Combat Feedback for Update 5.

[Tamerlane]

Acrobats might of already been addressed, but I haven't seen it, so heres my questions. . .

What about the speed boosts Acrobats get with their staves?

it was said earlier that this was one of the haste effects that would not change.

but if you have a acrobat/monk build that uses wind stance, that is being changed to only affect offhand attacks so it would no longer stack

[Arkat]

...
You acknowledge this bug..

Yet the fact that the capstone is bugged in another major way - it never increased glancing blow damage.. You never comment on, and leave out.

Fix the broken aspects making it worse before worrying about the broken aspects making it better.

Ah, there's the Shade we all know!

I don't suppose the word "please" is in your vocabulary, huh?

If you used it every once in a while, you might actually get better results both IG and IRL!

[Deamus]

actually your doing the same thing the crying rangers are .. .not seeing the big picture ...

17% dps loss for a ranger which is a class that is a high dps class right now means 17% less DPS that is it. actually only brings them to the dps they should have had all along before this tempest thing got out of hand.

14% dps loss to a monk which is already considered one of the weakest melee classes is HUGE ... then add the lack of ki generation which lowers dps and only benifits monks have in this game (attack speed) and then force them into wind stance (which is the slowest ki generating stance) and that 14% is massive ... pretty much screws monks over completely ...

and a 10% chance WHILE IN WIND STANCE for the dark fist finisher means nothing to a light monk who actually uses his ki frequently ...

dark monks dont lose as much as light monks do ... but even they need to have 50 ki to do there attack which will be 14% slower now if not more cause they cannot be in fire stance if they want there double strike.

if you cannot see how monks loss of 14% is way way greater dps loss and ability loss to a weaker dps class then the rangers loss then maybe you dont understand Monks at all ...

monks get the raw end of the stick in this exchange .. .pretty much puts us out of any party that isnt TOD or that we dont solo most of it ourselves first (aka gen point)

+1 Rep for you.
Consider also that they are looking also to give something like double strike to Rogue ..... So only Monks, which had problems with low dps get nerfed the most with the new changes.

PS. : Devs please do something for monks they need a boost like rogues did.

[bandyman1]

Ranged aggro is buggy. I believe you know well enough to know that, so I don't know why you hold raid bosses' (rightful) hate of ranged attackers up as a serious point. It certainly isn't because the ranged attacker is doing more damage...

Really???

Because I saw Shade lose Agro in a VoD elite on beta to a Dex-based TWFing rogue.

Who said anything about a ranged attacker???

[megathon]

How about you nerf the greensteel and not the monks.

Sounds like a monk lover that doesn't get GS handwraps.

[Aesop]

Your tables got a bit screwed up Aesop. Looks like you're suggesting making base off-hand proc a bit more frequent, and then reducing the amount the TWF line gives.

yeah I hate tables.. .they never come out right

[Vallen-Mabar]

it was said earlier that this was one of the haste effects that would not change.

but if you have a acrobat/monk build that uses wind stance, that is being changed to only affect offhand attacks so it would no longer stack

Well, i suppose the dex boost to sneak attack for the acrobats might still be worth that monk stance? Or maybe the capstones new +2 dex would be better. . .

[Boromirs]

I like the idea quite a bit.


It still works. We set it up to grab multiple targets as necessary in the initial check.

Eldarin-

Will you tell us the full scoop on what will be implemented before it goes live?

[Razcar]

Is 11 levels considered "splashing"?

A rather big splash I would say.

[Angelus_dead]

I would urge caution on any speed up of ranged attacks. Alot of players do not realize how powerful range combat is especially if you use improved precise properly.

Yes, one of the steps to improving ranged combat would be nerfing IPS. That way ranged combat can be given decent DPS against single targets without being overpowered against groups.

Probably the IPS arrow should do reduced damage against the 3rd enemy and beyond, rather like a glancing blow.

[Razcar]

Well, i suppose the dex boost to sneak attack for the acrobats might still be worth that monk stance? Or maybe the capstones new +2 dex would be better. . .

Those are meh and double-meh, respectively.

[sephiroth1084]

Still catching up through the thread, but a few comments:

I'm torn, Eladrin. I don't know whether to pat you on the back for slogging through the 80+ pages since you [apparently] last read, or give you a sympathy hug for slogging through the 80+ pages since you [apparently] last read.

[Tamerlane]

a couple of quick suggestions-

currently tempest II gives no rate of attack increase, but will if the proposed changes remain.

