I know this is a grey area, but a macro to spam the number one key would eliminate the possibility of human error or lapse in concentration. Just saying

A macro could make the hits more consistent but that's not what he was doing. But even still it doesn't explain for example how I could do two hits with a GCD of 1.109 seconds when I should only be able to do around 1.3 at best.

So the elements that make his experiment pointless are:

1) Human factor in pressing the key

2) Input delays for whatever technical reason

3) Delays or descrepancies between SWTOR and Starparse

Yes, you can reduce the human factor by using a macro but as I said if I really focus I can get very consistent times most of the time without needing 500 points more as he suggested. His biggest fault on the calculation side is to just take an average of 40 hits which will be inconsistent and hide what is going on.

The problem is not that the game is calculating things wrong, he is because he makes assumptions about the 3 points I listed above. I do not assume that any of the three are going to be perfect because the human and his keyboard, the PC, the server and Starparse are all connected via the internet and can suffer delays. Do mind that we're talking about tenths of seconds and even milliseconds. Those timings we can see in the numbers but we can't really deal with ourselves. I mean you or I won't know the difference in timing ourselves when they are that small.

But by using his method I could easily see that how focused I am and whether I used keyboard or mouse had a major effect on those timings. Try pushing that button consistently to get 40 consecutive hits, your finger won't like it and there will be speed differences. Using a macro will improve the precision but it still doesn't account for things such as server lag. For him to call proper math "magic numbers" is therefore unwarranted as his method only proves why the math is correct, but you do have to pay attention to what's going on.

Quote: "For him to call proper math "magic numbers" is therefore unwarranted as his method only proves why the math is correct, but you do have to pay attention to what's going on."

Again, knock it off with the damn strawman.

I took Trixie's suggestion of trying a macro setup.

Installed AutoHotKey

Wrote a script such that when a toggle key was pressed, the "1" key would be spammed.

I tried:

4 per second

5 per second

6 per second

10 per second

20 per second

100 per second <-- was bizzare, made the cooldown effect flicker...

Oddly enough, 20 per second and 100 per second actually caused all my 1.3 GCD toons to drop to 1.32-.133 raw average for a 1.4 GCD. The global cooldown sliding window effect on the ability bar would reset midway 3-4 times during a test.

So I kept it to 6 per second.

I would experience 1-2 resets during a test session at 6 per second. My toons would have a few test results where the average gave me a 1.3 GCD...and a couple where the average would give me a 1.4 GCD.

I tried 4 per second...but starting getting worse results...too slow I guess.

5 per second seemed to get me the best results...a few 1.3 GCD test results, and a 1.4 GCD test result roughly the same as when I use my finger.

So I guess nothing has really changed...

One thing I did notice is that each time I launched my AutoHotKey scripts, it caused the SWTOR UI to reset.

=8-|

*** Another Update ***

http://www.mrrabbit.net/misc/swtor/swtorglobalcooldown.php

I did several things in this update:

1. Lowered the Alacrity points for my Mercenary healer and Sorcerer healer to something in the ball park of the SWTOR 6.0 Theory Crafting Charts.

2. Leveled, geared and tested my Sniper - Marksmanship.

@3244 (Sorcerer - Corruption)

@3270 (Mercenary - Bodyguard)

Both will very rarely get from a client perspective a "clean" 1.3 GCD. They get mostly 1.31 to 1.32 results in a full test - and intermittently will experience slow downs that drop to to 1.37 / s RAW which are of course treated at 1.4 /s Actual.

My Marksmanship Sniper @1665 just broke into the 5.5 /s Energy Regen Rate territory and walks a very fine line from a local client perspective at 1.4032 /s Raw GCD for a 1.4 Actual GCD.

My Sniper has always lagged behind my Arsenal Merc . . . so I'm heading downstairs to look at my tactical and gear choices for him. He's my main...

=8-)

I took Trixie's suggestion of trying a macro setup.

 

Installed AutoHotKey

 

Wrote a script such that when a toggle key was pressed, the "1" key would be spammed.

