17755 5
19727 5
20039 5
22218 5
22737 5
26786 5
28343 5
31249 5
31976 5
32598 5
33429 5
33844 5
34571 5
35609 5
38619 5
41006 5
41629 5
42563 5
43497 5
44016 5
11610 4
12876 4
13499 4
15056 4
15575 4
15783 4
16094 4
17444 4
19416 4
20558 4
20973 4
21077 4
21492 4
21596 4
21699 4
22115 4
22426 4
23049 4
23672 4
25229 4
26993 4
27616 4
28031 4
28135 4
28758 4
29173 4
29796 4
30938 4
31353 4
31768 4
34986 4
35193 4
36128 4
36543 4
37892 4
38307 4
38515 4
39138 4
39864 4
40695 4
43290 4
12461 3
12669 3
14018 3
14745 3
15264 3
15368 3
15471 3
16198 3
16302 3
16821 3
17340 3
18274 3
18482 3
18585 3
18793 3
20454 3
20765 3
21388 3
21907 3
23256 3
23775 3
24813 3
25125 3
25540 3
25644 3
25748 3
25851 3
26267 3
27408 3
27824 3
27927 3
28550 3
28965 3
29692 3
29900 3
30522 3
30834 3
31664 3
32910 3
33014 3
33117 3
33221 3
33533 3
33636 3
34052 3
34778 3
35401 3
35816 3
36024 3
36750 3
36854 3
36958 3
37477 3
37581 3
37788 3
37996 3
38930 3
39345 3
39761 3
40072 3
40280 3
40591 3
40902 3
41110 3
41421 3
42875 3
42978 3
43809 3
12171 2
12472 2
13084 2
13188 2
13292 2
13375 2
14330 2
14433 2
14537 2
14849 2
14952 2
15160 2
15679 2
15990 2
16406 2
16613 2
16655 2
16717 2
16925 2
17132 2
17963 2
18056 2
18066 2
18378 2
18689 2
19104 2
20246 2
20610 2
21180 2
22198 2
22322 2
22530 2
22634 2
22862 2
23464 2
23879 2
24087 2
24191 2
24294 2
24403 2
24502 2
24917 2
25903 2
26080 2
26682 2
26889 2
27097 2
27512 2
28509 2
28784 2
30003 2
30107 2
30419 2
30626 2
30668 2
31041 2
31145 2
31436 2
31560 2
32287 2
32391 2
32702 2
33740 2
34155 2
34467 2
34674 2
35297 2
35920 2
36231 2
36439 2
37062 2
38100 2
38723 2
39449 2
39657 2
41733 2
41940 2
42044 2
42148 2
42252 2
42667 2
42771 2
43186 2
43601 2
43913 2
8875 2
10115 1
10230 1
10364 1
10380 1
10468 1
10489 1
10515 1
10551 1
10723 1
10816 1
10842 1
10904 1
11216 1
11273 1
11382 1
11423 1
11735 1
11942 1
12108 1
12435 1
12586 1
12918 1
12980 1
13032 1
13058 1
13126 1
13198 1
13328 1
13344 1
13395 1
13562 1
13816 1
13899 1
13956 1
14122 1
14241 1
14361 1
14413 1
14724 1
14786 1
15103 1
15170 1
15383 1
15508 1
15627 1
15669 1
15705 1
15835 1
16157 1
16317 1
16364 1
16478 1
16509 1
16816 1
17028 1
17122 1
17210 1
17267 1
17366 1
17464 1
17547 1
17610 1
17651 1
17729 1
17802 1
17900 1
18051 1
18108 1
18170 1
18253 1
18284 1
18336 1
18352 1
18430 1
18617 1
18772 1
18814 1
19001 1
19037 1
19286 1
19312 1
19354 1
19364 1
19447 1
19457 1
19691 1
19738 1
19758 1
19779 1
19831 1
19992 1
20028 1
20142 1
20298 1
20324 1
20350 1
20661 1
20869 1
20905 1
20921 1
21284 1
21772 1
21793 1
21803 1
21845 1
21980 1
22099 1
22177 1
22208 1
22234 1
22286 1
22385 1
22400 1
22639 1
22841 1
22904 1
22945 1
23007 1
23381 1
23402 1
23661 1
23718 1
23848 1
23921 1
23957 1
23962 1
23983 1
24035 1
24118 1
24170 1
24243 1
24253 1
24268 1
24279 1
24398 1
24606 1
24632 1
24710 1
24860 1
24938 1
24954 1
25021 1
25063 1
25197 1
25286 1
25332 1
25436 1
25592 1
25649 1
25670 1
25831 1
25872 1
25955 1
26059 1
26163 1
26241 1
26293 1
26474 1
26537 1
26858 1
27139 1
27248 1
27398 1
27580 1
27720 1
27787 1
27990 1
28166 1
28239 1
28446 1
28457 1
28529 1
28602 1
28862 1
29038 1
29069 1
29152 1
29168 1
29204 1
29256 1
29272 1
29277 1
29339 1
29344 1
29396 1
29588 1
29734 1
30211 1
30315 1
30481 1
30496 1
30730 1
30808 1
30813 1
30886 1
31021 1
31228 1
31457 1
31742 1
31872 1
31960 1
32079 1
32349 1
32365 1
32495 1
32521 1
32723 1
32816 1
33299 1
33325 1
33668 1
33735 1
34031 1
34124 1
34135 1
34218 1
34301 1
34311 1
34363 1
34882 1
35214 1
35339 1
35505 1
35619 1
36003 1
36335 1
36647 1
37166 1
37186 1
37207 1
37269 1
37685 1
37877 1
38204 1
38411 1
38702 1
38795 1
38826 1
39034 1
39076 1
39242 1
39262 1
39553 1
39823 1
39968 1
40383 1
41214 1
41318 1
41525 1
41837 1
43082 1
43394 1
8154 1
8309 1
8704 1
8932 1
8989 1
9275 1
9638 1
9804 1
9929 1

