((149+FLOOR((432+192-540)*20/360))*2.1)+(.9 * 50)
((153)*2.1)+45=366.3
Relic hat is 50 tp per shot, 90% of the time.
Samurai Roll maxed is 40 for 11, 20 for +5 roll, and 10 for bonus that has 50% chance of being applied even if SAM is not in the party making 70. An additional 40 STP from gear becomes 110 STP, which is represented as 2.1 in the calc.
Power Rangers: A Guide To Pewing
Power Rangers: A Guide to Pewing
Verda said: »
((149+FLOOR((432+192-540)*20/360))*2.1)+(.9 * 50)
((153)*2.1)+45=366.3
Relic hat is 50 tp per shot, 90% of the time.
Samurai Roll maxed is 40 for 11, 20 for +5 roll, and 10 for bonus that has 50% chance of being applied even if SAM is not in the party making 70. An additional 40 STP from gear becomes 110 STP, which is represented as 2.1 in the calc.
((153)*2.1)+45=366.3
Relic hat is 50 tp per shot, 90% of the time.
Samurai Roll maxed is 40 for 11, 20 for +5 roll, and 10 for bonus that has 50% chance of being applied even if SAM is not in the party making 70. An additional 40 STP from gear becomes 110 STP, which is represented as 2.1 in the calc.
Ok awesome! Thanks for the clarification, I thought I was doing something wrong.
Phoenix.Darkspawn said: »
sorry for the delay response, yep you were right, was using melee tab /kick myself out of the server. sorry to give false alarm
Glad it's working for you :)
eliroo said: »
Ok awesome! Thanks for the clarification, I thought I was doing something wrong.
no problem glad I could help~
How does Apex arrow line up compared to Jishnu's?
This is just to sate my curiosity.
I noticed that Apex > Jish > Jish should make double light. Just wondering if Apex > Jish > Apex would be stronger since that will close Radiance.
This is just to sate my curiosity.
I noticed that Apex > Jish > Jish should make double light. Just wondering if Apex > Jish > Apex would be stronger since that will close Radiance.
It's got a high ws mod, that's all AGI, it is one hit so it benefits a lot from WSD, but not as much from fotia though that's still best to use for it. It's pretty much the same as the Last Stand set posted on the prior page, though you'd want to go for fern stones for 1 more wsd and 5 less AGI. The one thing I don't like about it is that it's a 1 hit WS, and multihit on ranged attacks doesn't happen unless you use a multihit ws. That means the 95% accuracy cap can hurt it for skillchain potential. Damage wise, even in situations that favor it (high def targets where pdif isn't capped) it will end up doing about 70% of the damage of jishnu's. Jishnu's can crit, and ranged crits work differently but they still are good in high def situations. So all in all, if you're using it to make a longer chain, yes it's worth it the damage loss over time isn't going to compare to the damage gain over time of more skillchains that do more damage because they are further in the chain, just watch out for that 95% accuracy cap on a 1 hit WS. Doing apex -> jishnu -> jishnu will result in 2/3rd of WS skillchain instead of 1/2, so a gain already even if you don't consider the fact that you're going to have a very nice skillchain damage boost on the 2nd jishnu's. Combine this with the fact it can play into aeonic aftermaths and it's very nice indeed, you need to be careful use that aftermath though or it might make the second jishnu's not chain as with AM up Apex Arrow is light property so would make double light from the first 2 ws. Ideally you'd want 2k tp for apex arrow, then the 2nd Jishnu would make radiance, AM would be consumed and then you could go again.
[+]
The description for Apex Arrow says it's four hits. I have never tested it. I assumed it was 3.0 + 3 x 1.0. Is the description wrong?
Apex is one hit.
SE is a little inconsistent with the -fold in English; poor translation I guess. -fold is supposed to just represent the damage multiplier (fTP) of the hit, generally when it's static. Sidewinder used to say "four-fold" too. In Japanese, it was corrected to "five-fold," but English doesn't seem to mention a number anymore.