 

I tried:

 

4 per second

5 per second

6 per second

10 per second

20 per second

100 per second <-- was bizzare, made the cooldown effect flicker...

 

Oddly enough, 20 per second and 100 per second actually caused all my 1.3 GCD toons to drop to 1.32-.133 raw average for a 1.4 GCD. The global cooldown sliding window effect on the ability bar would reset midway 3-4 times during a test.

 

So I kept it to 6 per second.

 

I would experience 1-2 resets during a test session at 6 per second. My toons would have a few test results where the average gave me a 1.3 GCD...and a couple where the average would give me a 1.4 GCD.

 

I tried 4 per second...but starting getting worse results...too slow I guess.

 

5 per second seemed to get me the best results...a few 1.3 GCD test results, and a 1.4 GCD test result roughly the same as when I use my finger.

 

So I guess nothing has really changed...

 

One thing I did notice is that each time I launched my AutoHotKey scripts, it caused the SWTOR UI to reset.

 

=8-|

When you press a key there is a message sent to the server that you pressed that key. However, with a GCD in place I wonder what happens to that message when you send multiple commands like that within the GCD window. What I mean is that I wonder if they can get in the way of each other and actually cause odd responses. A bit like when people hit a button multiple times before it has time to do what it needs to do.

So that could affect your results as it is.

Also there is a setting in game where you can push a key before the GCD finishes and if it's within that set time frame it will execute after the GCD finishes. This setting could also affect results.

Bottom line is still that you're refusing to acknowledge all kinds of variables that are clearly affecting your test. It's like you are trying to test something in an enclosed area but that area isn't actually enclosed because there's holes in various places and you try to act like those holes aren't there or aren't really holes.

The input of your keyboard and any lag in between and people hitting buttons multiple times and connection delays between your PC, the server and Starparse are all part of a range of things that affect the timing of things. And these fluctuations are themselves in flux and as such change constantly. That's why you can't get a clear consistent result.

Your experiment is trying to exclude variables in its approach and calculations but without actually excluding these variables. Keep going as you will but you will continue to find things that don't make a lot of sense as they haven't so far. In the end you're on the road to discover that there are a number of variables that affect your outcomes that you can't control. You already should be there in fact. This is the reason why you keep getting odd results. It's just odd that you haven't gotten to this conclusion yourself yet though. It makes some of us think that you just don't want to see it.

It seems rather clear that human input is flawed, our connections via the internet aren't as precise as you seem to assume and there are a number of things that can have different effects, like GCD times that shouldn't happen. Now, I do not know what it's going to take to break through, but if anything, getting GCD times that are better than they should be able to be, should tell you that there's more going on than you can control. You can blame it on the game coding but that's too easy and unproven since there are many other things that affect it.

You see, it could very well be that the game has some countermeasures against delays which could affect the timings you get through in your starparse. So when there's a delay it recognises it may adjust for it etc.

Again too many things could be the reason for these inconsistencies and you can't ignore them simply because they are unknown or unmeasurable by yourself.

Strawman.

You really should pay attention...

=8-|

Strawman.

 

You really should pay attention...

 

=8-|

If I got a cartel coin for every time you clutch that strawman I'd be rich in game.

There is no strawman going on here because your set-up is flawed as you refuse to acknowledge that there are variables in play that affect your outcomes. It's natural that you cannot do much with those variables because in order to take those variables out you either have to be able to entirely exclude them or include them fully into your set up.

You can do neither. That's not your fault but acting like they don't exist or don't matter for your results is your fault. The only reason why you would call that a strawman argument is because your cognitive dissonance doesn't allow you to do anything else.

What you do is widen the range of alacrity to the point where (almost) all these variables no longer affect the outcome you seek. That's fitting the evidence to the conclusion.

I'm giving valid criticism on your set-up. I tried your set up and saw that how I pushed the buttons made a difference. That told me right then and there that the measuring method is flawed. If your point is to prove that humans are imprecise then we could've told you that beforehand and saved you the trouble.

Again...strawman.