brogett
12-29-2013, 12:52 PM
That last set seems to match DB=2080, Dam multiplier 1.73 (where? 55 from resolute, what is the remainder?), damage value 3000 as expected, and damage table as 4.05.

12460 = 2080 + 1.73*2*3000
44120 = 2080 + 1.73*(2*3000*4.05+1)
42018 = 2080 + 1.73*(1.9*3000+4.05+1)
etc

Where is the 73% extra damage coming from? Do you have AAs that multiplicatively modify your hits somehow? Maybe kick only? Eg like Scout's Mastery of Piercing is +0.02 multiplier per AA rank, giving +0.18 at rank 9.

As for crits, the commonest value you see is your 2x weapon damage spike:

19734 = 2080 + 1.7*(5 + 1.73*2*3000)

So 70% crit mod for kicks as I'd expect - no specials are modified by Veteran's Wrath.
If my theory of no crit mod for any base value above 20k (the bit in the brackets after the 1.7) then the max we'd see would be 2080 + 1.7*(20000), ie 36080.

In reality we see 35971, so close enough. That comes from damage table roll 1.92 with the common DI roll of 20: 35971 = 2080 + floor(1.7*(5 + floor(1.73*floor(2*3000*1.92+1)))). Note that a damage table roll of 1.93 instead would have put the inner portion as 20036 and so outside the hypothesised crit cutoff value. That doesn't mean it's not possible to get higher. You could have a DI roll < 20 combined with another damage table roll, but then you're looking for two rare rolls rather than just 1 so it's less likely to turn up.

All things considered, every parse I look at is saying the same thing - 20k cap in an earlier part of the formula (which is why it looks different post-crit sometimes).

