19 replies to this topic

[spweasel]

I'm programing a rough DPS simulator to help me get a feel for Ranger (optimal build/rotation) before making a final decision between it and BM (and because I've been wanting a good excuse to learn Python).

What I'm not sure of is how much time to put as the delay between attacks. So while PA might be "instant cast", I'm fairly certain that it cannot be followed by DS on the very next frame. Currently, I'm listing all animation/delay times as .9 seconds except for Arrow Volley, which I have down as 2 seconds. But since I only have an Archer currently, I have no way of knowing what the general downtime after most attacks really is.

Any suggestions/guestimates? Also, just to confirm, Haste and Vigor have no effect on animation speed/ "global cooldown", right?

[indignation85]

Haste affects casting bar times i.e. Charge Arrow and auto attack skills. Vigor affects GCD and cooldowns of long abilities which is why I personally believe the crafted items are bad for any Ranger/Rogue.

Regards,

[Finraziel]

If you're doing it as an excercise in python then more power to you, but what exactly are you hoping to determine in regards to ranger/BM DPS? Ranger DPS kicks the out of BM DPS, I can tell you that right away. A good ranger can make me have some trouble overtaking them in threat in grizzly mode (meaning I get triple threat) if they attack before I do. I still outthreat them, but if the DPS would be equal I would have three times the threat (what I did have a lot earlier when a rogue was so happy with his DPS... decided not to burst his bubble ) or at least double, since bearform would increase my damage of course.

Edit: Oh and don't just use the times listed by the skills or the global cooldown (which is 1 second). At least for beastmaster the animation time matters a lot. Rage smash takes 1.4 second, brutal strike 2.4 seconds, wild crush and cruel bite I'm pretty sure are 1.4 second as well. I don't know about ranger, but I know these times I found for beastmaster (by timing how long it takes me to do 10 attacks with the skill) are not the same for other classes. My priest and sorc can use instant skills 10 times in 10 seconds for instance, having exactly the 1 second global cooldown for each skill.

Indignation85, are you sure on vigor affecting global cooldown? I used to think so as well because my lightning bolt was faster in colo, but when I tested with putting on vigor equip, I couldn't detect any difference in castingtime so my guess is colo just does something else (maybe directly altering GCD) but it's not the vigor that speeds casting up.

Edited by Finraziel, 22 June 2013 - 03:56 AM.

[indignation85]

100%, I have 17.05% Vigor and casting WW 5/5 (10% Haste, 10% Vigor) I can see a difference in GCD).

Regards,

[spweasel]

If you're doing it as an excercise in python then more power to you, but what exactly are you hoping to determine in regards to ranger/BM DPS? Ranger DPS kicks the out of BM DPS, I can tell you that right away.

I knew that much. I'm not testing BM DPS at all, I'm just trying to get a feel for Ranger before commiting. (BM DPS has already had a pretty lengthy thread created about it, too). BM's higher defenses and self-heals make it a very appealing class for soloing, not to mention Tanks have been a precious commodity in every WoWlike I've ever played.

If Ranger's optimal rotation and build simply doesn't look like fun, I'm not going to bother with the class. I tend to prefer Archer-style characters in games to heavy brawlers, but RO2 makes those Brawlers look fun.

As for the topic, I think I'm going to stick with a .9 second delay between attacks (aside from AV, which will be 2 Seconds) to account for human reaction time and latency.

Edit: I tried to make this post a little less rude. I had no ill will, I just didn't realize until I reread it that it sounded way more jerkish than I had intended.

Edited by 544130611113304573, 22 June 2013 - 05:28 PM.

[Finraziel]

Ah I see, that makes a lot more sense And don't worry, I read your post before you edited it, it didn't seem that rude to me and I probably had it coming
.9 second might be a little optimistic though, generally global cooldown (meaning cooldown before you can even try to start a new skil, regardless of if that skill is on cooldown itself or not) is considered to be 1 second and I myself can't get faster than 1 second either. I do live in europe which increases my latency, but I'm measuring this by just mashing one skill button constantly so I don't think the latency will matter much. If you actually have to react to procs and the like, you're probably going to be a bit slower still no matter how quick your reflexes are..