Apex in Japanese says "three-fold." In comparison, Last Stand says "1-2 attack rounds based on remaining ammunition."
SE is a little inconsistent with the -fold in English; poor translation I guess. -fold is supposed to just represent the damage multiplier (fTP) of the hit, generally when it's static. Sidewinder used to say "four-fold" too. In Japanese, it was corrected to "five-fold," but English doesn't seem to mention a number anymore.
Apex in Japanese says "three-fold." In comparison, Last Stand says "1-2 attack rounds based on remaining ammunition."
[+]
Ya, what flippant said, though I didn't know about the translation differences.
In general if it doesn't say hits (and instead say something like X-fold or X damage), it means the damage multiplication from http://fTP. To my knowledge the only multihit ranged WS are Last Stand and Jishnu's, and (possibly) Blast Arrow and Blast Shot, which need some more testing done to really know quite a few things about them :/ They are also the two best physical damage options in part because of that and their interaction with fotia but they have much more going for them than that, especially Jishnu's for 1k spam due to it also being able to crit and having the most hits (3).
In general if it doesn't say hits (and instead say something like X-fold or X damage), it means the damage multiplication from http://fTP. To my knowledge the only multihit ranged WS are Last Stand and Jishnu's, and (possibly) Blast Arrow and Blast Shot, which need some more testing done to really know quite a few things about them :/ They are also the two best physical damage options in part because of that and their interaction with fotia but they have much more going for them than that, especially Jishnu's for 1k spam due to it also being able to crit and having the most hits (3).
Blast x2, Sidewinder/Slug, Numbing, Refulgent, and Apex are all x-fold in Japanese. But Sidewinder's help text was wrong for nine years...so entirely possible that there could be a mistake~.
The description doesn't say fold, it says hits.
Ah I haven't read the descriptions much in game I was going by the wiki.
Upon further inspection, it does say fold.
Hey guys I am just returning but had a quick refresher regarding set swapping for normal shots. I was searching the posts but not really getting the answer.
I know that pre-shot gear (snap shot ect) needs to be put up first followed by another set about a second later but what I'm confused about is seeing a stp set and acc sets being posted. which should I have for the arrow release? or do all the sets need to be used in some order? or is it based on the circumstances of the event?
I know that pre-shot gear (snap shot ect) needs to be put up first followed by another set about a second later but what I'm confused about is seeing a stp set and acc sets being posted. which should I have for the arrow release? or do all the sets need to be used in some order? or is it based on the circumstances of the event?
Enucleation said: »
Hey guys I am just returning but had a quick refresher regarding set swapping for normal shots. I was searching the posts but not really getting the answer.
I know that pre-shot gear (snap shot ect) needs to be put up first followed by another set about a second later but what I'm confused about is seeing a stp set and acc sets being posted. which should I have for the arrow release? or do all the sets need to be used in some order? or is it based on the circumstances of the event?
I know that pre-shot gear (snap shot ect) needs to be put up first followed by another set about a second later but what I'm confused about is seeing a stp set and acc sets being posted. which should I have for the arrow release? or do all the sets need to be used in some order? or is it based on the circumstances of the event?
It's based on event, NM, etc etc. if your r.acc is high normally, you can use an STP build, which makes your TP gain and WS rate fast. However, if you use an STP build, but a 50% r.acc rate, then your overall damage will suffer.
Hey thanks!
Hey everyone.
I'm new to playing Ranger, and had a question about Flurry. Is it doing something beyond providing snapshot? I should be snapshot capped, yet I'm pretty sure I'm shooting faster with Flurry up. Here's my current precast with Velocity Shot up:
I must be doing something wrong. Hope someone can help me out here.
I'm new to playing Ranger, and had a question about Flurry. Is it doing something beyond providing snapshot? I should be snapshot capped, yet I'm pretty sure I'm shooting faster with Flurry up. Here's my current precast with Velocity Shot up:
Code
Name Snapshot Rapid Shot Velocity Shot Snapshot
Head Artifact +1 14%
Body empy+1 22%
Hands Carmine Fin. Ga. 7% 10%
Back Belenus's 10% 2%
Waist Yemaya 5%
Legs Adhemar Kecks 9% 10%
Feet persuers 10%
Job Merits 10% 5%
Job Traits 30%
Velocity Shot 15%
V.Shot Gift I 5%
-----------------------------------------------------------------------
Totals 36% 84% 44%
w/ Velocity 80% 84%
I must be doing something wrong. Hope someone can help me out here.