You keep attributing a lack of acknowledging of variables to me - which indicates that you aren't paying attention or are deliberating misrepresenting me.

Intended or not . . .

You're creating a position I have not taken and attacking it - that's a strawman.

And again, I'm not arguing anything...just posting my observations from my client perspective.

Stating observations from a test session is not arguing...just stating observations.

=8-|

Clearly, he isn't going to listen and change anything. Or back his claims with some real proof. I suggest we all just stop posting in here and let him have the time of his life. Let him post whatever he wants for whatever reason he does this whole thing. If some people are going to follow his stat distributions for some reason and do lower DPS, that's on them. Plenty of real guides out there that are backed by real math and real parses. Smart people will find those if they want.

As he states himself, he is just "observing" things, even if his methods are flawed. So, whatever, let him continue his observations. May be one day he will finally arrive to the real numbers.

I took Trixie's suggestion of trying a macro setup.

 

Installed AutoHotKey

 

Wrote a script such that when a toggle key was pressed, the "1" key would be spammed.

 

I tried:

 

4 per second

5 per second

6 per second

10 per second

20 per second

100 per second <-- was bizzare, made the cooldown effect flicker...

 

Oddly enough, 20 per second and 100 per second actually caused all my 1.3 GCD toons to drop to 1.32-.133 raw average for a 1.4 GCD. The global cooldown sliding window effect on the ability bar would reset midway 3-4 times during a test.

 

So I kept it to 6 per second.

 

I would experience 1-2 resets during a test session at 6 per second. My toons would have a few test results where the average gave me a 1.3 GCD...and a couple where the average would give me a 1.4 GCD.

 

I tried 4 per second...but starting getting worse results...too slow I guess.

 

5 per second seemed to get me the best results...a few 1.3 GCD test results, and a 1.4 GCD test result roughly the same as when I use my finger.

 

So I guess nothing has really changed...

 

One thing I did notice is that each time I launched my AutoHotKey scripts, it caused the SWTOR UI to reset.

 

=8-|

Wait, AutoHotKey? I thought macro's were against the ToS? Not only that, but you posted here that you installed and used one? lol Good luck.

Did you forget sarcasm tags?

Did you forget sarcasm tags?

 

Wasn't being sarcastic at all. I was being serious. Why would you post on a SWTOR Forum that you were using a macro in their game? That is against the ToS is it not?

Wait, AutoHotKey? I thought macro's were against the ToS? Not only that, but you posted here that you installed and used one? lol Good luck.

Idk what "A***H**K**" is but macro's are allowed if they just do 1 thing. One button equals one action. Dev post on it. Do a search you should be able to find it.

So if that is doing just one action, he's fine per Dev's comment.

Wasn't being sarcastic at all. I was being serious. Why would you post on a SWTOR Forum that you were using a macro in their game? That is against the ToS is it not?

It’s a gray area. Plus he isn’t using it in combat or doing content. He is testing on a dummy. I think intent plays a big role in how they look at it.

It’s a gray area. Plus he isn’t using it in combat or doing content. He is testing on a dummy. I think intent plays a big role in how they look at it.

Not at all. It's not a grey area. They have stated what they mean on it. If it's not abiding by their rules, he can get in serious trouble for it. No matter if he's just sitting on his ship feeding gifts to a comp or in combat.

Don't give false information that can warrant someone a ban, especially when going against a Dev's official comment.

Again...strawman.

 

You keep attributing a lack of acknowledging of variables to me - which indicates that you aren't paying attention or are deliberating misrepresenting me.

 

Intended or not . . .

 

You're creating a position I have not taken and attacking it - that's a strawman.

 

And again, I'm not arguing anything...just posting my observations from my client perspective.

 

Stating observations from a test session is not arguing...just stating observations.

 

=8-|

And to what end? That is the question I asked some time ago already. I have said that the problem with your "findings" is that they are entirely personal and since you do not and I daresay cannot calculate all the factors that contribute to your conclusions, these conclusions by themselves are useless.