As for vigor affecting GCD, I tested it again and I still can't see any effect. What I did is I took my sorc, removed all equipment to get a vigor of 0 and cast 20 lightning bolts using a timer to see how long it takes. With 0 vigor I came to 20.1 seconds. Then putting on my equip I got 4.85% vigor and timed 19.6 seconds, which might indicate an effect, or I might have stopped the timer slightly sooner than I should. 4.85% of 20 seconds should be about 1 second. Then I enabled my guardian, which unfortunately is only lvl 1 so it adds 10% vigor bringing me to 14.85%. TIming it again gives me 20.0 seconds, when it should be more like 17 seconds if vigor would be fully applied to GCD.
To verify I now switched to my priest who has a lvl 2 guardian. Timed 10 renovatios to be 28.1 seconds with 0% vigor. With 20% vigor from my guardian, I timed 28.4 seconds, again within any measuring error I would make manually timing if there is no effect, but if vigor would fully apply it should have been more like 23-24 seconds.
How did you actually test this Indignation? If I'm not mistaken, ranger only has one skill without any cooldown on it, arrow shower. If you tested with any other skill then you may just have noticed the skill's cooldown itself being reduced. I don't have a ranger myself though, so just going by ro2base...

[spweasel]

Good, then. I was afraid it may have come across as condescending, and some people get uppity at the word "Duh" for some reason.

Are Haste and/or Vigor applied as an increase in speed, or as a decrease in time? That is, for an attack with X seconds cast time or CD, would 20% Vigor/haste be:
A ) X * 0.8
or
B ) X / 1.2

I have a working version of the sim, though I'm keeping the elapsed time to a relatively low 1000 Minutes (16 Hours, 40 Minutes) to keep run time reasonable (that's still 3 million frames at 50 fps). Different runs with the same parameters are still within about 1% of each other, though.

Damage breakdown for the impossible "maxed everything" (hereafter "ME") build with a natural 20% Haste/Vigor and 35% Crit from pre-IC Agi:

Total DPS: 103.14% DPS
AutoAttack (AA): 8.00% of total damage
PA: 18.03% of total damage
FA: 9.48%
CA: 23.91%
DS: 26.38%
SS: 3.36%
AV: 10.84%

Thoughts:
1) All skill builds deal similar damage. Using arbitrary runs, the lowest DPS build I tested (92.6% DPS) and the highest (100.9% DPS), are both within roughly 10% of the ME build (103.1% DPS), which would indicate that specific builds are generally less important than things like gear stats or how often you can turret.

2) Haste is mediocre. Since it only increases AA damage and one part of one attack that one makes up only one part of your final DPS... yeah, the returns per point gained are pretty meh.
2b) As such, FB is kinda meh, and you're better off getting AV on cooldown ASAP than waiting for FB stacks to almost expire.

3) Poison Arrow deals way more DPS if used every CD rather than waiting for the DoT to expire. The damage to time ratio on the initial attack tends to favor PA until higher Haste levels, but the Concentration gain to time ratio on PA is much better until obscene Haste levels (CA's Cast Time would need to be around .25 seconds), since you are functionally comparing 24 Con with a cast time to 19 Con instantly (assuming 5/5 RM).

4) One sizable issue with the sim is that I don't have it as able to Double Falcon (since I implemented FA as a single DoT that can Crit). If you have WW, it isn't a massive deal (since you'll want to wait for WW to be put on CD before popping TS anyways), but for WW-less builds you are pushing back your next FA by roughly 30 seconds.

5) SS is better than expected. That 1% extra from maxing it out adds up since it applies to absolutely everything.

6) Of what I tested, this build performed best. And I recorded tests for 7 different non-ME builds. I can post the builds/DPS if anybody wants them.

Disclaimer: I'm not claiming that my sim is correct, and any or all of the above assertions might be complete bullhonky. Indeed, I'm fairly certain that there are some pretty big issues with it, not least of which being the FA issue mentioned above.

[indignation85]

Since I run 17.05%, I'll respec 5/5 Fear Breeze to see if 20% haste will affect my GCD. If it does, then it's Haste, if not, then it's Vigor. (5/5 Fear Breeze = 10% Haste, 5/5 Wind Walk = 10% Haste for a total of 37.05% Haste) No gear guardian gets me up to 30% Haste as well and you can definitely see the difference in GCD reduction, although it's important to note that guardian also gives 30% Vigor, so it's hard to specify what helps at this point.

Now do not mix up animation times with GCD, because they are completely different. I've seen numerous posts that specify that animation is not affected by either Haste or Vigor, regardless of what the tool tips say (just google it).

I'm more interested in what affects GCD, which I'm certain is Vigor since I've ran a full Darkness crafted set (full of haste, no vigor, and I do not see a difference in GCD).

Regards,

[Chiffon]

I'm more interested in what affects GCD, which I'm certain is Vigor

So here you say you're sure vigor affects GCD,

(full of haste, no vigor, and I do not see a difference in GCD).

and immediately afterwards, say that you saw no difference between having and not having vigor.

What.

