FFXIclopedia
Register
Advertisement
      • please research and post... Haste does not reduce cast time in the least! ONLY recast timer.

Fast Cast Affects both Cast time and Recast Timer.

Test:

I'm a RDM and I did some solo tests while fighting lizards in the jungle. I noticed no difference in casting times. On one of about 10 tests, there was a slight reduction in recast time, but that my havee been a timing accident.

I'm not fully convinced that haste does anything for spells, but I'd like to some tests on longer spells before I say for sure. --Kraftlos 04:16, 22 January 2007 (EST)


From my testing as BRD75/WHM:

Effects Mazurka/Reraise Recast Stoneskin Recast
Base 60 30
1% haste 59 29
2% haste 58 29
3% haste 58 29
4% haste 57 28
5% haste 57 28
6% haste 56 28
7% haste 55 27
8% haste 55 27
9% haste 54 27
10% haste 54 27
11% haste 53 26
12% haste 52 26
13% haste 52 26
14% haste 51 25
15% haste 51 25
16% haste 50 25
17% haste 50 25
18% haste 49 24
19% haste 48 24
20% haste 48 24
21% haste 47 23
22% haste 47 23
23% haste 46 23

Equipment for 23%: Rune Chopper, Angel Lyre, Walahra Turban, Sha'ir Manteel, Swift Belt, Sha'ir Crackows. Just need to borrow someone's Dusk Gloves to hit the reported 25% cap.

I should go back and test Utsusemi: Ni recast times with NIN sub to get the numbers for a 45 sec spell.

Note: By simple calculation, 17% Haste should give 49 sec recast, and 22% Haste should give 46, but they don't, so it's more complicated than a simple percentage reduction. For 17% I tried both Rune Chopper, Walahra Turban, Angel Lyre, Sha'ir Crackows (9+5+2+1) and Rune Chopper, Patroclus's Helm, Sha'ir Manteel, Swift Belt (9+2+2+4), with the same result.

I suspect that this means that the bonuses from the various pieces of armor are applied separately: 60*0.91*0.95*0.98*0.99 works out to 50.34, which would be truncated to 50, as observed. So let's check that:

Effects Mazurka Recast (Observed) Calculation
Base 60 60
1% haste 59 60*0.99=59.40
2% haste 58 60*0.98=58.80
3% haste 58 60*0.97=58.20
4% haste 57 60*0.96=57.60
5% haste 57 60*0.95=57.00
6% haste 56 60*0.94=56.40
7% haste 55 60*0.93=55.80, 60*0.98*0.98*0.98*0.99=55.90
8% haste 55 60*0.92=55.20
9% haste 54 60*0.91=54.60
10% haste 54 60*0.90=54
11% haste 53 60*0.89=53.40
12% haste 52 60*0.88=52.80, 60*0.95*0.96*0.98*0.99=53.09
13% haste 52 60*0.87=52.20
14% haste 51 60*0.86=51.60
15% haste 51 60*0.85=51.00
16% haste 50 60*0.84=50.40
17% haste 50 60*0.83=49.80, 60*0.91*0.95*0.98*0.99=50.32
18% haste 49 60*0.82=49.20
19% haste 48 60*0.81=48.60, 60*0.91*0.95*0.96*0.99=49.30
20% haste 48 60*0.80=48.00, 60*0.91*0.95*0.96*0.98=48.80
21% haste 47 60*0.79=47.40, 60*0.91*0.95*0.96*0.98*0.99=48.31
22% haste 47 60*0.78=46.80, 60*0.91*0.95*0.96*0.98*0.98=47.82
23% haste 46 60*0.77=46.20, 60*0.91*0.95*0.96*0.98*0.98*0.98=47.34

OK, guess that theory doesn't work either; it breaks down at 12% and again at 19, 20, 21 and 23%. Seems something more complicated is going on. --Valyana 17:26, 7 March 2007 (EST)

Haste is calculated in 256ths, not 100ths. Similar to damage reduction gear.

Here's theoretically derived "actual haste" values Rune Chopper: 23/256
Angel Lyre: 5/256
Walahra Turban: 12/256
Sha'ir Manteel: 5/256
Swift Belt: 10/256
Sha'ir Crackows: 2/256

Your "12% haste" was crackows (2) swift (10) turban (12) and manteel (5) which adds up to 29/256 reduction, or a mazurka recast of floor(60 * 227/256) = 53. Compare this with a single item that gives 30/256 reduction (black belt), floor(60 * 226/256) = 52.