Your observations are like looking through binoculars, meaning that there are many things you don't see outside your focus that DO affect what you are seeing.

What's happening here is that you are mispresenting your observations and that's why my comments are anything but a strawman. You also conveniently use the strawman defense as a way to not address my points thereby maintaining the illusion, at least for yourself, that your observations have any value or meaning.

The math behind what you call a "magic number" is sound. It also excludes variables because there are too many variables to take into account and because these variables are not the same for everybody and not the same everytime for anyone either.

I also commented that your calculation is flawed because it relies on some assumptions that you cannot actually make. When I see two moments where the basic attack strikes for example, you cannot assume that the time between them is the actual GCD if there are any number of delays that could affect it, even as miniscule as a micro second. In fact you equate the interval between strikes to the GCD. I calculate the interval between strikes as GCD + X where X represents the variables and could indeed be a negative number as well. This is why in my version of the calculation there are shorter intervals possible than the GCD should be able to be (for example 1.1 where it should not be faster than 1.3). This is the core mistake in your calculation: you equate the interval between strikes to the GCD and this is simply false. It has to include X to account for the variations in your measurements, because your measurements do not exclude these variables.

What you are displaying here is a personal experience that has little meaning for others and does not in fact debunk the calculation of GCD. You said it yourself here: it's from your perspective but it's imprecise on top of being eprsonal. Which proves my point but also begs a question...

What the hell is the point of your calculations if they are entirely personal and the variables that cause variations are not by themselves representative of anything but your experience at the moment of testing?

I also commented that your calculation is flawed because it relies on some assumptions that you cannot actually make. When I see two moments where the basic attack strikes for example, you cannot assume that the time between them is the actual GCD if there are any number of delays that could affect it, even as miniscule as a micro second. In fact you equate the interval between strikes to the GCD. I calculate the interval between strikes as GCD + X where X represents the variables and could indeed be a negative number as well. This is why in my version of the calculation there are shorter intervals possible than the GCD should be able to be (for example 1.1 where it should not be faster than 1.3). This is the core mistake in your calculation: you equate the interval between strikes to the GCD and this is simply false. It has to include X to account for the variations in your measurements, because your measurements do not exclude these variables.

I have resisted coming back into this thread but Tsillah's post made we want to provide some feedback. Tsillah is of course correct. As I have tried to indicate in multiple responses to the OP involving multiple recommendations in multiple threads, the correct variable to measure isn't GCD per se... the correct variable to measure is Abilities Per Minute. They are, of course, intricately related, as a shorter GCD means more potential APM. However, APM is measured over a longer period, and can be statistically analyzed to determine if there are differences between different alacrity levels. It was this testing that dipstik did during PTS which confirmed a) that GCD rounding is still occurring at tenths of a second and not hundredths, and b) that the different GCD thresholds do result in a statistically different number of Abilities per Minute. It was also the basis for my personal testing, which I linked earlier in the thread, that was at the very least consistent with the PTS.

The OP is correct that different disciplines may have different alacrity rating needs. But the OPs testing is not designed to answer that question, since all the OP does is spam the basic attack, which is nothing like a real rotation, especially for healers. In his defense, he is not claiming that he is presenting an "optimal" alacrity rating. But I truly have no appreciation for what, exactly, the OP is providing here. This is especially true in light of the fact that the OP has claimed they are not trying to replicate or validate any of dipstik's testing. I think the OP is trying to present some real examples of his effective GCD when spamming 40 basic attacks or spamming for a minute, whichever is longer. That's fine, but the way he is discussing his data, in terms of "raw" vs "effective", and "excess points in alacrity" is what is confusing.

I made an incorrect assumption about the conclusions the OP was drawing. He wrote:

This character is in the opinion of some carrying 90-100 excess points in Alacrity Points for a 1.3 GCD.

