The .NET documentation leads me to believe that the same numbers will be generated across different computers, from "The random number generation starts from a seed value. If the same seed is used repeatedly, the same series of numbers is generated." I don't have anyway to test that currently though.
As long as the seed is perfectly static (i.e. you explicitly feed it a specific # every time) and not somehow picked when Rawr opens, you're golden. It sounds like this is the case. The only case it would be a problem is if you let the system randomly pick once (i.e. on opening the program) then repeatedly use that number.
The 1 haste issue can occur even without 2piece T10. There are plateaus where 1 haste is the decider for one more autoattack in a short timeframe (average fight length), thus causing a DPS jump. If you want to increase accuracy I'd agree with Neraya regarding multiple runs (with multiple static numbers) - 3 to 5 of your normal length, rather than one really long run.
Originally Posted by Neraya
Ideally... I would expect you to calculate a random seed when Rawr starts, then use that same seed as long as rawr is running. Quit rawr, relaunch and you get a different seed. Something like this is how a lot of statistical/military/economic simulations are run. ... It also leads to different people using rawr getting different results, not necessarily a problem unless different people expect to have comparable data.
Bold is mine - I think this is precisely the problem (BIS list uses Rawr, not the spreadsheets) and why he should stick with static (or allow it to be manually set). My spreadsheet uses a static number (3.3 at default, for the patch) which can be altered by the user if desired.
Regarding reaction time - I've argued to call Latency by the name "Delay" to subsume both. I gathered the same ~200 ms data from various sources. Testing in live by several people, however, seems to show reaction time is often concurrent with latency, not additive. I.e. if you've got 100ms latency and a 200ms reaction time, you react in about 200ms total - effectively eliminating your latency issue. My default value of delay is 200ms. I seem to recall Rawr's default latency is 0.1 or 0.2 - if it's set to 0.2 then it's going to be pretty accurate for most (at least high-end) players.
Edit:
Originally Posted by Trakor
Assuming lag + reacting time of 0.5secs, still gives you 1sec to change your mind and cancel whatever ability you were going to do and use DS instead.
Not if you use ability A and 1.3 seconds later your autoattack resets DS - you now have 0.2 seconds to prepare. As has already been discussed, clcret will immediately detect and advise. Humans may still miss procs by a GCD in times like this - likely to already have a key half pressed in preparation.
Rock: "We're sub-standard DPS. Nerf Paper, Scissors are fine."
Paper: "OMG, WTF, Scissors!"
Scissors: "Rock is OP and Paper are QQers. We need PvP buffs."
Rawr 2.3.1 came out which has the 3.3 changes I made. Please make sure to report any bugs or issues you have with Ret here. Since I don't play WoW anymore I don't actively use Rawr anymore. So while I try to test everything I implement the best I can, there is a larger chance of bugs slipping by me.
I don't see the benefit of having a different static seed used on each startup of Rawr. With the length the simulator is run, the dps difference between using different seeds is much less then 0.1%. So it then prevents the ability to compare results from other people, just to gain a really small amount of accuracy. Which is not a tradeoff worth taking in my opinion.
I forgot I never supported trinkets like [Victor's Call], I'll add that. I'll update the default buff set used. I'll also look into having "delay" effect the usage of 2T10 autoattack procs.
It appears +4 Strength bonuses from Yellow sockets will be worthwhile to obtain using [Inscribed Ametrine]s instead of forgoing the bonus and using a [Bold Cardinal Ruby] once we drop 2pieceT9. I presume it's because the frequency which we deliver attacks that can proc RV is increased through the 2pieceT10 bonus as well as, slightly less so though, our 4pieceT10 bonus increases Judgement damage.
Would it be possible to implement the framework for Tiny Abomination in a Jar? We won't know the proc chance on melee hit until early January, but I would think that this trinket may have a shot at beating out an ilvl 245 Death's Choice.
Would it be possible to implement the framework for Tiny Abomination in a Jar? We won't know the proc chance on melee hit until early January, but I would think that this trinket may have a shot at beating out an ilvl 245 Death's Choice.