Ishtass
12-29-2013, 01:03 PM
Has Dev responded to these findings yet?

brogett
12-29-2013, 04:04 PM
No, but they're probably on holiday atm. I'll give them a prod in the new year.

tanecho
12-29-2013, 07:54 PM
Removed all AAs, no buffs except previously mentioned ones to make results come faster.

Repurchased wu and pressure points for hits and accuracy.

My gear gives +190 damage from cleave, +76 FK damage, and +34 hStr bonus, which should put my innate bonus to kicks at an even 300.

Normal
2x spike - 426
Max spike - 817

Crits
2x spike - 524
Max spike - 1188

Repurchased Veteran's wrath, and no change from above values, so it does not affect our kicks.

Repurchased Kick Mastery, increase of 405 to all above figures, so it only touches the bonus.

Normal
2x spike - 831
Max spike - 1222

Crits
2x spike - 929
Max spike -1593

Repurchased Stonefoot, increase of 950 to all above figures, again only to bonus.

Normal
2x spike - 1781
Max spike - 2172

Crits
2x spike - 1879
Max spike - 2543



Repurchased Heel of Brithrax.. And these numbers just look odd now. Maybe this is the multiplier you were looking for? It has 9 ranks, so 9 * 0.02 would be the .18 you were looking for earlier, but I think it was .14 in one of the earlier formulas.

Normal
2x spike - 2229, increase of 448
Max spike - 2690, increase of 518

Crits
2x spike - 2343, increase of 464
Max spike - 3126, increase of 583

The only other possible factor at play (didn't have time to test before raids) was boot AC. I didn't vary it during any tests, but monk folklore is that it plays a part in FK damage. I've just never seen any hard and fast evidence for how much and in what manner.

tanecho
12-29-2013, 08:03 PM
If that theory holds, it looks like that's going to make Brithrax the most important +kick damage AA. Wu/thunderfoot are still better, but of the ones that just boost kick damage, it will give the best synergy with things like thunderfoot, ironfist, and any of our other large multitude of damage increasing discs.

brogett
12-29-2013, 08:38 PM
Good data. It looks like Hell of Bithrax works in the same way that the other weapon mastery AAs do that rangers (I think) and rogues get. However it is also adding a flat damage amount too.

The .14 I saw earlier was a confusion over the 280% damage mod which was only working when I factored in 294% instead. Are you sure it's 280% and not 276% say? Or maybe the .14 there was from something totally different.

Regarding the numbers, it fits better kick damage 64 with bonus 298, not 300. So with Kick Mastery + Stonefoot it's

2x = 298 + 405 + 950 + 2*64 = 1781
Max = 298 + 405 + 950 + 2*64*4.05+1 = 2172
Max crit = 298 + 405 + 1.7*(5 + 2*64*4.05+1) = 2543

Heel of Bithrax looks to be adding 425 as well as the 18% bonus.

2x = 298 + 405 + 950 + 425 + 1.18*(2*64) = 2229
Max = 298 + 405 + 950 + 425 + 1.18*(2*64*4.05+1) = 2690
Max crit = 298 + 405 + 950 + 425 + 1.7*(5 + 1.18*(2*64*4.05+1)) = 3126

425 doesn't divide evenly into 9, so likely the ranks aren't the same, at least between expansions.

tanecho
12-29-2013, 08:55 PM
That's my bad, I read 260 as 280 when I viewed on Lucy. Their 6s are especially curvy :P.

http://raidloot.com/Spells.aspx?name=ironfist+discipline+rk.+iii&class=Monk

That's strange though on the 298 vs 300 piece, because that 300 is fully explainable, and it's likely that there is a slight bit more from boot AC.

Mris
12-29-2013, 11:05 PM
To the best of my knowledge, boot AC is increments of 25 for FK and 30 for regular kick. not CERTAIN on the bonus, but I have it in my head that it's 10 points per increment.