Which, while slightly confusing grammatically speaking, I interpreted as:

This character is of the opinion that you need to carry 90-100 excess points in alacrity rating to reliably get a 1.3s GCD

In other words, I assumed the OP was hypothesizing that a player needs to have a certain amount of excess alacrity rating, beyond the number required for a specific GCD, in order to "offset" things like lag, latency, fatigue, human delays, etc. ... all those variables to which Tsillah is referring. It is an interesting hypothesis, and one worth exploring. Dipstik didn't find a statistically different result between an alacrity rating just above the threshold versus one hundreds of points above. However, it appears that my assumption was wrong, and what the OP was really trying to say in the quote above was this:

This character has 90-100 excess points in alacrity rating above what is needed to get a 1.3s GCD

which isn't an opinion, its a fact. At least, that's what I am inferring from his replies subsequently and his accusations against me.

Now, those results don't necessarily mean that an individual player won't observe some difference with 90-100 excess points of alacrity. But, does the hypothesis itself make sense? In other words, would we expect that some excess alacrity rating would offset those execution variables Tsillah is discussing?

The GCD is calculated on the fly, after an ability activation. We know this because abilities, relics, or adrenals which affect alacrity cause changes to the GCD right then and there. For prevention of hacking, we can safely assume all those calculations are occurring server side. Therefore, I suspect that the server is determining the "delay until the next ability can be activated", which is the definition of GCD, based entirely on the alacrity rating and alacrity percentage, and by its very nature cannot include variables it cannot measure, like lag, latency, or human response time. The way to validate that is to, as I have outlined before, and as dipstik did, test the APM at two different alacrity ratings: one very close to the threshold and one several hundred points above, and see if the APM after 4.5 minutes is statistically different.

Lastly, I'll just clarify that I have no agenda and no personal grudge against the OP. I am very invested in this topic, as really most of us should be, because how we allocate our stat budget is relevant to just about everyone. Therefore, I strive for accuracy in both testing methodology and in conclusions from the results. I have made my case for why the OP does neither, and he has accused me of trolling him and providing strawman arguments. As a result, I decided to step away until now. I hope others find the discussion informative in some way.

don't you wish you could simply ban people from your thread.

You can, indirectly, by reporting posts and then the moderators delete posts that are offensive.

One last comment on the OP’s more recent methodology. Temple Of Sacrifice aside, the greatest common factor of 1.3, 1.4, and 1.5 is only 0.1 seconds. Any delay value higher than that will result in input to the game that will not align with any GCD interval, which in turn will interfere with your resulting APM. My fear is that input that quick might trigger a response from the Revanities, regardless of the scientific intent. Either way, the method reduces human response time error but introduces a new error if the delay is not a greatest common factor of the GCD values being compared. So, take my methodology comment as you wish.

Back on topic,

Can anyone tell me what alacrity to run for Arsenal Merc? It’s been 3 years since I’ve played one and I’m not sure wether my goal is 1,4 or 1.3 GCD.

Also what alacrity percentage am I looking for to get these?

Back on topic,

 

Can anyone tell me what alacrity to run for Arsenal Merc? It’s been 3 years since I’ve played one and I’m not sure wether my goal is 1,4 or 1.3 GCD.

Also what alacrity percentage am I looking for to get these?

Gonna disappoint you here Trixie...

Short Answer:

Doesn't seem to matter at all.

My Arsenal Merc has the priming shot tactical and the gear that boosts damage for standing still.

At the minimal 1.4 GCD (1220 actual) and the higher Heat Dissipation Rate 1.4 GCD (1444 actual) and at the minimal 1.3 GCD (2376 actual) . . .

She pulls 15-16k off dummy and 7.5-8.5k in fight. Same basic rotation...irregardless of which GCD Tier.

Sure the 1.3 GCD allows for faster heat dissipation and faster recovery - but she overheats easily if you're not careful.

Long Answer:

Talked to a few NiM raiders in private whispers while in SM and HM OPs, and it seems they are ignoring base GCD almost entirely.

Little or no base alacrity.

Devastating Vengeance Relic

Alacrity Relic

Stacking 3500+ Critical Points

And when the Dev Relic triggers, they pop the Alacrity relic OR an Alacrity boost ability as a matter of rotation . . .