If you remove the 10% offhand from tempest II and include both hands in all bonuses (fighter capstone becomes 110/90. zeal becomes 110/90, wind IV becomes 110/90, tempest I becomes 110/90 and tempest III becomes 120/100) everyone still winds up with less total attacks (by about 10%) but its about the same decrease for everyone.

it would be possible to get more then 100 percent for off hand attacks with creative multiclassing or the addition of future items or bard songs, but it could be made into a chance for an offhand doublestrike.

[xxScoobyDooxx]

Your comments about code efficiency are worthwhile ; however, the DPS lag issues also occur in Europe, when there may be neither 500,000 nor 100,000 online --- but 50 or 100.

I'm not saying that so many people online won't cause some sort of lag, but the issue being addressed by these changes is NOT a population lag ; but a specific lag appearing in certain scenarios irrespective of how many people may be online at the time.

Just to clarify .... the numbers used were purely examples and not meant to be an accurate representation as I have no idea as to the numbers of users online involved.

Maybe if I rephrase .... the points I wanted to make were.

1. LAG and Packet loss are directly related to network traffic within the ISP (or the Internet in general) or at the DDO hosting site (i.e their Internet connection) If they have a 500Mb connection and 700Mb of traffic is trying to pass through that point you will get LAG or Packet loss depending what method of rate limiting their ISP uses to restrict them to 500Mb. The ISP will either delay sending the extra 200M by storing it in buffers = LAG or they drop the extra 200M = packet loss.

2. Code has a limit on how efficient it can be made. (NOTE if you are NOT changing any functionality!!) 10 tasks in 100 lines of code ... if you still have the same 10 tasks in 80 lines of code then it is more efficient code. If you have 8 tasks in 100 lines of code or 80 lines of code then it is different code not more efficient code. (which I am sure can be argued but the point is don't confuse rewriting the same code more efficiently with rewriting code with less functionality.)

3. If you increase the number of users and the code remains the same you will reach a point were the servers become the bottleneck.

This is a plain and simple networking issue. Reducing the amount of network traffic and or CPU processes by simplifying TWF is one way of addressing it. It does seem like the cheapest way for Turbine (utilising salaried staff to change code which is zero additional cost for them) but unfortunately it's the one that will upset the most people.

IT problems are best solved by changing one thing at a time and remeasuring the results. Scattergun approaches invariably lead to creating new issues while often not resolving the original ones.

It would be fascinating to know if Turbine can recreate the in game lag in the test lab and if they can't find out why not.

[Coldin]

Yes, one of the steps to improving ranged combat would be nerfing IPS. That way ranged combat can be given decent DPS against single targets without being overpowered against groups.

Probably the IPS arrow should do reduced damage against the 3rd enemy and beyond, rather like a glancing blow.

Or just only hit the first 3 targets.

[Blivit]

This thread is over 145 pages long now, and I don't have time to read through it all. So, sorry if I've missed previous discussion that is important. I did at least read all of Eladrin's responses so far.

I play a healing amp monk, and as has been discussed recently in the thread, decreasing the number of hits per second decreases ki generation, which is very undesirable (to me at least), since I use it as a major source of healing. I'm wondering if there are other ways to optimize the attack calculations without simply reducing the number of attacks.

What are the major sources of lag? Is it computations on the server side? Packet exchanges between cilent and server? Which contributes more?

I will assume that random number generation itself, for rolls, is not a significant source of calculation time. However, if we assume that Turbine is using a statistically acceptable random number generator such as the Mersenne Twister, rather than something horribly bad like a linear congruential generator, then perhaps the calculations can be a little costly if the server load is really large (?). If the random number generator is taking a significant amount of time, and isn't already using a very fast method, then I would suggest using an XOR-shifter such as those of George Marsaglia (http://www.jstatsoft.org/v08/i14/paper), which pass most or all the DIEHARD tests for randomness and are very simple and fast to calculate. The period 2^32 - 1 method is super-simple (I suggest [<<6,<<5,>>13], most of the other combinations don't actually work correctly), and is PLENTY good enough for dice rolling purposes, since it passes all but two of the most stringent DIEHARD tests. I might also assume that if random number generation would have been a significant facor, a large number of dice rolls would be pre-rolled ahead of time during lighter load and cached for when they are needed later, thus making any discussion of random number generation time moot. Since I don't know anything about how things are coded behind the scenes, I'm just throwing this out there as possible problems and solutions.