--Aurikasura 18:30, 7 March 2007 (EST)

It's likely that the game is using fixed-point numbers to represent percentages instead of floating-point numbers for a trade of speed vs precision. It's probably using 8-bits to represent decimals which means each +haste % gear may be off by up to 0.00390625 (1/256) or 0.39%.

--JKL 18:36, 7 March 2007 (EST)

I just read the entire alla thread, looks like they've already been to the "everything in 256ths" theory and supposedly found some data which contradicts expectations generated using that theory. However I'd double-check base assumptions before tossing the theory...like for example when it was found that defending ring is actually 26/256 dmg reduction instead of 25/256.

Given that an apparent contradiction is lyre+turban (7% "listed", predicted 12+5=17/256-->56.01~sec recast, but in actual testing gives 55) I would try to double-check turban and lyre individually to see if they are mislabelled like d.ring was.

As for invoking Occam's Razor in the thread, treating everything as 256ths is actually the simplest explanation from a computer engineering perspective. --Aurikasura 18:51, 7 March 2007 (EST)


Perhaps in general the conversion to x/256 rounds to the nearest integer rather than taking the floor? 10% of 256 is 25.6, rounds up to 26. Similarly, 5% is 12.8, which would round up to 13. 2% is 5.12, rounds down to 5. 13+5=18/256->55.78 sec recast.

1% -> 2.56 -> 3/256
2% -> 5.12 -> 5/256
3% -> 7.67 -> 8/256
4% -> 10.24 -> 10/256
5% -> 12.80 -> 13/256
6% -> 15.35 -> 15/256
8% -> 20.48 -> 20/256
9% -> 23.03 -> 23/256
12% -> 30.72 -> 31/256

That way my composite 12% Haste (5+4+2+1 -> 13+10+5+3 = 31/256) works out the same as your BB 12% Haste. And indeed, my composite 12% Haste gives 52 second recast. As does 4+2+2+2+2 -> 10+5+5+5+5 = 30/256.

I also tested 2+2+2+1 and 3+4 for 7% Haste, and all gave 55 sec recast.

But this still doesn't explain the 17% number. Using the x/256 numbers here, 9+5+2+1 -> 44/256 and 9+4+2+2 -> 43/256, both of which still work out to a 49 second recast. You'd have to drop to 42/256 to get a 50 second recast time. But how can you do that without messing up another measured time? I also get 50 secs for 9+5+3 and 9+2+2+2+2.

So far I have been unable to find an example where how I arrive at a particular Haste number makes any difference, all the different combinations I try give the exact same result.

It just occurred to me that I could get better granularity using Hasso to extend casting times:

Hasso Mazurka Recast
Haste Expected Observed Components
0 90 90
1 89 89 1
2 88 88 2
3 87 87 3, 2+1
4 86 86 4, 2+2
17 74 75 9+5+3, 5+4+3+2+2+1
18 73 74 9+5+4, 5+4+3+2+2+2

...and that's all the testing I feel up to tonight. More numbers at 17% and 18% that don't fit the expected value for plain % haste, but again splitting up the Haste across as few or many pieces of armor as possible doesn't make a difference.

Does anyone have an example where it does make a difference? --Valyana 04:37, 3 April 2007 (EDT)


Perhaps they did something like they did with accuracy from weapon skill levels over 200? If you want to floor each value of the formula 'time * (1 - (haste * .98)),' each value fits your tests. -- Yen 13:11, 3 April 2007 (EDT)


I still think the right thing to do is to test individual pieces alone to determine the exact fraction on each piece. On dmg reduction items, every single one of them is rounded down except d.ring which is a very odd distinction. Who knows what metric they used for haste items. --Aurikasura 13:55, 3 April 2007 (EDT)


The problem is that it's impossible for me to test individual pieces with enough precision, as the longest recast I have to work with is 90 seconds. Perhaps a DRK could do some testing with Dread Spikes recast time; with Hasso I believe it rises to 270 seconds, which is nice and close to 256.

Hmm, actually I have access to a friend's DRK...I'll check this out later. --Valyana 17:32, 3 April 2007 (EDT)

...