Should have support in for that for the build that is going out tonight. I am going to assume a 35% proc rate on MeleeHit with no ICD.
Would it be possible to implement the framework for Tiny Abomination in a Jar? We won't know the proc chance on melee hit until early January, but I would think that this trinket may have a shot at beating out an ilvl 245 Death's Choice.
I have it modelled in Rawr now. Assuming 35% chance on MeleeHit to gain a charge, no ICD, the proc causes SoV and making full use of the hit, it is around as much DPS as Greatness.
So Bryntroll is modeled off of a MeleeHit trigger, but it's been looking like a lot of people suspect it to also be procing on a SoV refresh due to how much dps the proc is providing
Is there any trigger that could account for that other than SoVTick? I'm sure tick is too high, what if I just double the proc rate since most of the time I'm using SoV anyway?
SoV (the refresh or the direct damage) isn't taken into account for being able to proc stuff. Does the refresh (or the direct damage) count as a melee hit for that stuff?
From PTR data it was suspected that the proc rate was 15% across the board for physical hits. I did a few longer tests to see if this was still the case.
Perhaps someone else can make more sense of the data than I could. The Judgement, CS, and DS tests were done such that I could not melee the target dummy.
SoV (the refresh or the direct damage) isn't taken into account for being able to proc stuff. Does the refresh (or the direct damage) count as a melee hit for that stuff?
The DoT application or refresh count as a hidden melee strike(I am not sure whether it can miss/be parried/dodged independently from the white hit that triggered it), basically making each white hit count as 2 as long as one uses SoV. The damage proc does not count as a melee strike.
I think I may have found a "bug" in the database regarding Engineering buffs on gear.
I currently use Flexweave Underlay and Nitro Boosts on my current set of gear, but when I try to select it as my back and feet enchant respectively, it is unavailable. When I use the Optimize tool, it comes up as not being found and chooses "unenchanted" as the best choice.
I tried looking for an option to add this in on my own, but couldn't locate the proper tool. Is anyone else having this problem?
It's not clear that refreshing the SoV stack effectively doubles the number of Drain Life proc chances. In my 60 minute test of auto attacks the proc rate was observed to be 10%. In my 60 minute test of auto attacks with SoV active the proc rate was observed to be substantially higher versus auto attacks without SoV active.
17.81% if melee + stack refreshes can proc drain life.
15.25% if melee + stack refreshes + drain life can proc drain life.
11.88% if melee + stack refreshes + seal proc can proc drain life.
10.68% if melee + stack refreshes + seal proc + drain life can proc drain life.
I don't know how to discern which framework is correct. You'd figure with a sample size of 2000-3500 proc opportunities that the difference is statistically significant. The raw numbers and WMO links are in my previous post. One would expect that if refreshing the SoV stack effectively doubles the number of Drain Life proc opportunities then the 17.81% number should be closer to 10%.
From PTR data it was suspected that the proc rate was 15% across the board for physical hits. I did a few longer tests to see if this was still the case.
Perhaps someone else can make more sense of the data than I could. The Judgement, CS, and DS tests were done such that I could not melee the target dummy.
I found no occurrence of Drain Life procing of Drain Life in your log of auto attacks without SoV, the smallest time frame of 2 procs was 2.8 seconds, which should roughly match your attack speed. However, there was a triple proc when using SoV. To me the data look like a roughly 11% proc chance, which procs of white hits, melee skills, the SoV refresh as well as proc. Also possibly one or more of the debuffs applied by judging, since your data for Judgment is decidedly odd.
Perhaps you could do another test where you use just Judgement, but without any talents. Also, which Judgement did you use?
Edit: Perhaps the RV DoT also is somehow capable of triggering the proc. Since there is no log data for CS, DS and Judgment, I cannot look whether crits can produce additional procs, but that would explain the extremely high proc rate for judgment as well as the still quite high proc rate for DS(unless you hit more than one target with DS, but I assume you were careful).
