THF vs SWF vs TWF

[Khurse]

or is that WT F? Ah I kill myself...

Anyway planning on coming back yet again, and saw the proposed changes to the ED's, but still no fix/changes to the basic weapon styles.

So question- I hate the TWF graphics, I'm not a huge fan of the other ones, but I can live with them. Now the Guild I run with (unless they've kicked me out, I probably should check in..) is pretty good for running endgame on high difficulty.

So- as it currently stands, I know that TWF is still king for single target (and probably AOE king in some circumstances..) but given proper equipment, are SWF and/or THF close enough to be useful? last time I tried a THF toon the DPS was...unfortunate. But I'm wondering if I actually gear correctly and pay attention , are they close? Or is TWF still far enough ahead that it's really just silly to not use it.

Class would probably be fighter or Paladin, but I understand Paladins are still in a bad spot right now, although KotC will be looked at..someday.

[droid327]

or is that WT F? Ah I kill myself...

Anyway planning on coming back yet again, and saw the proposed changes to the ED's, but still no fix/changes to the basic weapon styles.

So question- I hate the TWF graphics, I'm not a huge fan of the other ones, but I can live with them. Now the Guild I run with (unless they've kicked me out, I probably should check in..) is pretty good for running endgame on high difficulty.

So- as it currently stands, I know that TWF is still king for single target (and probably AOE king in some circumstances..) but given proper equipment, are SWF and/or THF close enough to be useful? last time I tried a THF toon the DPS was...unfortunate. But I'm wondering if I actually gear correctly and pay attention , are they close? Or is TWF still far enough ahead that it's really just silly to not use it.

Class would probably be fighter or Paladin, but I understand Paladins are still in a bad spot right now, although KotC will be looked at..someday.

I think the only competitive thf build is a maul wolf, which only becomes competitive by bypassing the thf mechanic entirely

Swf has some options, swashbuckler and such, vkf variations.

Don't forget ranged too. Gxb and Dxb builds have a lot of rolespace right now.

[Kaboom2112]

With fighter you can go both THFing and TWFing, you have the feat. TWFing on bosses, THFing for trash.

It's okay.

Not as strong as a wolf.

[Hawkwier]

Really think they missed a trick on the ED pass.

Could have fixed adrenaline cleaves. For THF only would prefer an interesting option.

Could have upped glancing blow damage. Or made it not scale with reaper.

Something to give THF a fighting chance..

But no. Nothing improves. THF remains sub optimal.

Pity.

[droid327]

Really think they missed a trick on the ED pass.

Could have fixed adrenaline cleaves. For THF only would prefer an interesting option.

Could have upped glancing blow damage. Or made it not scale with reaper.

Something to give THF a fighting chance..

But no. Nothing improves. THF remains sub optimal.

Pity.

They specifically said that they were not going to try and fix THF through the ED system, nor create a situation where THF was pigeonholed into one destiny to be viable. And I agree with that...THF needs to be actually fixed, not simply compensated for.

[korgzz_bloodaxe]

I think the only competitive thf build is a maul wolf, which only becomes competitive by bypassing the thf mechanic entirely

Swf has some options, swashbuckler and such, vkf variations.

Don't forget ranged too. Gxb and Dxb builds have a lot of rolespace right now.

If you by competitive you mean able to solo endgame high reaper then yes. Meanwhile my THF barbarian non-maul non-wolf build has no problem soloing sharn up to around r4-5. This is with minimal past lives (less than 30) and an average amount of reaper points.

[TedSandyman]

If you by competitive you mean able to solo endgame high reaper then yes. Meanwhile my THF barbarian non-maul non-wolf build has no problem soloing sharn up to around r4-5. This is with minimal past lives (less than 30) and an average amount of reaper points.

I agree with this. A lot of times, when someone says "broke" in the forums, what they really mean is something else is better, so why bother.

THF is decent, especially at lower levels. You can make some good builds that are fun to play if you like melee.

There are some pretty good weapons out there that aren't too hard to get.

But if you only care about optimal DPS or running with people who only care about optimal DPS or running the absolute hardest levels of content, then THF isn't for you.

But if you are running low to moderate reaper and tr-ing a lot, a THF barbarian is probably one of the easiest classes to play and get descent DPS from.

[droid327]

If you by competitive you mean able to solo endgame high reaper then yes. Meanwhile my THF barbarian non-maul non-wolf build has no problem soloing sharn up to around r4-5. This is with minimal past lives (less than 30) and an average amount of reaper points.