I don't remember seeing testing done but I'm sure someone did it and I'm just ignorant. If it hasn't been done it would be worth testing.
It's still untested, and not well understood. I've been doing snapshot testing and about 40% done with it. But I haven't got to velocity shot testing either. What we know was taken from blue gartr player testing, the JP wiki, as well as logical assumptions. The 15% slow is easy to test, but the snapshot was harder, people used to think it was base 10% for snapshot etc, snapshot has never been easy to test. I'm all for seeing if perhaps velocity shot is another term, or free of caps. It is very hard to test snapshot on ranger though due to rapid shot procs, random latency, and just general inconsistencies. Anyone that can help testing I'm thankful for it's a lot of work.
As far as chocobro's question though, I should edit the wiki to be more clear. Velocity shot with empy +1 is 22% total, 7% from the body, 15% from the job ability itself, so you counted velocity shot's base twice.
As far as chocobro's question though, I should edit the wiki to be more clear. Velocity shot with empy +1 is 22% total, 7% from the body, 15% from the job ability itself, so you counted velocity shot's base twice.
I feel like this should be the goto preshot build.
ItemSet 345751
With Velocity Shot, augments, merits and gifts it comes out to 87 Rapid Shot and 72 Snap Shot. I don't consider capping Rapid Shot to be nearly as important as Snapshot as it doesn't have increasing marginal returns like Snapshot does. If you use Kustawi +1 you're at 90~94 Rapid Shot.
ItemSet 345751
With Velocity Shot, augments, merits and gifts it comes out to 87 Rapid Shot and 72 Snap Shot. I don't consider capping Rapid Shot to be nearly as important as Snapshot as it doesn't have increasing marginal returns like Snapshot does. If you use Kustawi +1 you're at 90~94 Rapid Shot.
Also that's assuming what we assume know about snapshot/rapid shot to be true.
Honestly, I wouldn't be shocked if Snapshot/Flurry was like gear Haste and Spell Haste, where they have individual caps, as well as total caps. And where Coursers Roll fits in to all this, who knows.
Fenrir.Snaps said: »
I feel like this should be the goto preshot build.
ItemSet 345751
With Velocity Shot, augments, merits and gifts it comes out to 87 Rapid Shot and 72 Snap Shot. I don't consider capping Rapid Shot to be nearly as important as Snapshot as it doesn't have increasing marginal returns like Snapshot does. If you use Kustawi +1 you're at 90~94 Rapid Shot.
ItemSet 345751
With Velocity Shot, augments, merits and gifts it comes out to 87 Rapid Shot and 72 Snap Shot. I don't consider capping Rapid Shot to be nearly as important as Snapshot as it doesn't have increasing marginal returns like Snapshot does. If you use Kustawi +1 you're at 90~94 Rapid Shot.
Did SE fix the issue with the Arcadian Jerkin? My RNG is still back in 2013 (Divine Might II era) and I'm updating it presently, but if I recall correctly, we used a different body for Snapshot/TP. Offhand I just don't remember what it was, but I do remember we avoided using Arcadian due to the stats literally being broken (and not the good broken).
Come to think of it, it might have been Sylvan body that was used for our precast. I'm not sure ilevel Empy (Amini) was released yet at the time of the RNG DM II spammage. If someone could elaborate on that I'd appreciate it.
Verda said: »
As far as chocobro's question though, I should edit the wiki to be more clear. Velocity shot with empy +1 is 22% total, 7% from the body, 15% from the job ability itself, so you counted velocity shot's base twice.
Okay. That 22% was pretty confusing. Glad to have that cleared up.
Thanks for the replies everyone.