If RV is doing something weird, I would expect that the CS data would display the same phenomenon. However the proc rate of Drain Life on CS is nearly identical to that of auto attacks without a SoV active. Perhaps it's related to Heart of the Crusader? The other possibility is that because CS is 4 seconds, in the test you are less likely to run into a situation where you do not crit before Righteous Vengeance wears off.
I have the Judge WMO available but it was split into 4 sections for some reason. 1,2,3,4
I haven't been able to observe a double proc from melee hits without a seal active. I was able to observe several Judgement double procs. Example screenshot.
Judgement
9 seconds
Judgement (crit) with Drain Life proc
1 second
Additional Drain Life proc
2 seconds
RV ticks begin
So far it seems fairly reproducible. Judgement crits that apply Righteous Vengeance act as two opportunities to proc Drain Life.
Indeed, there are a few almost double procs in your judgment data. Notice however that all are caused by a judgment crit, and when you look at the time stamps, the first proc is simultaneous to the judgment crit(or hit, but then it is no double proc)
Originally Posted by Log
2009-12-11 19:21:21.433 Heroic Training Dummy's Heart of the Crusader ( Bluedeep's debuffer ) is removed.
2009-12-11 19:21:21.433 Bluedeep's Drain Life hit Heroic Training Dummy for 2380 damage.
2009-12-11 19:21:21.433 Bluedeep's Judgement of Corruption(R1) crits Heroic Training Dummy for 3622 damage.
2009-12-11 19:21:21.433 Heroic Training Dummy affected by Bluedeep's Heart of the Crusader.
The second is simultaneous to the application of righteous vengeance
Originally Posted by Log
2009-12-11 19:21:22.299 Heroic Training Dummy affected by Bluedeep's Righteous Vengeance.
2009-12-11 19:21:22.299 Bluedeep's Drain Life hit Heroic Training Dummy for 2435 damage.
2009-12-11 19:21:22.299 Bluedeep gains 1098 Mana.(Bluedeep's Sanctified Judgement)
I also found a case where there is no recorded application of RV, but in that case there was RV still on the dummy, so it just got refreshed and not applied, which is not recorded in the log.
To me this is a strong indication that the application or refresh of RV indeed can cause a proc.
Edit: Ah, you came to the same conclusion as I. Do you also have to logs of your CS and DS runs? It should be quite easy to look for double procs of this type after crits, and if it is the case, that would be proof. With Judgments there is so much going on that there still is some doubt, even though I am now convinced that the application of RV can trigger the proc.
15.25% if melee + stack refreshes + drain life can proc drain life.
That one. I believe your double Drain Life from a crit Judgement is just Drain Life procing Drain Life.
I don't have one on my Pally, but managed to pick one up on my DK in a PuG. Unholy does have a "hidden" melee strike (Blood Caked Blade) that similar to SoV refresh can proc Drain Life.
Millions of words are written annually purporting to tell how to beat the races, whereas the best possible advice on the subject is found in the three monosyllables: 'Do not try.'
That one. I believe your double Drain Life from a crit Judgement is just Drain Life procing Drain Life.
No, I do not think so. When you check Gluttons Logs(the one of just auto attacking without any Seal active), you will see that out of more than 1000 white hits there are no double procs. If it were indeed able to proc of itself, then there should have been on average 1078*0.15*0.15 = 24.3 double procs(assuming the 15% proc chance). Since we observe 0, we can safely discard the idea that it can proc of itself.
To summarize what information we get through analysis of the logs:
:Log of just auto attacks, no seal active:
--it cannot proc of itself
--proc chance is about 10%-12%, no internal cooldown
:Log of auto attacks, SoV active:
--it can proc of SoV DoT refresh (we expected this, since it works as a hidden melee strike)
--it can proc of SoV damage proc (there is one instance of a trippe proc in the log, and since it cannot proc of itself, we must find three seperate causes of the proc. 1. auto attack, 2. SoV refresh, 3. SoV damage proc. Also, if we assume that with SoV each auto attack counts as 3, we observe a proc chance of 11.8%, which matches the pure auto attacks)
:Logs of Judgment attacks plus Log of Crusader Strikes:
--it can proc of RV application or refresh, meaning CS, DS and Judgment crits have an additional chance to proc.