OK:

Hasso Dread Spikes Recast
Haste Expected Observed x/256 Components
0 270 255 ??
5 256 255 ?? WT
6 253.8 254 15 SB+DL
7 251.1 251 18 WT+DL, SB+DG, QB+DL+DG
8 248.4 248 20 WT+DG
9 245.7 246 22 RC, WT+SB, WT+QB+DL, SB+DG+DL
10 243.0 243 25 WT+DG+DL, WT+DG+QB
11 240.3 240 28 RC+DL, WT+SB+DL
12 237.6 238 30 WT+SB+DG
13 234.9 235 33 RC+SB
14 232.2 233 35 WT+SB+DG+DL
15 229.5 230 37 RC+SB+DL

OK, from the 270 recast we can actually conclude exactly what x/256 factor would give the observed recast time, except for <6% Haste, where the recast is capped at 255 seconds.

But there's a problem. Between 7% and 9% Haste, adding 2% Haste via Quick Belt or Dusk Ledelsens is 4/256, but between 8% and 10% it's 5/256. And then between 9% and 11% Haste it's 6/256.

Which seems to suggest that Haste doesn't work by converting each piece to x/256 and adding them together like -damage gear. --Valyana 02:21, 4 April 2007 (EDT)

Apparently it works for /512 rather than /256: [1] --Valyana 17:59, 3 November 2007 (UTC)


While this has been determined elsewhere (BG had a thread on it, but the link is dead), there's no info on it here, so I figured I should note it for general reference.

First, haste is calcualted in fractions of 1024 (well, 512 also works, but for practical application 1024 is easier to use).

Second, any given haste % is internally represented as haste*10/1024. So, 1% == 10/1024, 2% = 20/1024, 3% = 30/1024, etc.

Third, all haste totals are additive. 3% in gear plus 8% in gear gives 11%, and is totalled as 30/1024 + 80/1024 = 110/1024.

Fourth, the formula for figuring out recast delay can be written as:

FLOOR(((1024-hasteFraction)*baseDelay)/1024)

So 12% haste applied to the 60 second recast of Reraise would be floor((1024-120) * 60 / 1024) = 52 seconds.

Aside from the apparent maximum value of 255 seconds for recast timers from the Dread Spikes test, this formula exactly matches the observed results of all info on this talk page.

The 43.75% cap on magical haste implies a cap of 448/1024 for magic haste. Gear haste and JA haste each cap at 256/1024. I'm not sure whether the overall haste cap of 80% is 800/1024 or 820/1024; I suspect the former.

--Motenten 04:52, December 3, 2009 (UTC)


Should Victory March and Advancing March be listed under haste? The only thing they speed up are melee attacks, not spells, and they're shown as a song effect. Just a thought. --Xalioniaf 21:30, 7 April 2007 (EDT)

Song covers all song effects, as far as a status effect goes. --Chrisjander 22:19, 7 April 2007 (EDT)

March improves spell recast times the same way as other sources of haste. --Valyana 23:46, 7 April 2007 (EDT)


I don't know if this is common knowledge but is ranged delay not affected by the haste/slow effect? I haven't tested it personally but the more I shoot while hasted/slowed it seems there is no affect either way. Wondering if this should be mentioned somewhere on the page. --Valient 00:19, 11 July 2007 (CDT)


The status effect given by the White Magic spell Haste has NEVER affected recast timers of spells while in weakness.

Examples from 2005:

Raise III in weakness without Haste

Raise III in weakness with Haste

--Teorem 18:52, 1 August 2007 (CDT)

- I'm sorry, but what you show, does not show anything proving or disproving your point. --Docstu 9:22, 24 August 2007 (CDT)

I beg to differ, you cannot just arbitrarily remove the information from the article just because you have different thoughts. If you are dead set on this, go experiment and bring back your results. Teorem took the initiative to show something, but all we have is you saying "No" without having anything to back it up. --User:Charitwo/Sig 10:37, 26 August 2007 (CDT)

As an avid player in Dynamis as WHM75, Thorem's information is correct. As an example, Raise 3 while weakened was 2 minutes recast with or without haste while weakened. If not weakened, Raise 3 is 1 minute and less than one minute if hasted. So the statement that haste has no effect while weakened is 100% true. --Wayka †Talk† 12:38, 26 August 2007 (CDT)

I get 2:00 recast on reraise while weakened and no haste and 1:51 while weakened with haste. —The preceding unsigned comment was added by Yoel (talkcontribs).

