Login

russian armor

ROF doesn't quite work as we thought it did

13 Apr 2020, 16:37 PM
#1
avatar of Hannibal
Senior Moderator Badge

Posts: 3102 | Subs: 2

After some more extensive testing I think this is something worth its own thread.

For a different project, I tested the rate of fire (ROF) in the game (context can be found from here onwards). In short, I noticed that the ROF is quite a bit longer than it should be on the basis of all formulas used by Cruzz and probably also the websites used for DPS calculations. There are four values that determine the ROF of a tank. My first gut feeling told me that for every stat used, there is an additional 0,125 second delay independent of the stat value itself.

So, working hypothesis (for tanks only):
ROF = wind_up + fire_aim_time + wind_down + reload + 0,125 * number_of_stats_used



I modded a T34 and a OST P4 for simultaneous testing to speed up the process.


Note: I called the rate of fire (ROF) a different name - time between shots (TBS).

The following table is a bit chaotic but I'll take you through it:


What we can see is, that my new model is way more accurate than the current one. So I assume my model to be true. My explanation for this is that the game runs in ticks of 0,125 seconds. So every time one step of the process is done, it has to wait for the next tick to start the following process.
Wind_up - 0,125 second wait - fire_aim_time - wait......

Now the T34 seems to have major issues if it has no reload time. I don't know why. The P4 doesn't though, and for all other tests even the T34 worked fine.

So why is this important?
The errors generally generated by the "wrong" ROF are in a range of 3-8% (see current model error below), depending on how many stats a unit uses. So it's not a super duper difference, but still noticeable for comparison of units.



Also I only tested tanks. How this applied for infantry that uses basically more stats I do not know. I assume the issue is the same, and I expect the differences to be larger than for tanks. This could mean that we might need to recalculate the DPS of many units, especially since also the balance team uses these DPS to compare and rework units.
13 Apr 2020, 16:43 PM
#2
avatar of Vipper

Posts: 13476 | Subs: 1

13 Apr 2020, 17:22 PM
#3
avatar of BlueKnight

Posts: 320

Could it be that infantry ROF also hides some secrets similar to the ones presented in op? Nobel prize for the first guy to explain suppression calculations.

Am I correct to assume then, that the discrepancy in TBS between the JP4 and Jackson is larger than expected and to disadvantage of Jackson?
13 Apr 2020, 18:04 PM
#4
avatar of Hannibal
Senior Moderator Badge

Posts: 3102 | Subs: 2

Could it be that infantry ROF also hides some secrets similar to the ones presented in op? Nobel prize for the first guy to explain suppression calculations.

Am I correct to assume then, that the discrepancy in TBS between the JP4 and Jackson is larger than expected and to disadvantage of Jackson?


I am not 100% sure that the JP4 does not have wind up and wind down. But if it does not, then you are correct. It would generate an additional difference of 0,25 seconds.
13 Apr 2020, 18:17 PM
#5
avatar of elchino7
Senior Moderator Badge

Posts: 8154 | Subs: 2

Slightly unrelated.

Whether this has a real impact or not, I THINK that i found this discrepancy in performance due to tickrate in income generated by abilities such as scavenging.

The UKF scavenge engineers siphon resources from wreck based on type and the HP it has (which always starts at 50%).

By using the cheatmode v2 which let's you reset the income to 0 and resources as well, i end up having different results after repeating the test on same wreck.
MMX
14 Apr 2020, 09:53 AM
#6
avatar of MMX

Posts: 999 | Subs: 1

Great find, you're definitely onto something here. I've done a bit of testing and I see the same discrepancy between the in-game ROF and theoretical values.











Nametheoreticalin-game
Pershing6.7507.000
M8 Greyhound4.0004.250
Pack Howi AF9.75010.125
Brumm8.5008.613
Zis-35.8636.245
KV-8 45mm5.0005.112
HM-38 120mm8.2388.542
PM-41 82mm6.9507.280


I've tried correcting the theoretical TBS by your method, but I always get 0.25 s more than in-game. Pretty strange, but I agree this can be pretty important and deserves a closer look!
The more you know...
14 Apr 2020, 10:03 AM
#7
avatar of Vipper

Posts: 13476 | Subs: 1

suggest that you ask a moder to create a mod that measures damage and ROF then it can calculate DPS in game.
14 Apr 2020, 10:28 AM
#8
avatar of JibberJabberJobber