...and during that short window of 1.3 / 1.2 / 1.1 GCD and Supercrit . . .

Burn Torque to a crisp in two minutes flat, laugh at Master/Blaster and leave Tyrans dead before he even finishes his "famous quotes".

It's funny to watch - just can't help but laugh.

When it's a bunch of Sorcs together in a 16m, they burn so hard, no heals required, tanks are ignored....it's that crazy.

This is in Legacy OPs, not the Dxun OP.

I'm guessing it'll continue this way until Bioware finally gets around to their next nerf.

A nerf is coming...count on it.

=8-(

until now i had ( due to some people suggesting it ) a little overflow on alacrity, like 0,5% above the gcd-thresholds to fight lags from connection/pc/human/whatever to still reach/get the shorter gcd i aimed for

now this thread made me think about it again

regardless which problems occurs (connection/pc/human/whatever), the gcd will everytime the same rounded number

IF there are lags, than the ability (and here gcd) will be executed later or there is a short window after her gcd and the next ability will be delayed BUT the gcd of this ability will STILL get the calculated gcd

An Overflow on alacrity will not affect those delayes and it will not shorten the gcd further to reach the calculated gcd with lags taken into calculation

thats why the apm cannot change/get better with an overflow on alacrity (ignoring further reduction on 3sec-casts for the moment please) when problems/lags occur

would you say there is something useful in this thoughts?

this would mean every little poor single point above the thresholds are wasted, right?

this would mean every little poor single point above the thresholds are wasted, right?

Not necessarily. Remember that the OP is not claiming to be providing an optimal number per discipline or testing any rotational abilities. There may be specific circumstances or where you may target an alacrity percentage higher than the 15.4% required to have a 1.3s GCD. For example, if you push the alacrity percentage up to 16.36%, sages and sorcerers can achieve a 1.1s GCD when under the buff from Mental Alacrity/Polarity Shift because those abilities add 20% alacrity, and 36.36% alacrity percentage is the breakpoint for 1.1s GCD. Now the window is short, but it does allow for abilities like Healing Trance and Wandering Mend to be ready an entire GCD sooner than if you were at a 1.2s GCD during MA/PS. Of course I haven’t re-done the calculations with the 4-piece revitalized mystic set bonus cooldown reduction for Healing Trance.

But my point is that specific disciplines may have casts, channels, cooldowns, or buffs that would prompt someone to run an alacrity percentage more than what is required for a specific GCD.

Not necessarily. ...for a specific GCD.

ok, certain classes have abilities that push alacrity further for a short time and to optimize this i need an alacrityvalue which calculates with those abilities to reach the next gcd breakpoint

maybe for my sorc and the speed marauder i will look into this

But if i only want the "standard" 1.4 / 1.3 gcd i only need the exact values for 7,15% / 15,39% ( yes some classes have a bonus on alacrity :-) and no additional points, right?

i was unsure if my considerations where right in my last post :-)

So, I'm reading here:

1. The math says we should have 1213 alacrity for the 1.4sec threshold on classes that have no inherent alacrity buff, assuming there's no alacrity guild perk active. This is based on equations derived from the game code or released by BioWare or both--I don't know which off the top of my head.

2. The OP has noted he needs more than that to get to the 1.4sec threshold.

3. Others have noted this increase could be due to computer processing time, humans being human, server latency, internet lag, etc. etc. etc.

So the take-home message:

Reaching the alacrity breakpoints is a great starting point based on the math, but individual players may experience some loss due to the factors listed in point 3 above and may need to add more alacrity to compensate.

The amount of alacrity one might need to add could vary tremendously, however. My hand arthritis will undoubtedly make me a little slower hitting the keys than others, and client and internet latency can vary tremendously, so there's no good way to predict just how much alacrity an individual player might need to overcome those issues to reach the 1.4 or 1.3sec breakpoints.

I'd recommend anyone who likes to do end-game content download StarParse and learn how to interpret the results. Then, figure out what you individually need to make the 1.4 or 1.3sec breakpoint happen reliably for you, and simply adjust alacrity accordingly.