Småland fire rate testing - DD skills don't work as advertised?

[Gebe_]

After some heated discussion with @Hirohito I decided to do some fire rate tests on Småland.

 

I tested all the captain skills that affect the fire rate in two different "modes": sequential firing vs salvo firing. The test subject was Vermont at about 1 km distance. I fired HE shells at his broadside continuously for 3 minutes for every test. I double checked all results in the battle results screen. I took my expected results for RPM from WoWs Fitting Tool but they are easily verifiable with a calculator.

 

Here are the results:

Spoiler

 

-- Baseline (no captain skills affecting fire rate; I ran this test twice and got identical results both times)

 

Expected result: 40.11 RPM or 481 shell hits over 3 minutes.
Test result (shell hits over 3 minutes):
424 sequential fire
448 salvo fire
= 1.7-1.6 seconds actual reload

 

-- Swift in Silence (+5% reload)

 

Expected result: 38.20 RPM or 458 shell hits over 3 minutes.
Test result (shell hits over 3 minutes):
424 sequential fire
424 salvo fire
= 1.7 seconds actual reload = 0% to 6% slower reload compared to baseline

 

-- Main Battery and AA specialist (-5% reload)

 

Expected result: 42.22 RPM or 506 shell hits over 3 minutes.
Test result (shell hits over 3 minutes):
508 sequential fire
500 salvo fire
= average 1.42 seconds actual reload = 11.25% to 16.5% faster reload compared to baseline

 

-- Fearless Brawler (-10% reload)

 

Expected result: 44.56 RPM or 534 shell hits over 3 minutes.
Test result (shell hits over 3 minutes):
508 sequential fire
492 salvo fire
= average 1.42 seconds actual reload = 11.25% to 16.5% faster reload compared to baseline

 

-- Fearless Brawler + Main Battery and AA specialist (-14.5% reload)

 

Expected result: 46.91 RPM or 562 shell hits over 3 minutes.
Test result (shell hits over 3 minutes):
508 sequential fire
536 salvo fire
= average 1.38 seconds actual reload = 13.75% to 18.8% faster reload compared to baseline

 

 

 

TLDR:

Some or all destroyer captain skills affecting the gun fire rate seem to be broken. None of the results match the expected theoretical fire rate but this could be due to lag/desync. However, if we assume that the base fire rate is not broken, then Main Battery and AA Specialist (the new BFT) seems quite certainly broken and gives the same reload speed improvement - 11% to 16% - as Fearless Brawler (which might also be broken). Swift in Silence might also be broken, but I'd value this particular test result as inconclusive (0% to 5% reload speed difference).

 

Regarding sequential firing vs salvo firing, results seem inconclusive. In some tests sequential fire produced slightly faster or the same fire rate, in others salvo fire produced significantly (5% to 10%) faster fire rate.  I ran the baseline test twice to double-check the result and got identical numbers. There could be many factors influencing this (like desync) so decide for yourself.

 

It would be great if someone (or preferably several people) could validate the results as this is only one data point. I'm also quite tired after a long day - I hope all my calculations are correct, but a double check would be nice.

 

I know that for now I will have Main Battery and AA Specialist (the new BFT) on most of my DDs.

 

[Kartoffelmos]

I can confirm that the "Main Battery and AA Specialist" is not working as intended at least (or that there are some other shenanigans at work):

On 2/1/2021 at 8:32 PM, Kartoffelmos said:

One minute of shooting:

172 shells with sequential fire

174 shells with salvoes

 

Totally as expected, since the reason why you want to fire salvoes is because of the initial strike (where those 2 extra shells came from) and because you will lose DPM if you are using a ship with many turrets and fast reloads. Småland, due to her two turrets, do not benefit much by the alpha strike and do not suffer the negative effects of sequential firing of the turrets.

 

What was not as expected however, were the results without using BFT (I tested this one multiple times and the results were similar):

144 shells with sequential fire

152 shells with salvoes

 

Not only is the gap larger, the difference between BFT and non-BFT does not add up to 5%. Even if I remove 4 shots from the BFT values (or add 4 shots to the non-BFT values), the numbers do not add up. It almost appears that the reload above 1,5 seconds has a slower refresh rate than those below this value. In fact, I went back to verify the BFT values and I noticed that the reload bar only occasionally turned green when sequential firing. Without BFT, the bar turned green between each reload.