Nedrom
12-30-2013, 01:49 AM
Can you please explain to me how heel of brithrax is any different than stonefoot or kick mastery?

brogett
12-30-2013, 06:08 AM
Can you please explain to me how heel of brithrax is any different than stonefoot or kick mastery?

Look at the values Tanecho posted:

Base non-crit
2x spike - 426
Max spike - 817

+Kick Mastery
2x spike - 831
Max spike - 1222

+Stonefoot
2x spike - 1781
Max spike - 2172

+Heel of Brithrax
2x spike - 2229
Max spike - 2690

Now look at the difference between the max spike and the 2x spike:
Normal: 391
+Kick Mastery: 391
+Stonefoot: 391
+Heel of Brithrax: 461

So it's added more to the top end than the bottom end. Why? The only way that can be true is if it's both a flat damage boost and also a damage multiplier. The numbers I gave fitted the equations; +425 and *1.18

tanecho
12-30-2013, 07:23 AM
That multiplier will make a bigger difference on our special FKs like synergy, stunning kick, and thunderfoot.

sojuu
01-09-2014, 04:08 PM
Just wanted to let yall know thanx to Brogett this has been looked into by devs. And as stated on the soe forums for those that caught the post cant really give details on whats coming with it because of NDA agreements. Just didn't want yall to think it has gotten forgotten about.

Obiziana
01-09-2014, 07:11 PM
Great news! Thank you to everyone that participated in bringing this to the attention of the dev team.

Chumpp01
01-09-2014, 08:33 PM
Just wanted to let yall know thanx to Brogett this has been looked into by devs. And as stated on the soe forums for those that caught the post cant really give details on whats coming with it because of NDA agreements. Just didn't want yall to think it has gotten forgotten about.


didn't they say there was no NDA with this beta ?

Nedrom
01-09-2014, 10:48 PM
Just wanted to let yall know thanx to Brogett this has been looked into by devs. And as stated on the soe forums for those that caught the post cant really give details on whats coming with it because of NDA agreements. Just didn't want yall to think it has gotten forgotten about.

/nods

Ishtass
01-10-2014, 12:24 AM
didn't they say there was no NDA with this beta ?

The fix is not related to beta, but it appears that's where it's going into it. They wanted to get it ironed out before any promises were made, but now it's public.

Kaliaila
01-10-2014, 09:48 AM
I think that Sojuu is referring to the CRT NDA, not Beta.

Ishtass
01-11-2014, 12:41 AM
Supposedly this change is making rogues insane dps (will likely come with another retune imo, a friend parsed pretty hard in beta).

Anyone compare the change in monk dps?

Nedrom
01-11-2014, 04:56 AM
Supposedly this change is making rogues insane dps (will likely come with another retune imo, a friend parsed pretty hard in beta).

Anyone compare the change in monk dps?

Rogues should be doing insane dps.

Gorkeyah
01-11-2014, 10:50 AM
I've been wondering what Nerf is going to go with this fix you all wanted. :)

KIILLZ
01-11-2014, 12:12 PM
I've been wondering what Nerf is going to go with this fix you all wanted. :)

needs a like button.

brogett
01-11-2014, 01:09 PM
Supposedly this change is making rogues insane dps (will likely come with another retune imo, a friend parsed pretty hard in beta).

Anyone compare the change in monk dps?

I'd need to see the parses, as they don't match my math.

On our best burn combo we may have 70% coming from backstab (that's the 30sec Frenzied Stabbing disc), but for all the rest it's half that. Of that backstab damage, most of it is normal backstab skill. Maybe a third comes from Massive Strike procs (what your Thunderfoot was based on) and Incursion skill (20sec reuse ability), with the rest being normal backstabs. The normal stabs couldn't trigger this bug so don't gain anything from the fix. Normal non-crits couldn't gain anything either. So we're looking at gaining damage only for the cases that would have critted and didn't, which is only a few percent of our hits.