We are using the same physics detect as your main hand attack, and then doing the attack rolls for additional hits.

---

When an off-hand attack triggers, it has a standard attack and confirmation roll. Instead of always getting off-hand attacks on certain attacks, you instead have a % chance of getting an off hand attack on every swing.

Next, is the lag packet exchange related? If off-hand attack physics checks will now be based on the primary hand, why not just roll both attacks into a single damage packet, since they are both occuring at the same "time"? Rather than getting two sets of number scrolling above the enemy's head, why not add the damage results of the primary and secondary attacks together and send them as one packet of information, rather than two separate packets of information? So, you'd get only 4 sets of floating numbers per round (assuming normal attack speed), but each set would contain the total damage from both the primary and off-hand attacks that went with the primary attack. Some players might miss seeing the individual damage results, but I think most would prefer lumping them together rather than a nerf. I don't know how much overhead goes into sending/receiving the information that goes into causing the separate packets of damage to be displayed over the enemy's head, but perhaps consolidating them into bundles for each corresponding primary attack would help? Again, since I don't know the details, I'm just throwing this out there as an idea.

There's actually been a lot of very useful feedback in this thread.

Some things are very light, performance-wise. "Chance to proc" is one of the lightest calculations available.

Edit: Actually, even those chance to proc checks are improvements, since it's no longer checking your feats constantly, but instead looking at a cached value.

This got me to wondering if the weapons themselves are cached, or need to be looked up for every attack. If they, like the feats, were not cached before, are they cached now (updating cache on each weapon change), and if not, could this help too? The same goes for stats and other bonuses. Were they previously uncached, and are they now cached (updated only on equipment, stance, etc. change) to improve performance further?

I don't mean to question the technical competency of the developers, especially since I have no idea how things are actually implemented. These were just thoughts that came to mind, and when the nerf bat comes a-swinging, I just wanted to throw out any ideas that might help shield against it.

[Auran82]

It would be fascinating to know if Turbine can recreate the in game lag in the test lab and if they can't find out why not.

It would be to me as well, as from memory, the following are situations where you have a raid where there is fairly consistent DPS against a single target:

VON 6 Epic
Shroud Part 4 *
Shroud Part 5
ToD Part 2 *
Tod Part 3

Where only the 2 with *'s actually have the major lag issues, and from what others have said, they have tested things like Shroud part 4 with TWO melee hitting harry (from the sounds of it both using THF) with the rest of the party either healing or arcanes doing their thing and they still experienced the same lag as you get with 9 or 10 people hitting harry, that has to put up alarm bells that there is something else happening there.

Shroud part 1 does sometimes have some lag, but it's not really the same type, some peoples computers/connections for whatever reason sometimes just do not like that area. I have seen shroud part 5 get some DPS lag, but from memory, it's usually in a group with less than spectacular DPS and it's usually after the fight has been going for a while and the blades are out in force.

[DarkSim87]

maybe its the raids that need to be changed..

[Aesop]

well since I can't get tables to line up and they all explode lets try something different. The way I see it this idea of having things proc could add some diversity and what not to the game in terms of viable functionality and styles.


I'd call this the Triggered Style Effect. The triggered style effects are those that are attached to a specific combat style. The most commonly thought of styles are:

Two Weapon Fighting
Two Handed Fighting
Weapon and Shield
Unarmed Combat
Ranged Combat

there are a couple of seldom considered other styles:

Single Weapon Combat
Thrown Weapon Combat


Two Weapon Fighting
Started the discussion because E says that the many rapid collision detections are wreaking havok with the calculations and cause certain types of Lag. TWF also is usually the highest form of DPS in the game. In fact TWF caused enough problems that they needed to upgrade THF and then people started to realize that the other acknowledged style S&B was even further behind. Apparently THF is still behind as well because the devs want to reduce TWF a bit as well. I have a number of twf and I really don't mind this all that much. really now we are just haggling for the average number of attack we will have.

So the current suggestion gives the following cumulative bonuses.

NO Feats: 20% (menaing 20% of the time a main hand attack will trigger the roll of an off hand attack.)
TWF: +20% (each tier of the TWF chain grants an additional 20% upto a total 80% of the time a main hand attack will trigger an off hand attack.)

After that there are the Prestige Enhancements

Tempest 1&2: (Each tier granting a 10% bonus bringing the grand total to every attack with the main hand triggers an off hand attack.)

No other listed bonuses actually add to or subtract from the Off Hand Attacks.