Asura.Avallon said: »
Fenrir.Snaps said: »
I feel like this should be the goto preshot build.
ItemSet 345751
With Velocity Shot, augments, merits and gifts it comes out to 87 Rapid Shot and 72 Snap Shot. I don't consider capping Rapid Shot to be nearly as important as Snapshot as it doesn't have increasing marginal returns like Snapshot does. If you use Kustawi +1 you're at 90~94 Rapid Shot.
ItemSet 345751
With Velocity Shot, augments, merits and gifts it comes out to 87 Rapid Shot and 72 Snap Shot. I don't consider capping Rapid Shot to be nearly as important as Snapshot as it doesn't have increasing marginal returns like Snapshot does. If you use Kustawi +1 you're at 90~94 Rapid Shot.
Did SE fix the issue with the Arcadian Jerkin? My RNG is still back in 2013 (Divine Might II era) and I'm updating it presently, but if I recall correctly, we used a different body for Snapshot/TP. Offhand I just don't remember what it was, but I do remember we avoided using Arcadian due to the stats literally being broken (and not the good broken).
Come to think of it, it might have been Sylvan body that was used for our precast. I'm not sure ilevel Empy (Amini) was released yet at the time of the RNG DM II spammage. If someone could elaborate on that I'd appreciate it.
I do believe that it was the arcadian legs which were broken, not the body, where their snapshot did not seem to be working. Issues with the body are new to me.
Bismarck.Snprphnx said: »
Honestly, I wouldn't be shocked if Snapshot/Flurry was like gear Haste and Spell Haste, where they have individual caps, as well as total caps. And where Coursers Roll fits in to all this, who knows.
This has been said before and is on the wiki too but SE officially stated there's no categories and it caps at 70%.
Asura.Avallon said: »
Come to think of it, it might have been Sylvan body that was used for our precast
Fenrir.Snaps said: »
I feel like this should be the goto preshot build.
Code
sets.precast.RA = {
head="Orion Beret +1", --R:14
body="Amini Caban +1",--V:22
hands={ name="Carmine Fin. Ga. +1", augments={'Rng.Atk.+20','"Mag.Atk.Bns."+12','"Store TP"+6',}},--S:8 R:11
--waist="Yemaya Belt",--R:5
waist="Impulse Belt",--S:3
legs=ah_legs,--S:9 R:10
feet="Meghanada Jambeaux +1",--S:8
back=belenus, --V:2
--Kustawi+1 and Kustawi R:14
}
--Snapshot: 70
---- Merits:10
---- Gear: 22+8+9+8+3=50
---- Gifts: 10
--Rapidshot: 84
---- Merits:5
---- Traits:30
---- Gear: 14+11+10+14 = 49
sets.precast.RA.Flurry = {
head="Orion Beret +1", --R:14
body="Amini Caban +1",--V:22
hands={ name="Carmine Fin. Ga. +1", augments={'Rng.Atk.+20','"Mag.Atk.Bns."+12','"Store TP"+6',}},--S:8 R:11
waist="Yemaya Belt",--R:5
legs=ah_legs,--S:9 R:10
--feet="Meghanada Jambeaux +1",--S:8
feet=p_feet,--R:10
back=belenus, --V:2
--Kustawi+1 and Kustawi R:14
}
--Snapshot: 74
---- Merits:10
---- Gear: 22+9+8=39
---- Flurry1: 15
---- Gifts: 10
--Rapidshot: 94
---- Merits:5
---- Traits:30
---- Gear:14+11+5+10+5+14=59
sets.precast.RA.Flurry2 = {
head="Orion Beret +1", --R:14
body=arc_body,--R:12
hands={ name="Carmine Fin. Ga. +1", augments={'Rng.Atk.+20','"Mag.Atk.Bns."+12','"Store TP"+6',}},--S:8 R:11
--waist="Yemaya Belt",--R:5
waist="Impulse Belt",--S:3
legs=ah_legs,--S:9 R:10
feet=p_feet,--R:10
back=belenus, --V:2
--Kustawi+1 and Kustawi R:14
}
--Snapshot: 72
---- Merits:10
---- Gear: 8+9+2=19
---- Gifts: 10
---- Flurry 2: 30
--Rapidshot: 106
---- Merits:5
---- Traits:30
---- Gear:14+12+11+5+10+10+14=71
Yes that was a non-Flurry build. I haven't ruminated over gear enough for Flurry I/Flurry II. I'd also rate a Flurry I build as more useful than Flurry II as that is going to be much more common.