 

[Hirohito]

Nice posting your findings @Gebe_.
I will test the same BFT as well for Småland, and if the rate of fire increases significantly above 5% (in the interval you mentioned) I might go ahead and test it for other rapid-firing DDs as well (starting with Halland and Daring).

I'll also post my own finding regarding sequential vs salvo fire here, in case anyone is interested:
I did two separate tests, one involving 30 second intervals and the other with 60 second intervals (1 minute).
In both tests, I parked next to a Kremlin and shot into the belt armour for the entire tests, using a stopwatch to time the durations.

 

Thee first test had me testing twice for sequential fire (holding the left mouse button down for 30x2 seconds) and twice for salvo fire (clicking the left mouse button rapidly for 30x2 seconds).
The results were 64 shell hits in both of my 30-second tests for sequential fire, and 76 and 78 shell hits respectively for salvo fire.
Those gave a whopping 20,3% increase(!) in rate of fire.

The second test had me testing twice for sequential fire again (60x2 seconds) and twice for salvo fire (60x2).
The results were 144 shell hits in both of my 30-second tests for sequential fire, and 152 shell hits for both of my salvo fire tests.
This worked out to be a meagre 5,5% increase in rate of fire.

This necessarily warrants more testing, but so far my evidence shows that rapidly clicking should produce a few more shots during a certain timeframe.
Most likely this is due to lag between client and server, and how the client will delay re-firing the guns since it believes the guns are still on cooldown.
I'm not 100% sure on this, but either way it seems that the guns behave very inconsistently depending on mode of fire, which they in theory should not.

[Hirohito]

25 minutes ago, Gebe_ said:

TLDR:

Some or all destroyer captain skills affecting the gun fire rate seem to be broken. None of the results match the expected theoretical fire rate but this could be due to lag/desync. However, if we assume that the base fire rate is not broken, then Main Battery and AA Specialist (the new BFT) seems quite certainly broken and gives the same reload speed improvement - 11% to 16% - as Fearless Brawler (which might also be broken).

 

As I haven't tested the new BFT myself, it might be that the game cannot handle fractions of reload time, or that main battery reload mod (-12% reload) is applied incorrectly, giving cumulative effects (perhaps combined with a rounding error).

Meaning that whenever the game attempts to reduce the reload time through a percentage, the already rapid-firing nature of Småland guns could cause the game to round down to the nearest whole integer in order to make it easier to calculate.
Since the actual reload time should be handled server-side, the displayed reload on the client side might not be the actual value used by the server.
In this case, there is a possibility that BFT in reality reduces the number down to 1,4s reload, even though it should be reduced to 1,42s (or 1,41s depending on whether percentage effects are subtracted cumulatively or not).
Thats still a low difference in effect compared to your results, so it should possibly warrant some further testing, but it might serve as a starting point to explore whether or not the game can handle smaller fractions of reload time, or whether or not it rounds the number down (which in effect, increases dpm).

[Hirohito]

Either way, it seems that for ultra-rapid firing guns, the game seems to behave very wonky and calculate rate of fire incorrectly.
There is absolutely no way that BFT (or salvo fire for that matter) should produce such large differences in results from each test to the next.

While Småland is about the fastest-firing DD I can come to think of (noteworthy competitors being Friesland and Halland), any further results should apply to other DDs as well (though likely to a lesser extent).
It remains to be seen if this can be exploited in some way.

[GulvkluderGuld]

Interesting test, thanks for posting it! 

Confirms my impression that BFT was definitely still effective in Ranked and seemed to eclipse Dazzle. Might try to repeat the test and add here in a day or so - quite tired atm!

 

Definitely some spaghetti code is involved with the interactions!

 

Maybe WG didnt bother changing the code for BTF and just rebranded it - Like the Vittorio that still has the Roma nameplate 

 

[gopher31]

Interesting. Thank you for sharing. Perhaps time to respec my destroyers.

[Han_Sh0t_First]

I run Fearless Brawler + Main Battery + AA Spec. 

 

I thought I was going out of my mind when playing, as it seemed that Salvo firing was WAY more effective (More hits on targets) when engaging other DD players (Which makes sense, as those are usually quick sub 30 second fights). 

 

20% more rounds? Damn... 