I'm not sure that these numbers will really work out in the end but one thing I'd like to see changed is...


Fighter Capstone currently gives a 10% alacrity bonus to every style.

The proposed change is that it grants 10% Double Strike

Counter Proposal: Grant a 5% (or maybe higher?) bonus to all the various styles and Double Strike

TWF it would grant a 5+% bonus to Off Hand Attack proc
THF it would grant a 5+% bonus to Glancing Blows proc*
S&B it would grant a 5+% bonus to Shield Bash/Block procs*
Ranged it would grant a 5+% bonus to Many Shot proc*

*read on to answer the question "What the hell is Aesop talking about?"


It has been hinted that Two Handed Fighting may soon go to a system similar to what is proposed for TWF. The current thought, that seems to stem from a desire to reduce the effectiveness of the twitch style of play, is to remove Glancing Blows while moving. This could pose great disadvantages to the THF as that will reduce the DPS of THF significantly (so sayeth the big THF proponents)

My suggestion on this is to take THF into the realm of Triggered effects and have Movement REDUCE, but not eliminate the number of Glancing Blows. Currently movement means getting glancing blows on half of the attacks but it is a set attack. Perhaps a 20% reduction in Triggered effects while moving would help reduce the difference between Twitch and non twitch enough that twitch was still useful but not to the point of making non twitch useless. (I know wishful thinking to actually find a balance everyone could respect). You could also make the Feat Spring Attack reduce this movement penalty by say half.

Additionally Attack Sequence could be used to determine modifiers as well. For example, if the first attack had a penalty to Triggered effects while later attacks had a bonus to triggered effects then twitch play, while still dominant (as befits the greater effort), would not quite so greatly dominate the style.

The Attack Sequence currently grants a bonus to attack rolls. First and Second Attack Rolls in the Sequeence have a bonus of zero, the third attack in the sequence has a bonus of +5 and the fourth has a bonus of +10. What if, for determining the triggered effects, the first attack in sequence had a -5% penalty, the second had no modifier, the third had a 5% bonus and the fourth had a 10% bonus to rate of triggered effects.

This would have the effect of Twitch play having a chance of triggering Glancing Blows 25% lower on the first swing and 20% lower on the second swing. If they had Spring Attack they'd only have a 15% penalty on the first swing and a 10% on the second.

These ideas of Penalties and Bonuses based on Attack Sequence and Movement could be used accross the board of Triggered Styles Effects.

Another style that could make good use of Triggered Style Effects... likely more so than the first two is:

Weapon and Shield Combat

Weapon and shield combat is "supposed" to be offense and defense in harmonious balance. Currently the offense is lacking and the defense is only rarely of actual use. The problem defensively is that you must stop attacking to be able to truly use defense. Insread make use of Triggered Style Effects to use that shield without having to actively use them.

Shield Bashing is something that you do uring a combat not something that you get ready to do and think about. You read an opening and swing with the shield. Additionally you don't wait for an attack with your shield over your head you anticipate an attack and while swinging you bring up a shield to defend your self.

Give shield use a 20% chance to either Block an incoming attack (ie apply Blocking DR) or to Bash after a successful Main Hand Attack (similar to twf). Improved Shield Bash could grant a bonus 20% chance to bash away at your enemy. While the Shield Mastery and Improved Shield Mastery Feats could grant additional % chance to block incoming attacks.

this would bring the defense of Shield Fighting back into the game and add a little more DPS to the style that is near the bottom of efficiency.

Speaking of Bottom of Efficiency. Ranged Combat has some issues as well. Its too slow and its big gun only operates 20 sec (less if you include the start up time) and has a 2 min cooldown as well. instead how about turningthat ability (Many Shot) into a triggered effect.

Say once you get Many Shot you have a 15% chance to launch an extra arrow in your attack. once you get your next extra arrow (3 total) this chance increases as wellto say 30%. When you get your fourth and final arrow you have a total of 40% chance to launch upto 4 arrows at a time. Perhaps the Ranger Capstone could add 10% to that chance while the Fighter Capstone adds 5%. Numbers can be changed of course, butthis was just the base idea. The Movement penalties could certainly play a roll in this... and make kiting a little less attractive I suppose.


I think I lost track of what I was saying

Aesop

[SteeleTrueheart]

Still catching up through the thread, but a few comments:

1) I've been convinced that STWF is not a benefit to the game. Instead of opening up new options, its presence instead would close many.

Then there will be some other way that non rangers will be able to get 100% off hand attacks?