I don't think Carmine Finger Gauntlets +1 is asking for much. They're like 15m on my server. I don't think anybody who actually cares about their gear should have problems fronting such a small amount of gil. I also disagree with your tiers in regards to Ambuscade capes. Self cap Snapshot and nearly self capped Rapid Shot is amazing. You're basically freeing up another person's role or changing it (/RDM vs /SMN /WHM /SCH) to something that can end up being much more useful.
I don't think Carmine Finger Gauntlets +1 is asking for much. They're like 15m on my server. I don't think anybody who actually cares about their gear should have problems fronting such a small amount of gil. I also disagree with your tiers in regards to Ambuscade capes. Self cap Snapshot and nearly self capped Rapid Shot is amazing. You're basically freeing up another person's role or changing it (/RDM vs /SMN /WHM /SCH) to something that can end up being much more useful.
Also in regards to Snapshot testing, I'll try and find some time to test on Uraganites in the year future. I have a very simple addon that issues /shoot commands every .1 seconds. I can run that for a few hours with the build I posted + relic bow with/without Flurry II and see if it makes a difference.
Anyone have an up to date namas set? I realize the enmity is higher than coro, but already have a few bows so figured I'd give master trial a run with it and see how hate works out before dumping plutons into guns.
[+]
So are you saying Amini +1 pieces are no longer useful for pre-shot setups? I thought it was pretty amazing because of the bonuses from it.
I was asked to share the most recent version of my ranger lua, I also had to finish developing the flurry mode toggle. If flurry is active it will shoot with a flurry set, I or II depending on the toggle. If no flurry is active it uses the no flurry set.
RNG.lua: http://pastebin.com/8JNQH5VL
I also use this for my augmented items: http://pastebin.com/u64UXBXt
It is for my light armor jobs though so might have things you don't need in there.
I still haven't finished my flurry testing it's not fun or easy to do.
Two things I have found for certain, there isn't an outgoign packet for finishing ranged attacks. Your attack starts, a packet goes to SE, you get a packet back saying your ranged attack can start, then SE sends another packet saying when your ranged attack ends (there is no packet sent to SE that I have found that says your ranged attack ended). This would make sense since you want all that sort of stuff done server side as much as possible.
This is what I got through so far testing flurry on THF/RDM:
I still need to test extremely high packet delay and see how that affects flurry. What seems to be going on is the actual divisor is closer to 120, than 106 with about 116 being what I am able to record. Ruling rapidshot out is hugely helpful for testing flurry or snapshot, it does not however make packet delay and network spikes non existent. Samples with higher standard deviations are probably, but not necessarily, less reliable. There does also appear to be about a .4 second server update artifact making it harder to tell exactly where you are and this artificat could explain why lower delay weapons like Tsoabichi seem to get about 4% less benefit on average for 15 snapshot and failnaught (notice spelling it's one THF can use) and Culverin about 1% difference. These results are unfinished but I might never finish them if I don't kick my own butt and at least show the work I have done.
Things left to test:
1) Does snapshot really cap at 70?
- Have to make a 70 snapshot and over 70 snapshot set for THF, the most reliable way to do this is spend a lot on skirmish stones and combine it with flurry after I have solid flurry numbers, a less "good" way would be to just do it on ranger with tons of samples and only use the high delay numbers and try to distinguish those from just lag spikes (blech).
2) How much does your internet latency affect snapshot?
- Have to repeat tests I've done, using the worst VPN I can find to give very long responses from SE's server. This will also probably reveal some other interesting things about how it works.