[SirTogII]

Interesting. Time to look over the cappy skills for the gunboats

[Hirohito]

20 minutes ago, Han_Sh0t_First said:

I run Fearless Brawler + Main Battery + AA Spec. 

 

I thought I was going out of my mind when playing, as it seemed that Salvo firing was WAY more effective (More hits on targets) when engaging other DD players (Which makes sense, as those are usually quick sub 30 second fights). 

 

20% more rounds? Damn... 

Those were the extreme numbers I found for 30 second engagements, yeah.
The effect dropped off for 60 second engagements, but it was still positive.
The results were too high to be explained by me mis-timing the stop watch, since that should have created one extra salvo at most, not 3-4 extra salvoes which I got in my 30-second tests.
And since I ran the tests multiple times, the chance that I mis-timed one of the tests is even slimmer.

If my hypothesis is correct in that lag between server and client causes this, it means that you could likely improve the numbers further by using a rapid-firing mouse (holding the mouse button down with rapid fire activated) to spam out left click inputs which get transmitted to the server.
That should make sure that once the server-side reload is finished, the rapid-fire function on your mouse should immediately issue a shot command regardless of whether the client thinks the guns are reloaded or not.

Unfortunately I don't have a rapid-firing mouse, but if anyone who has one wants to test this further, the results could be interesting.

That being said, @Gebe_ apparently found (in some select cases) that the shot number went down when he tested salvo vs sequential fire, even if other tests showed the number to go up.
What is certain is however that the stated in-game number for rate of fire is incorrect, and salvo fire for the time being tends to trend upwards (producing more salvoes than you should get from sequential fire).
If BFT is bugged as well (which seems to hold true), it seems to suggest that there is an very large difference in rate of fire that could be gained (used to win a DD vs DD fight mostly) if one works out how to maximize the number of shots, over someone who doesn't know how to do so.

[Han_Sh0t_First]

9 minutes ago, Hirohito said:

Those were the extreme numbers I found for 30 second engagements, yeah.
The effect dropped off for 60 second engagements, but it was still positive.
The results were too high to be explained by me mis-timing the stop watch, since that should have created one extra salvo at most, not 3-4 extra salvoes which I got in my 30-second tests.
And since I ran the tests multiple times, the chance that I mis-timed one of the tests is even slimmer.

If my hypothesis is correct in that lag between server and client causes this, it means that you could likely improve the numbers further by using a rapid-firing mouse (holding the mouse button down with rapid fire activated) to spam out left click inputs which get transmitted to the server.
That should make sure that once the server-side reload is finished, the rapid-fire function on your mouse should immediately issue a shot command regardless of whether the client thinks the guns are reloaded or not.

Unfortunately I don't have a rapid-firing mouse, but if anyone who has one wants to test this further, the results could be interesting.

That being said, @Gebe_ apparently found (in some select cases) that the shot number went down when he tested salvo vs sequential fire, even if other tests showed the number to go up.
What is certain is however that the stated in-game number for rate of fire is incorrect, and salvo fire for the time being tends to trend upwards (producing more salvoes than you should get from sequential fire).
If BFT is bugged as well (which seems to hold true), it seems to suggest that there is an very large difference in rate of fire that could be gained (used to win a DD vs DD fight mostly) if one works out how to maximize the number of shots, over someone who doesn't know how to do so.

I will give the rapid fire a twirl this evening mate. 

 

Have a test account, so can make a macro (For test purposes!) that just spams left click every 0.2s until deactivated. 

 

Actually, if this holds true I will probably run it with different intervals (0.2, 0.5, 0.8, 1.0s) and see if that yields any difference. I am sure there is a sweet spot. 

 

Vermonth, 1km away - 5 runs each. 

[Hirohito]

24 minutes ago, Han_Sh0t_First said:

I will give the rapid fire a twirl this evening mate. 

 

 Have a test account, so can make a macro (For test purposes!) that just spams left click every 0.2s until deactivated. 

 

Actually, if this holds true I will probably run it with different intervals (0.2, 0.5, 0.8, 1.0s) and see if that yields any difference. I am sure there is a sweet spot. 

 

Vermonth, 1km away - 5 runs each. 

I would spam it more frequently than every 0.2s, as that is already about as fast I can click.
If it would be possible to reduce the clicking interval down to 0,05s or something similar it should be interesting though, as that is way faster than a human can click.

