| Pages: 1 :: [one page] |
|
|
| Author |
Thread Statistics | Show CCP posts - 1 post(s) |
Synner Zerg
SVER True Blood
Unclaimed.
3
   |
Posted - 2013.02.12 03:07:00 -
[1] - Quote
The stats on the Militia Assault Dropsuit for CPU are:
150gF
I have Circuitry trained to LVL4, which gives me 4x5% = 20% CPU increase
150 x 1.20 = 180 gF
In-game, when I am fitting the just the suit (no other modules whatsoever), it is showing I have 179 gF
I tried the same with a Militia Scout suit, and same result, it is missing 1 gF of CPU after the basic calculation of skills only (no modules, no guns, nothing)
So, where is the leak?
Can anyone with higher-tier suits confirm if they are getting the same leak with their suits?
|
|
CCP Eterne
C C P
C C P Alliance
759
|
Posted - 2013.02.12 10:09:00 -
[2] - Quote
Moving from General Discussions to Technical Support/Bugs. |
|
Sword of Gallente
Villore Sec Ops
Gallente Federation
3
|
Posted - 2013.02.12 10:12:00 -
[3] - Quote
There is a -1 percentage mistake on nearly everything that i've noticed too. |
1LTNORFLEET
Lost-Legion
24
|
Posted - 2013.02.12 10:13:00 -
[4] - Quote
Synner Zerg wrote:The stats on the Militia Assault Dropsuit for CPU are:
150gF
I have Circuitry trained to LVL4, which gives me 4x5% = 20% CPU increase
150 x 1.20 = 180 gF
In-game, when I am fitting the just the suit (no other modules whatsoever), it is showing I have 179 gF
I tried the same with a Militia Scout suit, and same result, it is missing 1 gF of CPU after the basic calculation of skills only (no modules, no guns, nothing)
So, where is the leak?
Can anyone with higher-tier suits confirm if they are getting the same leak with their suits?
stats for guns are rounded numbers for example 153 cpu with 5% increase = 7.65 add to the 153 =160.65 rounded too 161 cpu |
KalOfTheRathi
CowTek
168
|
Posted - 2013.02.12 11:22:00 -
[5] - Quote
A standard Floating Point Unit (FPU) rounding option is truncate as are the more commonly used Round Down/Up.
Depending how the math is calculated it could actually increase the chance that something will get chopped off. Each additional calculation allows the FPU to truncate something off the total which is how the chance increases. This mean each and every add, subtract, multiply and divide in the equation.
If we have EVE players that look into this level of detail they could help clarify the subject for us in the DUST. If EVE calculations are truncated or Rounded Down I would expect all these would remain as it. Essentially being as designed and Not A Bug.
Sucks for us. |
Synner Zerg
SVER True Blood
Unclaimed.
3
|
Posted - 2013.02.12 14:30:00 -
[6] - Quote
I understand that rounding down can affect the results, especially when there are a lot of parts to the calculation, like extra mods, and multiple skills.
But in this case, it is only base stats x skill bonus:
150 x 20%
And the result is an integer, not a floating point number
Even if the skill bonus calculation is applied sequentially (which would be very inefficient, no?), and we round down every step, it is still 180, not 179 that the game shows.
LVL1 150 + (150 x 0.05) = 150 + 7.5 = 157 rounded down
LVL2 157 + (157 x 0.05) = 157 + 7.85 = 164 rounded down
LVL3 164 + (164 x 0.05) = 164 + 8.2 = 172 rounded down
LVL4 172 + (172 x 0.05) = 172 + 8.6 = 180 rounded down
((((150 x 1.05) x 1.05) x 1.05) x 1.05)
(((157.5 x 1.05) x 1.05) x 1.05)
((165.375 x 1.05) x 1.05)
(173.64375 x 1.05)
182.3259375 <- Unrounded result
((((150 x 1.05) x 1.05) x 1.05) x 1.05)
(((157 x 1.05) x 1.05) x 1.05)
((164 x 1.05) x 1.05)
(172 x 1.05)
180 <- Rounding down every step, same as the very top calculation
Non of these is 179... 
I just want a confirmation that this is a bug, and not something that I am missing, cause I am making a spreadsheet, and want to make sure my calculations are correct. Oh, and i've been playing EvE for over 6 years
|
Aqil Aegivan
The Southern Legion
50
|
Posted - 2013.02.12 14:52:00 -
[7] - Quote
Yeah, some people have posted similar issues when their bonuses should come out as an integer. It looks like they're finding the floor of fractional results using the next greatest number that's an integer. Unfortunately a floor of an integer is that integer, not a lower one. Of course, being devs, they already know better so I'm sure the actual problem is a lot more complicated.
Basically results that come out as whole numbers are reduced by one. That's really all I know. I hope they fix it but it never seems to be worse than one point. |
KalOfTheRathi
CowTek
168
|
Posted - 2013.02.12 16:33:00 -
[8] - Quote
Aqil Aegivan wrote:Yeah, some people have posted similar issues when their bonuses should come out as an integer. It looks like they're finding the floor of fractional results using the next greatest number that's an integer. Unfortunately a floor of an integer is that integer, not a lower one. Of course, being devs, they already know better so I'm sure the actual problem is a lot more complicated.
Basically results that come out as whole numbers are reduced by one. That's really all I know. I hope they fix it but it never seems to be worse than one point. Thanks, that is a much clearer answer than mine. As a software engineer I made sure the test cases verified the requirements. If they wanted rounding they got.
Just occurred to me, I wonder if they are pushing them through the SPE or the GPU? It would seem silly to be but Sony does love those processors.
And OP, Aqil is correct. The difference has never been more than 1. |
Epicsting pro
Planetary Response Organisation
Test Friends Please Ignore
11
|
Posted - 2013.03.27 09:19:00 -
[9] - Quote
Srry for the bump but before chromosome update it was never off by one. |
KalOfTheRathi
Talon Strike Force LTD
Sleepless Knights Alliance
278
|
Posted - 2013.03.27 11:02:00 -
[10] - Quote
It has been off for the entire Chromosome that is for certain. It was reported early on and has never been responded, replied or commented on. |
|
|
Banjo Hero
BurgezzE.T.F
Orion Empire
61
|
Posted - 2013.03.28 05:58:00 -
[11] - Quote
I don't know if this is what is happening with this, but, if you change your assumption that the reported 150 is precisely 150.00...
149.5 * 1.2 = 179.4
149.5 would usually be rounded up to 150
179.4 would usually be rounded down to 179
|
|
|
| |
|
| Pages: 1 :: [one page] |