3) Higher Samples are needed to be more conclusive, some of the results vary by 3% for the same tests. Testing in different zones might also be helpful, but any set of data needs to be collected on the same net around the same time as network latencies can even change if the network is on heavy load (like for me seems around 8pm at night when everyone connected is watching netflix or something).
4) Try to support or deny the .4 second artifact with data, find a way to quantify lag spikes as outliers then see if the variance is within the bounds of this .4 second artifact + latency.
I should note this is building on top of, rather than replacing testing done by others in the past, and it's always good to have more data to compare to. I have SQL files stored with the timings. Windower lets me do code when the ranged packet is received from SE and the ranged end packet is received from SE.
Usually it's more fun to just blow up stuff on one of my favorite jobs so I apologize if this is slow going.
RNG.lua: http://pastebin.com/8JNQH5VL
I also use this for my augmented items: http://pastebin.com/u64UXBXt
It is for my light armor jobs though so might have things you don't need in there.
I still haven't finished my flurry testing it's not fun or easy to do.
Two things I have found for certain, there isn't an outgoign packet for finishing ranged attacks. Your attack starts, a packet goes to SE, you get a packet back saying your ranged attack can start, then SE sends another packet saying when your ranged attack ends (there is no packet sent to SE that I have found that says your ranged attack ended). This would make sense since you want all that sort of stuff done server side as much as possible.
This is what I got through so far testing flurry on THF/RDM:
Code
Average shots sample size 10, internet connection #1 average delay to a pol server: 189ms Average std: 116.373 Culverin Raw 6603.77 std_dev: 69.11 Predicted: 760/106 = 7.1698 Divisor Actual: 115.08 Culverin Flurry I 5478.51 Standard Deviation: 124.56 (82.9% ~= 17.03%) Predicted: 760/106 *.85 = 6.09433 Predicted w/ 116 divisor: 5.5689 seconds Culverin Snapshot 15 5507.30 Standard Deviation: 127.59 (83.3% ~= 16.6%) Predicted: 760/106 *.85 = 6.09433 Predicted w/ 116 divisor: 5.5689 seconds Tsoabichi Raw 2471.353 std_dev: 62.69 Predicted: 288/106 = 2.716 seconds Divisor Actual: 116.53 Tsoabichi Flurry I 2244.73 std_dev: 179.04 (90.8% ~= 9.16%) Predicted: 288/106 *.85 = 2.3086 seconds Predicted w/ 116 divisor: 2.11 seconds Tsoabichi Snapshot 15 2165.73 std_dev: 135.25 (87.6% ~= 12.36%) Predicted: 288/106 *.85 = 2.3086 seconds Predicted w/ 116 divisor: 2.11 seconds Testing on higher delay internet, higher sample sizes (20): average delay to pol server: 224ms Average std 115.15 Tsoabichi Raw 2489.59 std_dev: 59.279 Predicted: 288/106 = 2.716 seconds Divisor Actual: 115.68 Tsoabichi Flurry I 2168.9566 std_dev: 127.44 (87.12% ~= 12.87%) Predicted: 288/106 *.85 = 2.3086 seconds Predicted w/ 116 divisor: 2.11 seconds Tsoabichi Snapshot 15 2224.063 std_dev: 161.4736 (89.33% ~= 10.665%) Predicted: 288/106 *.85 = 2.3086 seconds Predicted w/ 116 divisor: 2.11 seconds Tsoabichi Flurry I Snapshot 15 1982.71 std_dev: 158.043 (79.64% ~= 20.359%) Predicted: 288/106 * .7 = 1.9018 seconds Predicted w/ 116 divisor: 1.7379 seconds Failnaught Raw 5013.80 std_dev: 71.725 Predicted: 582/106 = 5.49 seconds Divisor Actual: 116.08 Failnaught Flurry I 4255.206 std_dev:115.48 (84.86% ~= 15.13%) Predicted: 582/106 * .85 = 4.66 seconds Predicted w/ 116 divisor: 5.01724 * .85 = 4.264 Failnaught Snapshot 15 4280.50 std_dev:155.55 (85.374% ~= 14.625%) Predicted: 582/106 * .85 = 4.66 seconds Predicted w/ 116 divisor: 5.01724 * .85 = 4.264 Failnaught Flurry I Snapshot 15 3626.89 std_dev: 158.025 (72.33% ~= 