You disagree with my threshold for viable....I really disagree with your threshold for "minimal" past lives, if you think 30 is just a smattering lol

[HungarianRhapsody]

Hey, since this is a discussion of THF/SWF/TWF, could someone please post a video and a short explanation of twitching and what works/doesn't work wiht twitch fighting and THF?

Especially if there's anything that Wolves and Bears can/should do with THF twitch fighting since Wolf is the THF top of the heap at the moment (AFAIK).

[Grailhawk]

Hey, since this is a discussion of THF/SWF/TWF, could someone please post a video and a short explanation of twitching and what works/doesn't work wiht twitch fighting and THF?

Especially if there's anything that Wolves and Bears can/should do with THF twitch fighting since Wolf is the THF top of the heap at the moment (AFAIK).

https://www.youtube.com/watch?v=gWwZ8s6IOlU

Not my video but pretty good, Wolf/Bear don't use the THF animations so you don't get as much out of twitching making it not worth the effort.

[Bjond]

I know that TWF is still king for single target (and probably AOE king in some circumstances..) but given proper equipment, are SWF and/or THF close enough to be useful?

Basic "napkin math" puts it as SWF > TWF > THF. This is primarily due to MH doublestrike combined with the 1.5x Attribute bonus for SWF. IIRC, around 42% DS, SWF=TWF. Past that, SWF pulls further ahead. So, in heroics and most of epics, they're about the same, but 29+ it's SWF. This calc is only for basic TWF. I haven't calc'd it out for Monks and T5 Tempests, but intuition on those says they'd beat SWF; so, if you're not playing one of those and are trying to decide between TWF & SWF, go SWF.

However, anyone that's played with a good monk or tempest is definitely aware of how much more DPS and AE they can do than a good SWF, too. This is another of those situations where a good parser would answer the question. There's only so much you can consider all at once when running the numbers by hand.

[Cantor]

minimal past lives (less than 30)

This one made me chuckle. If something could drive away new players it would be this.

[palladin9479]

Basic "napkin math" puts it as SWF > TWF > THF. This is primarily due to MH doublestrike combined with the 1.5x Attribute bonus for SWF. IIRC, around 42% DS, SWF=TWF. Past that, SWF pulls further ahead. So, in heroics and most of epics, they're about the same, but 29+ it's SWF. This calc is only for basic TWF. I haven't calc'd it out for Monks and T5 Tempests, but intuition on those says they'd beat SWF; so, if you're not playing one of those and are trying to decide between TWF & SWF, go SWF.

However, anyone that's played with a good monk or tempest is definitely aware of how much more DPS and AE they can do than a good SWF, too. This is another of those situations where a good parser would answer the question. There's only so much you can consider all at once when running the numbers by hand.

I know this is a bit older of a thread but I figured this would be the best place to explain this. I've just finished a bunch of 20 Fighter eTR's where I played as THF, TWF, S&B Vanguard and SWF and I can say categorically SWF beats all of them followed by S&B, TWF then THF. The reason is the combat speed bonus and how haste percentages work in the majority of MMO's.

When we, as humans, think "Haste +30%" we imagine an increase in attack speed cause that is what the description says, while mechanically it's actually a decrease in attack delay. Directly increasing attack speed is actually kinda hard code wise because we start with the result and have to program backwards to get the modification. Decreasing attack delay is easy as it's just a simple subtraction. So what happens is that practically every MMO with a "% speed increase" value is a delay reduction while the ones with a "+integer attack speed" is an actual increase. Thing "Haste +10%" vs "Haste +100", DDO vs WoW. The integer can be added to an existing value that's then used to derive attack speed and could be any percentage value as a result. DDO's having different delays on the different attacks makes a simple delay reduction calc almost required.

To show case, assuming 100/100 is the base attack speed.

"30% speed increase"
(100 * 1.30)/100 = 130/100 = 1.30 = 30% increase

"30% delay reduction"
100/(100-30) = 100/70 = 1.4285 = 42.85% increase

This gets even more bananas when we start stacking, say a 15% enhancement increase with the 30% combat style bonus.

"45% speed increase"
(100 * 1.45)/100 = 145/100 = 1.45 = 45% increase
1.45/1.30 = 1.115, or 11.5% faster for a +15% increase, this is diminishing returns.

"45% delay reduction"
100/(100-45) = 100/55 = 1.8181 = 81% increase
1.8181/1.4285 = 1.2727, or 27.2% faster for a +15% increase, this is increasing returns.

Now lets add in AB Haste +30% to really see craziness.

"75% speed increase"
(100 * 1.75)/100 = 175/100 = 1.75 = 75% increase
1.75/1.45 = 1.2068, 20.6% faster for +30% increase, that diminishing returns starting to hurt now.