So we gain an extra 1.3x more damage on that few percent. Hardly going to be insane dps. However that's math and theory. I haven't raided on beta and haven't seen any parses yet. Without full support and discing on an arena dummy I managed 45k dps from backstab damage. Maybe double it with support, and add in autoattack and other skills.. sounds about where I'd expect tbh. My gut feeling here is monks will gain just as much as rogues, and it'll be around 5% boost.

Ishtass
01-11-2014, 02:15 PM
Rogues should be doing insane dps.

Did I say they shouldn't?

@Brogett - I don't have a full parse, but it looked pretty good while in burn mode. I don't want to share a parse I didn't make, just curious to see what others are experiencing. But if that parse was the norm and not a fluke, it's pretty impressive bump (which is why I would say rogues might have to be tuned a bit to compensate, but that's my own speculation).

I haven't heard anything about monks yet, so who knows how good it'll be for us. Before we get people crying that we might get nerfed, keep in mind that if anything, we're likely to come out better than we were before the change still. Also, this problem would just keep getting worse and worse as our levels and damage scale. You WANT this fixed now, trust me. This could eventually have been worse than any nerf.

Mris
01-11-2014, 08:29 PM
Also, this problem would just keep getting worse and worse as our levels and damage scale. You WANT this fixed now, trust me. This could eventually have been worse than any nerf.

Imagine the worst "Fix a 5 year old bug" scenario. That's where this could have headed. I agree with you, Ishtass. Better working now and tweaked a little now than working 2 years from now and tweaked a lot.

Nedrom
01-13-2014, 07:44 AM
Why defensive just making a statement :)

Ishtass
01-13-2014, 08:33 PM
Why defensive just making a statement :)

After all these years, still no ability to turn off riposte without switching weapons ;)

sojuu
01-14-2014, 03:38 PM
didn't they say there was no NDA with this beta ?



I think that Sojuu is referring to the CRT NDA, not Beta.

yea that was the NDA I was referring to

Obiziana
01-14-2014, 03:44 PM
All that said, anyone played with this in beta? See any noticeable gains in DPS?

Maesin
01-23-2014, 07:56 AM
Got tells tonight during the raids from angry wizards. "How the F did you hit for 665k?!"

Crane crit is working now, lol.

rapitiss
01-23-2014, 11:41 PM
Tonight Gnome Rogue pulls ahead of iksar monk in BG1 burn.

Going to have trip him next time we rush into combat!

( FYI - we're usually tied and he had me by a good 4-5k DPS )

Ishtass
01-23-2014, 11:46 PM
Tonight Gnome Rogue pulls ahead of iksar monk in BG1 burn.

Going to have trip him next time we rush into combat!

( FYI - we're usually tied and he had me by a good 4-5k DPS )

we'll never hear the end of it...

sensei savager
01-24-2014, 12:50 AM
[Thu Jan 23 20:42:42 2014] You kick Luclin for 796129 points of damage.

Ishtass
01-24-2014, 10:34 AM
[Thu Jan 23 20:42:42 2014] You kick Luclin for 796129 points of damage.

Damn, crane or thunder?

sensei savager
01-24-2014, 05:41 PM
That was crane with ironfist and shm epic running.

KIILLZ
01-24-2014, 06:06 PM
thats a crane crit I know couple of us did it on a burn last nite.....rogues shall not pass!!!

Kaliaila
01-25-2014, 09:55 PM
Looks like a great update so far; I will have to make a trip to Plane of Sky it sounds like.

Archus
02-05-2014, 12:23 PM
Wanted to thank Brogett and tanechco for doing the parses and analytical work that got this looked into and fixed so quickly. kudos!

Nedrom
03-10-2014, 11:13 AM
Wow, no one interested in testing this?

I'm happy that members of our community pushed through and tested this and provided feedback to the devs to get a fix in place.

Kudos to everyone involved for another monk bug being squished :) Even it it wasn't obvious we had it.

Thanks to the rogues who brought this issue to light.