 

Whether or not lag causes it (or something else) is irrelevant, the clue is to get the guns firing as soon as the server allows for another shot to be made, and the further one can reduce that window (more clicks per second) the better.

Edit: Make sure to post any results here (also whether you used BFT or not). The more we can find some consistent pattern to raising DPM, the better! 

 

[Hugh_Ruka]

try with something in the 4-5 second reload range ...

 

also if your dispersion is wonky and you hit water there will be no shell hits = no ribbons ....

[Hirohito]

4 minutes ago, Hugh_Ruka said:

try with something in the 4-5 second reload range ...

 

also if your dispersion is wonky and you hit water there will be no shell hits = no ribbons ....

You won't really miss at point blank range against a BBs belt armour though.
In my tests I made sure that I watched the ribbons and that they incremented in sets of 4, which they did.

Testing for reloads in the 4-5 second range would be interesting as well.
However, I strongly suspect that the game incorrectly handles fractions  of seconds of gun reload, meaning that Småland atm is the ideal candidate to test since it is the boat most affected by fractions of gun reload to reach its "true" listed dpm (or any incorrect number for that matter).
The slower the guns are, the more likely it is that they start behaving like "normal" (the extreme example being a 30s BB reload).

[Hugh_Ruka]

23 minutes ago, Hirohito said:

You won't really miss at point blank range against a BBs belt armour though.
In my tests I made sure that I watched the ribbons and that they incremented in sets of 4, which they did.

Testing for reloads in the 4-5 second range would be interesting as well.
However, I strongly suspect that the game incorrectly handles fractions  of seconds of gun reload, meaning that Småland atm is the ideal candidate to test since it is the boat most affected by fractions of gun reload to reach its "true" listed dpm (or any incorrect number for that matter).
The slower the guns are, the more likely it is that they start behaving like "normal" (the extreme example being a 30s BB reload).

Well if you counted each volley then ok ... I'll have to try in the Khaba as well with all the options ...

[Spectre_S_]

There shouldn't be any issues with misses , can always check the post battle result screen for the number of shells fired , also have a mouse with a button that produces 3 clicks when pressed , been using that when playing wows since forever , always felt that i outdmg other darings when in my own for the engagement time .

What i did notice however since the last patch sometimes one salvo is very innacurate at random times as if the guns weren't locked on targer , but guess this isn't the topic for that discussion .

[Meowgli]

29 minutes ago, Spectre_S_ said:

There shouldn't be any issues with misses , can always check the post battle result screen for the number of shells fired , also have a mouse with a button that produces 3 clicks when pressed , been using that when playing wows since forever , always felt that i outdmg other darings when in my own for the engagement time .

What i did notice however since the last patch sometimes one salvo is very innacurate at random times as if the guns weren't locked on targer , but guess this isn't the topic for that discussion .

big brain time here!
11/10
this is a violation of the official rules

"5.07. Using bots, clickers, macros, keyboard and mouse recorders, or any other similar methods that enables gameplay without the participation or with passive participation of the player in the battle."


"1.04. Wargaming may suspend, terminate, modify, or delete Accounts in accordance with the EULA at any time. Accounts terminated by Wargaming for any type of abuse, including without limitation a violation of these Rules or the EULA, will not be reactivated for any reason. For purposes of explanation and not limitation, most account suspensions, terminations, and/or deletions are the result of violations of these rules, the EULA, or other Wargaming policies. Players may cancel any Account registered to themselves at any time by following the instructions in the Wargaming Privacy Policy. Wargaming may stop offering and/or supporting any service at any time."

[159Hunter]

Thing is, changes are in .1 sec range for reload times. If your test is automated, forget my comment, if not: how can you be sure you are actually firing at the exact time your reload is finished? Just a small missclick can easily be .5 sec delay? So your margin of error of the testing method is far bigger than the gains/losses in the reload times you want to confirm?

[Hirohito]

29 minutes ago, 159Hunter said:

Thing is, changes are in .1 sec range for reload times. If your test is automated, forget my comment, if not: how can you be sure you are actually firing at the exact time your reload is finished? Just a small missclick can easily be .5 sec delay? So your margin of error of the testing method is far bigger than the gains/losses in the reload times you want to confirm?