"75% delay reduction", hold on cause this gets stupid
100/(100-75) = 100/25 = 4.0 = 300% increase over base
4.0/1.8181 = 2.2, 120% faster for +30% increase

Now I think there might be some sort of hard limit to how low the delay between attack animations can get because at 100% "haste" you would end up dividing by zero and crashing the game. I would need some sort of frame capture to test what this limit is but the increasing returns is extremely easy to notice on SWF and S&B builds. Your attack rounds become a blur and all you see is a huge wall of numbers jump up on the screen and when you have 21 AB haste charges you can just keep that mode going non-stop.

THF winter wolfs +30% style bonus is the reason it's so powerful, it's basically the SWF mechanic for THF builds.

Both TWF and THF builds need some sort of innate combat style bonus to be competitive right now because they are being crushed by other builds raw number of hits, which is what DPS really comes down to, hits and crits.

[SpartanKiller13]

I know this is a bit older of a thread but I figured this would be the best place to explain this. I've just finished a bunch of 20 Fighter eTR's where I played as THF, TWF, S&B Vanguard and SWF and I can say categorically SWF beats all of them followed by S&B, TWF then THF. The reason is the combat speed bonus and how haste percentages work in the majority of MMO's.

When we, as humans, think "Haste +30%" we imagine an increase in attack speed cause that is what the description says, while mechanically it's actually a decrease in attack delay.

Can anyone confirm this? Would explain a few things, but it's a strong claim to make without any hits/second numbers listed here. Although I haven't seen any attack speed testing recently, and a number of animations and styles have changed since like Vanshilar's.

I always figured AS bonuses were really strong because they represented a boost to something that's hard to get other boosts to (like square law). Like how a +30 MP boost is great at level 4, but at level 30 with 250 MP it's a lot less significant.

If what you're saying is correct, AS bonuses are by far the strongest DPS boosts available, rather than just a fairly strong one.

[droid327]

at 100% "haste" you would end up dividing by zero

This caught my eye as a common error when calculating game mechanics. a 100% boost to speed means you're attacking at 200% base speed, or 50% delay. I think that highlights why you may be overexaggerating the effects of Haste boosts.

If you want to express a percentile boost in terms of delay, then its 100/100+boost. So a 50% boost would be a decrease in delay to 100/150=.667 of normal delay, not 50%. A 100% boost would be a decrease to 100/200 = 50%.

Otherwise, as you point out, at 100% you'd be dividing by zero, and above 100% you'd actually be attacking BEFORE you clicked the button, and naturally that's not how time works

[Grailhawk]

Can anyone confirm this? Would explain a few things, but it's a strong claim to make without any hits/second numbers listed here. Although I haven't seen any attack speed testing recently, and a number of animations and styles have changed since like Vanshilar's.

I always figured AS bonuses were really strong because they represented a boost to something that's hard to get other boosts to (like square law). Like how a +30 MP boost is great at level 4, but at level 30 with 250 MP it's a lot less significant.

If what you're saying is correct, AS bonuses are by far the strongest DPS boosts available, rather than just a fairly strong one.

This caught my eye as a common error when calculating game mechanics. a 100% boost to speed means you're attacking at 200% base speed, or 50% delay. I think that highlights why you may be overexaggerating the effects of Haste boosts.

If you want to express a percentile boost in terms of delay, then its 100/100+boost. So a 50% boost would be a decrease in delay to 100/150=.667 of normal delay, not 50%. A 100% boost would be a decrease to 100/200 = 50%.

Otherwise, as you point out, at 100% you'd be dividing by zero, and above 100% you'd actually be attacking BEFORE you clicked the button, and naturally that's not how time works

So the hypothesis is that a 45% increase to Alacrity is a 81% increase to the number of attacks.

If this is true we should be able to prove it from the data gathered by Vanshilar's his data actually measures attack rate (swing speed) per minute at different values of alacrity.

From Vanshilar's table we know that in 2010 at 0% alacrity at TWF (which Mr. Cow later in the thread states has the same speed as SWF) there are 86.7 swings per minute and at 45% alacrity its 133.3. 133.3/86.7 = 1.5374 so its only a 53.7% increase.

What I conclude from this is palladin9479 is not right in the magnitude of his hypothesis but might be right about alacrity being a delay reduction, at this time Vanishilar's data is the most accurate we have that I know of.

[palladin9479]

So the hypothesis is that a 45% increase to Alacrity is a 81% increase to the number of attacks.

If this is true we should be able to prove it from the data gathered by Vanshilar's his data actually measures attack rate (swing speed) per minute at different values of alacrity.