Posts: 1614 | Subs: 3

How much do reload bulletins affect the time between shots though :D
14 Apr 2020, 10:54 AM
#9
avatar of Vipper

Posts: 13476 | Subs: 1

How much do reload bulletins affect the time between shots though :D

The modifier must go no the reload itself so the effect would be different a cross units. For instance KV-2 has a very longreload and has a hidden bonus....
14 Apr 2020, 13:24 PM
#10
avatar of Hannibal
Senior Moderator Badge

Posts: 3102 | Subs: 2

jump backJump back to quoted post14 Apr 2020, 09:53 AMMMX
Great find, you're definitely onto something here. I've done a bit of testing and I see the same discrepancy between the in-game ROF and theoretical values.











Nametheoreticalin-game
Pershing6.7507.000
M8 Greyhound4.0004.250
Pack Howi AF9.75010.125
Brumm8.5008.613
Zis-35.8636.245
KV-8 45mm5.0005.112
HM-38 120mm8.2388.542
PM-41 82mm6.9507.280


I've tried correcting the theoretical TBS by your method, but I always get 0.25 s more than in-game. Pretty strange, but I agree this can be pretty important and deserves a closer look!
The more you know...


How did you measure the times?
I literally stopped them by a stop watch. So there should be some minor errors for stopping the time, but compared to the duration I measured overall they should be minor.
I also tested the Brummbär and my formula seems to fit pretty well.
14 Apr 2020, 13:25 PM
#11
avatar of Hannibal
Senior Moderator Badge

Posts: 3102 | Subs: 2

How much do reload bulletins affect the time between shots though :D

I think they just affect the reload stats as Vipper said.
But I assume that you always have to round them up to the next 0.125 second tick.
MMX
15 Apr 2020, 05:11 AM
#12
avatar of MMX

Posts: 999 | Subs: 1



How did you measure the times?
I literally stopped them by a stop watch. So there should be some minor errors for stopping the time, but compared to the duration I measured overall they should be minor.
I also tested the Brummbär and my formula seems to fit pretty well.


ye good olde trusty stopwatch here as well, haha. though if someone could create a mod as suggested by vipper things would be much easier and precise. i tried messing with the refresh rate of the in-game damage counter a bit since, in theory, setting the delay slightly higher than the time between shots would produce double damage ticks whenever 2 hits are registered between refresh. this is still far from optimal and quite tedious though...
15 Apr 2020, 09:19 AM
#13
avatar of Hannibal
Senior Moderator Badge

Posts: 3102 | Subs: 2

jump backJump back to quoted post15 Apr 2020, 05:11 AMMMX


ye good olde trusty stopwatch here as well, haha. though if someone could create a mod as suggested by vipper things would be much easier and precise. i tried messing with the refresh rate of the in-game damage counter a bit since, in theory, setting the delay slightly higher than the time between shots would produce double damage ticks whenever 2 hits are registered between refresh. this is still far from optimal and quite tedious though...


For the ZiS and Brummbär, we have measured very similar in-game times.
I am not sure if you corrected properly. For example the Brummbär has 8,25 reload and 0,125 fire aim, which would lead to (old model) 8,375 seconds. It uses two stats so I corrected for 0,25 seconds, which leads to 8,625 seconds. You somehow dropped 0,125 seconds at some point and got a corrected value of 8,5.

I think you only added the 0,125 s for correction once, regardless of how many stats are used.

For ZiS (just see my table), the correction is 0,5 seconds as it uses all four stats. You seem to have corrected by only 0,125 again.


From your table, I would predict Pershing and Greyhound for example use three stats, while the Howi uses all for. For the Pershing I know that it uses wind down I think, Greyhound and Howi no idea.
1 user is browsing this thread: 1 guest

Livestreams

unknown 5
Korea, Republic Of 5

Ladders Top 10

  • #
    Steam Alias
    W
    L
    %
    Streak
Data provided by Relic Relic Entertainment

Replay highlight

VS
  • U.S. Forces flag cblanco ★
  • The British Forces flag 보드카 중대
  • Oberkommando West flag VonManteuffel
  • Ostheer flag Heartless Jäger
uploaded by XXxxHeartlessxxXX

Board Info

429 users are online: 1 member and 428 guests
Snack_Master
6 posts in the last 24h
32 posts in the last week
86 posts in the last month
Registered members: 44638
Welcome our newest member, Leilichuu96
Most online: 2043 users on 29 Oct 2023, 01:04 AM