I'm not sure how this works myself, only that clicking seems to fire the guns slightly before an automated sequential fire would fire them.
It could be that the server uses a batching system, where it has these small windows (batches) where any input (gun shots in this case) are applied in the succeeding batch.

Meaning that if you spam click gun shots prematurely, the game might apply them the very next instance (batch) the guns come off reload, whereas an automated sequential fire might miss one or even several batches.
I know that a similar system was used server side in World of Warcraft back on their older versions, as the system should save a lot of processing power over smooth and constant client/server interactions.

Not sure about any of this though, but it warrants enough interest since the guns behave wildly inconsistent atm, with the difference in number of gun shots produced being larger than what could be explained from simple timing errors.
Even in my second test which only yielded a 5,5% increase (60second tests) on average, I ended up with two more shots (8 more hits) over 60 seconds.
If I had timed it slightly wrong, I would expect to see at most one more shot (4 shell hits), not twice that amount.
We are talking about 3 seconds worth of gun dpm over 60 seconds in this case, and this was just my "low" result.

BFT also seems to interact in an odd way here, producing results it shouldn't, even if we assume timing errors (on the stopwatch).
If the batching hypothesis holds true, it could be that the percentage increase pushes the reload down to an earlier batch, allowing it to fire much sooner than they are advertised to do (5%).

[Echo_519]

So I tested some builds with Fletcher and Yugumo, very interesting results:

 

Fletcher: 

no upgrades, no skills

Expected Reload time: 3.34

Testes Reload time: 3.4

0.06 difference

 

3mio reload upgrade, no skills

Expected Reload time: 2.94

Tested Reload time: 3.14

0.2 difference

 

no upgrade, Main battery and AA specialist+ fearless brawler

expected reload time: 2.86

tested reload time: 2.86

no difference

 

3mio reload upgrade, Main battery and AA specialist+ fearless brawler

Expected reload time: 2.51

Tested reload time: 2.55

0.04 difference

 

 

Yugumo:

3mio reload upgrade, no skills

Expected Reload time: 5.02

Tested reload time: 5.13

0.11 difference

 

no upgrade, Main battery and AA specialist+ fearless brawler

Expected Reload time: 4.88

Tested reload time: 5.14

0.26 difference

 

3mio reload upgrade, Main battery and AA specialist+ fearless brawler

Expected Reload time: 4.3

Tested reload time: 4.57

0.27 difference

 

 

Tested every build several times, biggest between one build was 0.01, so rather stable test...

I'll go test some more builds with those ships, but seems that there are certain ship/upgrade/skill combination that gives the ships huge nerfs to it's expected reload time... All tests were made with sequantial firing, btw....

[Spectre_S_]

2 hours ago, Meowgli said:

big brain time here!
11/10
this is a violation of the official rules

"5.07. Using bots, clickers, macros, keyboard and mouse recorders, or any other similar methods that enables gameplay without the participation or with passive participation of the player in the battle."


"1.04. Wargaming may suspend, terminate, modify, or delete Accounts in accordance with the EULA at any time. Accounts terminated by Wargaming for any type of abuse, including without limitation a violation of these Rules or the EULA, will not be reactivated for any reason. For purposes of explanation and not limitation, most account suspensions, terminations, and/or deletions are the result of violations of these rules, the EULA, or other Wargaming policies. Players may cancel any Account registered to themselves at any time by following the instructions in the Wargaming Privacy Policy. Wargaming may stop offering and/or supporting any service at any time."

Then i guess you should reread that statement , especially the part about :

"that enables gameplay without the participation or with passive participation of the player in the battle."  

 

I'm pretty sure I'm not afk'ing when in exchanging gun fire with other ships .

[Ydoum]

WG and STs not doing proper testing before release of a major rework?

No way.

[Hirohito]

Ok, I have done some extensive testing now, both with sequential fire and salvo fire, and with and without BFT.
The results are very(!) surprising and suggest that BFT is broken atm for some unknown reason, like @Gebe_ suggested.
Similarly, players might want to be mindful of the way they fire their guns, as firing mode matters for DPM.