From Vanshilar's table we know that in 2010 at 0% alacrity at TWF (which Mr. Cow later in the thread states has the same speed as SWF) there are 86.7 swings per minute and at 45% alacrity its 133.3. 133.3/86.7 = 1.5374 so its only a 53.7% increase.

What I conclude from this is palladin9479 is not right in the magnitude of his hypothesis but might be right about alacrity being a delay reduction, at this time Vanishilar's data is the most accurate we have that I know of.

The problem is they were testing for the wrong thing by measuring number of swings instead of the time in between each swing of the four swing combo, so the data could only fit a certain way (additive linear). I really wish they had done the time in-between swings on a decent computer averaged over several attack rounds, then we measure the chance in that time at various haste values to get the real formula that's being used.

This game is run on a computer, meaning an "attack speed increase" is the exact same as "swing delay reduction" with the question being how it's being applied, and in that thread they assumed the answer instead of testing for it which is how their numbers don't line up well. With a 45% cap (of that era) the difference between a additive linear model and a multiplicative curved model isn't that big, it's not till past 50% that the difference gets bonkers big. It's also assuming the developers took the steps back when the level cap was 12 to code for the possibility of balanced haste values beyond 15% since the difference between models at 15% is practically non-existent.

Both linear models fall short of the observed value at 45%
(1.15 + 1.30) = 1.45
(1.15 * 1.30) = 1.495

Passing the base delay reduced value directly into the combat calc is much faster then first having to back calculate the value that needs to be passed based on a percentage.

Basically this
RoundDelay = BaseDelay * (HasteValue)

Instead of
DelayMultiplier = (100/HasteValue)
RoundDelay = BaseDelay * (DelayReduction)

Lets say 4 attacks at 1.0 seconds each (just example). We want them to be "30% faster", so we need to find the new value for the 1s.

Directly passing it in as
NewDelay = 1.0 * ((100-30)/100) = 0.7s
so it's four attacks at 0.7s each.

The other way
DelayMultiplier = (100/(100 +30)) = 0.7692
NewDelay = 1.0 * (0.7692) = 0.7692s
so four attacks at 0.7692s each.

The difference is small but has dramatic effects when scaled up.

So they would have to transform the "30% Haste" into the 0.7692 multiplier for each attack swing as it has to accept arbitrary values. Multiplications and especially Divisions are expensive CPU wise which is why they are normally avoided in favor of lookup tables or bit shifting. It's why I'm suspicious of any game that lists "+ percentage" to something like "attack speed" because it's non-intuitive to add a negative percentage to something. Then we got the static BAB reduction to delay, which could be applied before or after the haste reduction.

Having just finished using all these different combat styles, I can see definitely that S&B and SWF are much faster then both THF and TWF, like stupidly faster. Both are using the exact same gear with only minor tweaks for offhand. My previous explanation was just a demonstration on the two models start at similar values but diverge dramatically once scaled past 50% which can have serious balance consequences. Likely there is currently some in-between situation, because I don't see a doubling of attacks whenever I hit AB Haste on SWF but it's more then it should be and if there wasn't a larger then expected haste impact then TWF *should* be performing better in comparison.

[Fedora1]

If you by competitive you mean able to solo endgame high reaper then yes. Meanwhile my THF barbarian non-maul non-wolf build has no problem soloing sharn up to around r4-5. This is with minimal past lives (less than 30) and an average amount of reaper points.

But if you are running low to moderate reaper and tr-ing a lot, a THF barbarian is probably one of the easiest classes to play and get descent DPS from.

Agreed. While my favorite and best performing melee build so far has been an Aasimar wisdom based TWF ranger using falconry, I also had a good time with my pure barbarian THF horc w/ blood strength. Soloing low reaper all through heroics (three times) was easy breezy. Since it was only for the horc racial past lives I did not run this in epics, so can't speak to that.

EDIT: By comparison, my SWF swash (bard/fighter/rogue) kills single targets super fast in the same content, but it still takes longer to take out the trash when there are a lot of them, and being squishy it doesn't take too many hits before he's doing 0 dps. In 3 heroic lives of running a pure barb horc I think I had 2 deaths total. So far, my PDK (started at L15) swash has died 3 times and I am only just now holding L17.

[palladin9479]

This caught my eye as a common error when calculating game mechanics. a 100% boost to speed means you're attacking at 200% base speed, or 50% delay. I think that highlights why you may be overexaggerating the effects of Haste boosts.

If you want to express a percentile boost in terms of delay, then its 100/100+boost. So a 50% boost would be a decrease in delay to 100/150=.667 of normal delay, not 50%. A 100% boost would be a decrease to 100/200 = 50%.