27.66%) Predicted: 582/106 * .7 = 3.8433 seconds Predicted w/ 116 divisor: 3.512 Trying to remove packet delay doubled: 224 + 224 = 448 Failnaught 4.5658 Divisor becomes 127.46 Failnaught 3.807206 Flurry I Predicted 3.8812 Failnaught 3.8325 Snapshot 15 Predicted 3.8812 Failnaught 3.17889 Flurry I Snapshot 15 Predicted 3.196 Trying to remove packet delay single: Failnaught 4.7898 Divisor becomes 121.5 Failnaught Flurry I 4.0716049 Predicted 4.071604 Failnaught Snapshot 15 4.0565 Predicted 4.071604 Failnaught 3.40289 Predicted 3.35308 Tsoabichi Raw 2.26559 Divisor becomes 127.119 Tsoabichi Flurry I 1.9449566 Predicted: 1.92575 Tsoabichi Snapshot 15 2.000063 Predicted: 1.92575 Tsoabichi Flurry I Snapshot 15 1.75871 Predicted: 1.585913 For Net 1 Culverin delay doubled: 189 * 2= 378 6603.77 - 378 = 6230.77 Divisor becomes 121.97 Flurry I 5.10051 Predicted: 5.296 Snapshot 15 5.129.3 Predicted: 5.296 For Net 1 Culverin delay single: 6603.77 - 189 = 6414.77 Divisor becomes 118.4 Flurry I 5.28951 Predicted: 5.456 Snapshot 15 5.3183 Predicted: 5.456 Testing Extremely High packet delay:
I still need to test extremely high packet delay and see how that affects flurry. What seems to be going on is the actual divisor is closer to 120, than 106 with about 116 being what I am able to record. Ruling rapidshot out is hugely helpful for testing flurry or snapshot, it does not however make packet delay and network spikes non existent. Samples with higher standard deviations are probably, but not necessarily, less reliable. There does also appear to be about a .4 second server update artifact making it harder to tell exactly where you are and this artificat could explain why lower delay weapons like Tsoabichi seem to get about 4% less benefit on average for 15 snapshot and failnaught (notice spelling it's one THF can use) and Culverin about 1% difference. These results are unfinished but I might never finish them if I don't kick my own butt and at least show the work I have done.
Things left to test:
1) Does snapshot really cap at 70?
- Have to make a 70 snapshot and over 70 snapshot set for THF, the most reliable way to do this is spend a lot on skirmish stones and combine it with flurry after I have solid flurry numbers, a less "good" way would be to just do it on ranger with tons of samples and only use the high delay numbers and try to distinguish those from just lag spikes (blech).
2) How much does your internet latency affect snapshot?
- Have to repeat tests I've done, using the worst VPN I can find to give very long responses from SE's server. This will also probably reveal some other interesting things about how it works.
3) Higher Samples are needed to be more conclusive, some of the results vary by 3% for the same tests. Testing in different zones might also be helpful, but any set of data needs to be collected on the same net around the same time as network latencies can even change if the network is on heavy load (like for me seems around 8pm at night when everyone connected is watching netflix or something).
4) Try to support or deny the .4 second artifact with data, find a way to quantify lag spikes as outliers then see if the variance is within the bounds of this .4 second artifact + latency.
I should note this is building on top of, rather than replacing testing done by others in the past, and it's always good to have more data to compare to. I have SQL files stored with the timings. Windower lets me do code when the ranged packet is received from SE and the ranged end packet is received from SE.
Usually it's more fun to just blow up stuff on one of my favorite jobs so I apologize if this is slow going.
[+]
All FFXI content and images © 2002-2024 SQUARE ENIX CO., LTD. FINAL
FANTASY is a registered trademark of Square Enix Co., Ltd.