[Finraziel]

You read that wrong, he says he had lots of haste but no vigor, so if vigor affects GCD then it makes sense that it would not be affected in that situation.
What I'm more interested in is how you're determining that GCD is affected since as I said, rangers only have 1 skill that does not have its own cooldown. I think it'd be very weird if both lightning bolt and renovatio are completely unaffected if GCD is in fact lowered. I guess it's possible that the animation time is exactly 1 second for those skills as well, it would be stupid, but then, so much about this game is stupid, it would make a weird kind of sense...

edit: Now that I read that back I think I may be misreading your post as well. Maybe you did get that there was no vigor but you mean you think he said that he saw no difference between lots of vigor and no vigor? No, he means he sees no difference between being naked and wearing the set that gives lots of haste but no vigor.

Edited by Finraziel, 24 June 2013 - 08:39 AM.

[indignation85]

So here you say you're sure vigor affects GCD,



and immediately afterwards, say that you saw no difference between having and not having vigor.

What.

Thank you for reading that correctly Fin.

Let me rephrase this in layman's terms.

I have a full Darkness Valkyrie Set (each piece runs a considerable amount of Haste, I believe around 200ish per) with no Vigor. GCD refers to the spin animation of your skill bar after using one move. Obviously, if you have a cast time, this is not as important because the GCD will have refreshed by the time you finish casting. It is only important for instant casts so you can immediately use the skill again.

Now my assumption is based off taking all my gear off and casting Guardian. Naked = 0% Haste & Vigor and level 3 guardian gives you 30% of each. I see a considerable difference in my GCD for instant casts allowing me to shoot out Double Strafing faster.

Like I stated earlier, I am in full colo gear and running 17.05% Haste and Vigor. Since I am in only 1/5 in Fear Breeze, I only have an extra 2% Haste with 5 stacks. I am going to respec 5 points into Fear Breeze, giving me a total of 10% in addition to 10% off Wind Walk. This should be more than enough to get me at the Haste with a naked Guardian. If my GCD is similar to that of the naked Guardian, then it proves GCD was affected by Haste. If not, then it's safe to assume it's Vigor which I am assuming it already is.

Now for the kafra/zenny to buy that respec....

[Finraziel]

Hmm, odd. Strictly speaking double strafing has its own cast time of course, but since it's less than the generally accepted GCD of 1 second, that shouldn't matter. If you're casting more DS's than 1 per second, then it would suggest that GCD is indeed being affected. I just don't get why it doesn't work like that at all for other skills (unless DS is just special in ignoring GCD edit: no you already ruled that out with the lots of haste/no vigor set).
I guess for now the only thing to do is just test it out per skill to decide what's best for a certain class/build. Thanks for sharing.

Edited by Finraziel, 24 June 2013 - 12:26 PM.

[indignation85]

I raid often to help out my guild mates and test DPS builds and during raids, guardian is often used. With that being said, yes I can cast 2 DS within 1 second, but that's because both my Haste and Vigor are at 57.05% (30% from guardian, 10% from WW). You can test it yourself too if you'd wish. I'm certain a full Colo set gives 17.05% Haste and Vigor, I don't believe there are any passive skills that raise that outside stacks.

[spweasel]

For what it's worth, I'm almost sure I saw mentioned somewhere else that there is no real "GCD", only the cooldown after attacks and animation times. Can I find it now? Of course not ;_;.

How often you can Double Strafe means next to nothing for testing purposes, since if there is no actual GCD more Vigor = more DS regardless of other factors. Instead, try timing something like PA -> SS, where the attacks have too long a cooldown to be completely spammable regardless of your Vigor/Haste Values.

And since I never really got an answer as to the whole "increased speed vs. decreased time" thing: With that 57.05% Vigor/Haste, what is the cooldown of FA? Can you tell if Charge Arrow takes closer to 0.43 seconds, or 0.64 seconds?

[Finraziel]

2) Haste is mediocre. Since it only increases AA damage and one part of one attack that one makes up only one part of your final DPS... yeah, the returns per point gained are pretty meh.
2b) As such, FB is kinda meh, and you're better off getting AV on cooldown ASAP than waiting for FB stacks to almost expire.

I thought this as well, and I was wondering why just about every build I saw included wind walk. It seemed like with haste being mediocre for other classes and vigor seemingly not affecting global cooldown (tested on my sorc and priest) that this skill would be kind of lackluster. But since this doesn't seem to hold true for different classes (or GCD perhaps in fact not existing as such), and me making a ranger now as well, I thought I'd try to test it out.