--judgements have 2 chances to proc, 3 if it crits (this is a bit uncertain, since there is no definite proof in the logs. however, only this way we can fit the number of procs to the number of judgments with a proc chance of 10%-12%)
If you treat CS hits and CS crits each as one chance to proc, then the overall proc rate is 14.36%.
If you treat CS hits as one chance to proc and CS crits as two chances to proc, then the overall proc rate is 10.57%
I observed one double proc on the CS test, but it's fairly clear that RV is giving a second chance to proc roughly one second after a CS crit.
However if you apply the same method to the Judgement data, you still have an inflated chance to proc. 27% using the first method, 17% using the second method. In fact you have to say that Judgement crits have 4x opportunities to proc Drain Life than Judgement non crits in order to approach a 10% overall proc rate. Alternatively, you get 10% if you say non crits have two opportunities to proc Drain Life and crits have three opportunities to proc drain life (non crit: X + Judgement; crit: X + Judgement + Righteous Vengeance). Now what is X? Judgement of Wisdom/Light/Justice? Heart of the Crusader? Vindication? I'm not sure.
Yes, your Crusader Strike data seals the deal, a crit is worth 2 chances to proc due to the RV application/refresh.
As to Judgments, I suppose a Judgment could count as 2 melee strikes, it certainly applies debuffs(Heart of the Crusader, Judgment debuff), and when RV counts, I see no reason why they might not. However, Vindication did not increase the proc chance, so it is not generally debuffs. I really wonder if this is intended behaviour, since at the moment it seems to heavily favor paladins. Or DKs/Warriors also have a bunch of additional proc chances, who knows.
I have edited my post above to add the confirmation, and add the hypothesis of judgments as 2 strikes.
Fury warriors have a high number of physical actions per unit time. At the moment I'd say the weapon may disfavor Unholy DKs, although they may make up for it somewhat by having superior scaling talents for the shadow damage portion such as 15% Blood Presence, 10% Rage of Rivendare, 10% Bone Shield, and 5% Desolation while we only have 3/6% from Crusade and 1% from Sense Undead.
I agree that these "hidden" melee attacks don't seem like they're intended.
Fury warriors have a high number of physical actions per unit time. At the moment I'd say the weapon may disfavor Unholy DKs, although they may make up for it somewhat by having superior scaling talents for the shadow damage portion such as 15% Blood Presence, 10% Rage of Rivendare, 10% Bone Shield, and 5% Desolation while we only have 3/6% from Crusade and 1% from Sense Undead.
Unholy DKs get around 6% of their damage from the proc and have the best scaling on all damage. While Warriors have many melee attacks, they don't have the all damage auras that Ret/DK has (Arms/Fury damage auras only affect physical damage) so they get the lowest benefit, around 4% of their damage. Ret gets about 7% of their damage from it.
Either way, the weapon is BiS (ignoring Shadow's Edge) for Unholy and Ret.
Millions of words are written annually purporting to tell how to beat the races, whereas the best possible advice on the subject is found in the three monosyllables: 'Do not try.'
The "definitive" Bryntroll's Drain Life proc opportunities per ability list:
Melee hit or crit: 2 (Melee damage compnoent, SoV stack refresh)
SoV seal proc: 1
Judgement hit: 2 (Judgement damage component and Judgement debuff application/refresh)
Judgement crit: 3 (Judgment damage component, Judgement debuff application/refresh, Righteous Vengeance application/refresh)
CS and DS hit: 1
CS and DS crit: 2 (CS and DS damage component, Righteous Vengeance application/refresh)
DS secondary targets are capable of procing Drain Life.
Seal of Command cleave secondary targets are capable of procing Drain Life.
Vindication and Heart of the Crusader do not appear to act as hidden melee attacks capable of procing Drain Life. The overall proc rate appears to be 10% when all hidden attacks are taken into account.
I wonder if this may increase the value of crit and AGI gear a bit despite the loss of 2P T9.
12/11/09, 9:12 AM
Similar Threads