I duplicated this result in a quick test in Dynamis. 1:51 Raise III recast with Haste. --Alephnot 23:30, 26 February 2008 (UTC)
  • After being Raised Haste will only reduce the base time not the double affect caused by Weakness. For example with 5% Haste the recast of Aspir is approx 57 seconds down from 60 seconds. With Weakness and 5% Haste the recast is 1 minute 57 seconds. I am currently unsure if this affects melee delay in the same way —The preceding unsigned comment was added by Jabbathehutt (talkcontribs).

Weakened[]

Haste from equipment (swift belt, walahra turban, ...) lowers your recasts while weakened. It's easy to test by using BLU's Self-Destruct spell. One possible explanation for magical haste not working while weakened is because all the haste from magical sources are added up together and then capped at 100% Slow. Equipment haste, being in a different category, is not affected.

Hundred Fists[]

Would Hundred Fists fall under the same category as Hasso and Desperate Blows? Because Hundred Fists have a much higher haste effect than the caps of equiment (25%) and magical (~43%) haste, it can't belong to those two types.

However, DRKs can obtain up to ~93% haste and Hundred Fists is estimated to be around 70-75% melee haste but extra haste is said to have no effect during Hundred Fists. Maybe there is a minimum delay limit for melee swings? --JKL 16:56, 5 August 2007 (CDT)

Random effect in Caedarva Mire?[]

Can someone elaborate this?--VZX 17:19, 9 August 2007 (CDT)

If you walk through the swamps in Caedarva, you will randomly receive some sort of effect (Haste, Flee, etc.) --User:Charitwo/Sig 10:33, 26 August 2007 (CDT)

I've actually played around in the swamps a bit there, I've gotten flee effect probably 4 times (one time lasted about 2 minutes), about 7 or 8 times weight, and never any other effect, though I wouldn't discount other effects if someone else reports them. Also note that the random effect sometimes happens almost instantly, and other times takes up to even a minute to gain, also I've gotten a flee effect then a weight effect (message for weight), but the flee effect canceled out the weight portion, or rather blocked it from actually being applied. (if that helps anyone looking for a haste effect)--Solarhawk 09:57, 12 May 2008 (UTC)

New Haste Cap[]

So I remember reading something about a change/cap to Haste or some such thing in the most recent update: has anyone been doing any testing to figure out what this meant? --Vycron 12:42, 15 March 2008 (UTC)

I did a test of this today with Utsusemi: Ni in aht urghan with a rdm and a brd. this is kinda useless knowledge but. as my 75 ninja with no gear, haste from rdm brought Utsusemi: Ni recast down to 37 seconds which is 18%(17.778) haste. With march, march from bard it goes from 45 to 34 which is 24%(24.445) haste. With haste and march, march it went down from 45 to 28 seconds which is 38%(37.778) haste. With W. turban, Dusk gloves, swift belt and fuma sune-ate it brought it down to 21 seconds which is 53%(53.333) haste. We also tried with soul voice march march haste and it was still 21 seconds. So the haste cap is 53%(53.333) not 50,(or it is 50% rounded up) Mobisuoneffxi 08:43, 12 May 2008 (UTC) Mobiusone Hades I tested this again using the x/255 method and (you round down no matter what) i come up with 52.5490, or 52.5490% haste that was done by taking the haste for march march + haste (37.778) and multipling it by 2.55(100x2.55=255) and come up with 96.3339(96)((15x2.55)+38.25 from gear(38) =134 so 134/255=52.5490% so that theory doesnt hold true cause the recast would be 23.64 seconds but it is 21. so that doesnt work. it si done in a percentage. not x/255. Mobisuoneffxi 09:18, 12 May 2008 (UTC) mobiusone hades

JAs have their own Haste cap[]

Job abilities have a 25% haste cap, whereas Hasso and Last Resort (with DB) do not give 35%, but are capped to 25%. Seen here http://www.jpbutton.com/?p=831 can anyone confirm? --Ranzear 05:53, 7 October 2008 (UTC)

Advertisement