So, I just became ranger and wanted to test it now without spending any skillpoints (since main ranger would immediately add in more random, for instance). Went to the PVP arena, equiped a crappy bow so I wouldn't kill the lvl 25 dummy too quickly (the higher level ones I miss too much) and tested how much damage I could do to it in 30 seconds.
I did this 8 times and while I had 2 lucky streaks where I got it down to around 900, on average it had 1235 HP left, meaning I did 3205 damage.
Then I used my guardian, which is level 1 and so grants me 10% haste and cooldown. Then I tried again and did 3600 damage. I waited half an hour for it to recharge, and got to around the same. This would mean I had about a 12% damage increase.

I may have been lucky, but I also tested how many charge arrows I could throw out in two minutes (on the big raid dummy this time so it wont die ). For this I removed my pandora gear to test with 0% haste and vigor, and with it to test with 8.13% haste and vigor. Without haste/vigor I got 52 casts out, with it I got to 57. That would be an increase of 9.6% in damage output. Even if there was some rounding error or something and it should be 53 vs 56, that's still a 5.7% increase, still not bad at all.

All in all, it does look like for rangers, haste and vigor, at least when they are equal, show much better scaling than for other classes I've tested. I'm not sure what that means for fear breeze, since it's haste without vigor... but at least it seems like wind walk would actually be a worthwhile investment.

Now if only this would help me in deciding on the build I want to make Still don't really have a good idea...

[spweasel]

Hmm... I'll have to look into this. Was the first set of tests pure CA spam, too?

[Finraziel]

CA + DS whenever I had the concentration available.

[spweasel]

If we assume a 1 second CT and a .9 second CD on CA (as per the skill description), that leaves 0.41 seconds unnaccounted for between shots, most likely due to animation, latency, human reflexes, etc. Applying 8.13% Haste/Vigor to those numbers (but not the 0.41 seconds, which should be consistent regardless of stats), you are looking at an expected 55.73 CAs in that same time period. Considering the existence of rounding error on both tests, this seems reasonable.

Looking at it from the other direction, a 0.36 second delay would have the second test come out to exactly the right number (but the first test would have had 53.1 CAs).

I'll try revamping my code later to accomodate these numbers. I'll use a .38 delay in order to get some hopefully more accurate numbers.

Edit: Code updated.

No huge differences, though I only tested 4 different builds (I actually tested several more, but screwed up fixing the numbers several times before I fixed things properly). At the stat level I tested (10% Haste/Vigor, 20% Crit), this build performed best.

I did some practical tests, but unfortunately they mostly confirmed one thing - I kinda suck at playing a Ranger. There's a little too much going on for me to play perfectly without much practice, especially if I'm trying to add PA to the rotation instead of only using it when the DoT is down.

Considering I still haven't maxed PA, AV, or MR yet, trying to work PA into my rotation actually caused the fight to last longer, sadly.

Edited by 544130611113304573, 30 June 2013 - 06:47 PM.

[6444130507205735953]

I've been looking for a dps simulator of sorts and I just came across this thread - is there a way you could share the simulator? I have tons of builds I'd love to try and see for myself the differences. If not, could you test this build if you haven't? http://www.ro2base.c...1.1.1/0.0.0.0.0

My build right now is the same except WW lvl 5 FB lvl 1 AV lvl 1 and FA lvl 5. I'm curious about WW lvl 5 and FB lvl 1 vs WW lvl 1 and FB lvl 5. The total haste is 12% either way with fear breeze at 5 stacks so I'm curious as to which is better. Sustained/accumulative haste or burst haste. Also, what does more DPS: maxed falcon and lvl 1 AV or AV lvl 4 FA lvl 2?

Thanks!

[spweasel]

I'd rather not share the simulator as it is unless I were passing it on to another programmer for more development. It simply is neither user-friendly nor accurate enough for general use, and my interest in the project isn't high enough to do the detailed testing needed to get it into perfect working order.

 

As for that build you linked in particular, I can tell you a few things:

 

1) Don't use 4 AV/2 FA. There is absolutely no reason to leave them both partially maxed. Either max both or drop one to 1 point.

 

2) FB is probably not worth investing any extra points into. My simulator indicates that Haste is pretty bad to begin with, and if you are going to use AV at all it should be done ASAP.

 

3) WW, while more useful than FB, isn't great from a sustained DPS standpoint. It does have the advantage of being on-demand, though, which means you can save it for moments when the boss feels like letting you just beat on it. There is also the speed boost to consider.

 

It's your call what you use, but the sim does indicate that 1 FB/1 WW/5 AV/5 FA is best under optimal conditions. Even 1 point in WW is better than dropping it for 3/3 SS, though.