Otherwise, as you point out, at 100% you'd be dividing by zero, and above 100% you'd actually be attacking BEFORE you clicked the button, and naturally that's not how time works

That was more about how developers can be lazy and use a direct reduction by converting the percentage into a negative addition to the bottom of a formula instead of correctly finding the reduction multiple by adding it to the bottom of division first. At lower values the difference is almost non-existent and the first is both computationally faster and programatically easier to implement in the combat code. I've played many MMO's and this is a common theme, initial "Haste" values are small and limited, much later due to power creep those values grow larger and suddenly a +50% Haste value equates into a nearly+100% DPS result (100/(100-50)). One or two builds dominate everything until a refactoring / rebalancing happens that corrects the situation.

[Grailhawk]

The problem is they were testing for the wrong thing by measuring number of swings instead of the time in between each swing of the four swing combo, so the data could only fit a certain way (additive linear). I really wish they had done the time in-between swings on a decent computer averaged over several attack rounds, then we measure the chance in that time at various haste values to get the real formula that's being used.

This game is run on a computer, meaning an "attack speed increase" is the exact same as "swing delay reduction" with the question being how it's being applied, and in that thread they assumed the answer instead of testing for it which is how their numbers don't line up well. With a 45% cap (of that era) the difference between a additive linear model and a multiplicative curved model isn't that big, it's not till past 50% that the difference gets bonkers big. It's also assuming the developers took the steps back when the level cap was 12 to code for the possibility of balanced haste values beyond 15% since the difference between models at 15% is practically non-existent.

Both linear models fall short of the observed value at 45%
(1.15 + 1.30) = 1.45
(1.15 * 1.30) = 1.495

Passing the base delay reduced value directly into the combat calc is much faster then first having to back calculate the value that needs to be passed based on a percentage.

Basically this
RoundDelay = BaseDelay * (HasteValue)

Instead of
DelayMultiplier = (100/HasteValue)
RoundDelay = BaseDelay * (DelayReduction)

Lets say 4 attacks at 1.0 seconds each (just example). We want them to be "30% faster", so we need to find the new value for the 1s.

Directly passing it in as
NewDelay = 1.0 * ((100-30)/100) = 0.7s
so it's four attacks at 0.7s each.

The other way
DelayMultiplier = (100/(100 +30)) = 0.7692
NewDelay = 1.0 * (0.7692) = 0.7692s
so four attacks at 0.7692s each.

The difference is small but has dramatic effects when scaled up.

So they would have to transform the "30% Haste" into the 0.7692 multiplier for each attack swing as it has to accept arbitrary values. Multiplications and especially Divisions are expensive CPU wise which is why they are normally avoided in favor of lookup tables or bit shifting. It's why I'm suspicious of any game that lists "+ percentage" to something like "attack speed" because it's non-intuitive to add a negative percentage to something. Then we got the static BAB reduction to delay, which could be applied before or after the haste reduction.

If you look at Vanshilar work there is a version where we can see values greater then 45%, we can see 55% when looking at his work with tempest (before U5 tempest had a 10% alacrity bonus it should have been removed by the time of Vanshilar's tests but it was bugged to still be in the game) we know from his data that 86.7 SPM is base we see that a Tempest is at base 97.1 and at Haste + Action boost is at 143.5.

143.5/86.7 = 1.655 or 65.5% increase
per your theory it should
100/(100-55) = 2.22 or 1.22% increase

Vanshilar's data may not be perfect but it is enough to give us an idea of what is going on, clearly they are not directly using the Alacrity% value as the delay reduction. The actual data on SPM is not anywhere near approaching that magnitude.

Having just finished using all these different combat styles, I can see definitely that S&B and SWF are much faster then both THF and TWF, like stupidly faster. Both are using the exact same gear with only minor tweaks for offhand. My previous explanation was just a demonstration on the two models start at similar values but diverge dramatically once scaled past 50% which can have serious balance consequences. Likely there is currently some in-between situation, because I don't see a doubling of attacks whenever I hit AB Haste on SWF but it's more then it should be and if there wasn't a larger then expected haste impact then TWF *should* be performing better in comparison.

The math that Bjond provided already show that SWF should be faster attacking than TWF for a fighter (and that doesn't need to invalidate Vanshilar's data to be true), further more the Kensei tree doesn't really do any thing to make one style better than the other so I can absolutely believe SWF Fighter is better then TWF Fighter. Similarly I could see a S&B Fighter preform better then a TWF Fighter on the bases of having 2 DPS trees to get stuff out of vs having only one when TWF.

All this said I'm not saying your 100% wrong but you need more data (specifically something that disputes Vanshilar's data) to make the claims your making true.