As usual, I ran two sets of tests - one of them lasting 30 seconds, and one of them lasting 60 seconds.
Both tests were run with sequential fire (holding down the mouse button), and with salvo fire (hammering the mouse button).
After these tests were made, I ran the same test with BFT.
Here are the results:

No BFT:

30 second sequential fire: 72 shell hits (test 1), 72 shell hits (test 2)

30 second salvo fire: 76 shell hits (test 1), 76 shell hits (test 2)

60 second sequential fire: 144 shell hits (test 1), 144 shell hits (test 2)

60 second salvo fire: 152 shell hits (test 1), 156 shell hits (test 2)

 

With BFT:

30 second sequential fire: 88 shell hits (test 1), 88 shell hits (test 2)

30 second salvo fire: 88 shell hits (test 1), 88 shell hits (test 2)

60 second sequential fire: 172 shell hits (test 1), 172 shell hits (test 2)

60 second salvo fire: 168 shell hits (test 1), 168 shell hits (test 2)

 

Conclusion:

The results were very surprising to be honest.
It seems that if you do not run BFT, salvo fire is consistently stronger than sequential fire.

For my 30 second tests, salvo fire turned out to be a consistent 5,55% DPM increase.
For my 60 second tests, salvo fire turned out to be on average a 6,94% DPM increase.

 

However, the results were even more surprising for BFT skilled.
With BFT, it seems that salvo fire is no longer as efficient.
In fact, you barely lose out a salvo (4 shell hits) over 60 seconds unless you perfectly mash the mouse button.
Sequential fire is therefore (very slightly) superiorto salvo fire.
The most surprising result however was the DPM increase for BFT.
For the 30 second tests, BFT yielded an insane 22,2% DPM increase for sequential fire, and a 15,78% increase for salvo fire.

For the 60 second tests, BFT yielded a 19,44% DPM increase for sequential fire, and a 9,09% DPM increase for salvo fire.

The results of this test suggest that if you do NOT run BFT (why wouldn't you? Its very strong), you should always fire in salvoes.
If you do run BFT (you should!), there is less of a difference between firing in salvoes and sequential fire.
In fact, unless you want a strained arm, I suggest you fire in sequential fire, its more likely to yield the best results regardless.
Either way: SPEC BFT RIGHT NOW, it is broken in its current form and gives between a 9,09% and 22,2% DPM increase!

[Echo_519]

Vor 46 Minuten, Hirohito sagte:

The results are very(!) surprising and suggest that BFT is broken atm for some unknown reason

 

Did you test this on several ships?

Becasue i don't think BFT is the only issue. 

All kind of skills/upgrades and what so ever that effect reload time are broken or dont work correctly....

For example, on fletcher, BFT alone does not give you any bonus at all. 

In my test, with stock Fletcher i got 3.428 second reload, with only BTF, i still had 3.427 seconds average. No bonus.

The 3 mio reload upgrae for slot 7 also does not work properly either, without any commander skills, this upgrade puts fletcher reload at  3.138 seconds.That's roughly a 8.5% buff, 2/3 of what you pay for...

 

Did a full test on Fletcher reaload times with the slot7upgrade, BFT and Fearless Brawler with all possible combinations:

Stock: 3.428

BFT: 3.427  (-0.045%)/-5%

BFT+Fearles Brawler: 2.866  (-16.396%)/-14.5%

Fearless Brawler: 3.145  (-8.263%)/-10%

Slot7Upgrade:3.138  (-8.476%)/-12%

Slot7+BFT: 2.861  (-16.537%)/-16.4%

Slot7+BFT+Fearless Brawler: 2.582  (-24.686%)/-24.76%

Slot7+Fearless Brawler; 2.846  (-16.984%)/-20.8%

In brackets the reloadbuff based on the stock reload, behing the "/" the reloadbuff this build SHOULD have...

 

BFT alone does not work alone, but BFT+Fearless Brawler works too good, Slot7upgrade only does 2/3 of what you pay for...

Thus, my conclusion: There is something seriously wrong with everything that affects reload, not just BFT...

 

Not done yet fully testing yugumo, but it seems rather random what combination of skills/upgrade does not work...

What i've got so far for yugumo:

Stock: 5.721

BFT: 5.445  (-4.833%)/-5%

BFT+Fearles Brawler: 5.123  (-10.452%)/-14.5%

Fearless Brawler: 5.152  (-9.946%)/-10%

 

Here BFT and Fearless Brawler work fine seperately, but in combination you only get 2/3....

 

 

This is so broken....