Pages: [1] :: one page |
|
Author |
Thread Statistics | Show CCP posts - 0 post(s) |
Jack McReady
DUST University Ivy League
1669
|
Posted - 2014.12.27 18:19:00 -
[1] - Quote
following fit:
Assault mk.0 Complex Light Damage Modifier Complex Shield Extender Complex Shield Extender Complex Shield Recharger Complex Armor Repairer Complex Ferroscale Plates Complex Ferroscale Plates Complex Shield Regulator Six Kin Assault Combat Rifle Six Kin Submachine Gun Core Locus Grenade Wiyrkomi Nanite Injector
I have the neccessary skills at MAX. with my skills it should use use 514.4625 out of 514.5 CPU and 83.625 out of 85.3125 PG
BUT due to some wonky rounding the fit ends with 515 out of 514 CPU and is invalid...
CCP u no math? |
Megaman Trigger
Knights of Eternal Darkness
165
|
Posted - 2014.12.27 18:53:00 -
[2] - Quote
Wouldn't be surprised if suit available CPU and PG is rounded down and the actual fitting cost rounds up.
Also, why can a Proto Assault (in theory) fit all Proto yet the Sentinels got nerf'd for it?
Purifier. First Class.
|
Jack McReady
DUST University Ivy League
1669
|
Posted - 2014.12.27 19:29:00 -
[3] - Quote
Megaman Trigger wrote:Wouldn't be surprised if suit available CPU and PG is rounded down and the actual fitting cost rounds up.
pretty much looks like it!
|
Kaze Eyrou
Molon Labe. General Tso's Alliance
1177
|
Posted - 2014.12.27 23:24:00 -
[4] - Quote
This is interesting because you gave a number that would not be rounded, but yet could be rounded.
Let me explain better with an example:
514.4625. Look specifically at the tens place -> 514.4. Now round -> 514.
However, it seems like it does rounding while accounting for the hundred place as well. That means the system might be doing this:
514.4625. Round to the tens -> 514.5. Now round -> 515.
It's a very minute detail that I thought we'd never see but it looks like you finally hit the absolute in min-maxing.
I don't have an Assault mk.0, but I do have a Logistics mk.0 which reduces the equipment CPU and PG. Using Protofits, I have come up with 3 different fits to test out to check exactly where the rounding happens. I'll post again when I've finished testing.
CB Vet // True Logi Bro // @KazeEyrou
Learning Coalition and RTG Mentor
|
Kaze Eyrou
Molon Labe. General Tso's Alliance
1177
|
Posted - 2014.12.27 23:52:00 -
[5] - Quote
Passive skills that are applied for the following tests: Electronics 5 (raises CPU) Engineering 5 (raises PG) Light Weapon Operation 5 (reduces CPU usage on Light Weapons) Combat Rifle Proficiency 4 (reduces PG usage on Combat Rifles) Minmatar Logistics 5 (reduces CPU and PG usage on equipment) Nanocircuitry 5 (reduces CPU usage on Nanohives) Handheld Weapon Upgrades (reduces CPU usage on Damage Modifiers)
TEST 1: Logistics mk.0 Six Kin Assault Combat Rifle X-3 Quantum Nanohive
CPU used according to Protofits: 94.425 CPU used according to Dust: 94
TEST 2: Logistics mk.0 Enhanced Light Damage Modifier Enhanced Light Damage Modifier
CPU used according to Protofits: 76.5 CPU used according to Dust: 77
TEST 3: Logistics mk.0 Combat Rifle Nanite Injector Nanite Injector Nanite Injector Nanite Injector
CPU used according to Protofits: 52.45 CPU used according to Dust: 52
TEST 4: (This is where I was able to get the exact decimal as OP) Logistics mk.0 BK-42 Assault Combat Rifle Ishukone Flux Nanohive Allotek Nanohive (R) Repair Tool
CPU used according to Protofits: 130.4625 CPU used according to Dust: 130
I think you overlooked something somewhere OP. One of your skills isn't leveled up and/or you don't have Protofits set up exactly to the skills of your character. I don't see a rounding error on my end.
Double check something for me Jack: What are your Light Weapon Operation skill and Sidearm Operation skill levels at?
CB Vet // True Logi Bro // @KazeEyrou
Learning Coalition and RTG Mentor
|
Jack McReady
DUST University Ivy League
1669
|
Posted - 2014.12.28 00:30:00 -
[6] - Quote
Kaze Eyrou wrote:
I think you overlooked something somewhere OP. One of your skills isn't leveled up and/or you don't have Protofits set up exactly to the skills of your character. I don't see a rounding error on my end.
Double check something for me Jack: What are your Light Weapon Operation skill and Sidearm Operation skill levels at?
Hmm, after modifying the skill levels of various things, I couldn't get the CPU to go to 515. At best, I got it to 516 if you missed a level in a skill somewhere.
Maybe there's something wonky with the Assault suit and/or the Assault bonus? You are right it is 516 ingame. Anyway all my skills are maxed for the fit. |
Kaze Eyrou
Molon Labe. General Tso's Alliance
1177
|
Posted - 2014.12.28 06:18:00 -
[7] - Quote
Uh, you read my posts in reverse...
CB Vet // True Logi Bro // @KazeEyrou
Learning Coalition and RTG Mentor
|
Jack McReady
DUST University Ivy League
1669
|
Posted - 2014.12.29 08:41:00 -
[8] - Quote
Kaze Eyrou wrote:Uh, you read my posts in reverse... I am a strange person :) |
Jack McReady
DUST University Ivy League
1676
|
Posted - 2014.12.30 09:44:00 -
[9] - Quote
could some dev maybe come here and confirm that my observation is correct and why rounding is working in that way? especially because this totally screws my great fit I worked for and calculated in advance!
why is the suits CPU 514.5 round down to 514 when if rounded it actually would be correct to round up? and why is used CPU amount of 514.4625 double rounded UP to 515? |
Zindorak
G0DS AM0NG MEN General Tso's Alliance
1459
|
Posted - 2014.12.30 19:09:00 -
[10] - Quote
Wat 514CPU?
ILOOMINATI How you get that much Electronic skill only give me 10 cpu per level
Pokemon master and Tekken Lord
Give me da iskiez
Gk0 Scout yay :)
|
|
Jack McReady
DUST University Ivy League
1683
|
Posted - 2014.12.30 20:48:00 -
[11] - Quote
Zindorak wrote:Wat 514CPU?
ILOOMINATI How you get that much Electronic skill only give me 10 cpu per level there are two skills that increase CPU, one gives 1%, the other 5% per level. |
Sequal's Back
Les Desanusseurs
52
|
Posted - 2015.01.02 10:31:00 -
[12] - Quote
Megaman Trigger wrote:Also, why can a Proto Assault (in theory) fit all Proto yet the Sentinels got nerf'd for it? Because Basic Sent OP.
514 CPU is why MinAss is best assault ;)
Rise? That's what they used to call me. Sequal Rise. That was my name.
Now I come Back to you, at the turn of the tide.
|
Jack McReady
DUST University Ivy League
1733
|
Posted - 2015.01.07 08:25:00 -
[13] - Quote
broken rounding FTW. and the "fun" begins...
I used my AUR for some respecs to check how the rounding could potentially be "abused", e.g. full proto shield AV fit: Assault gk.0 Complex Light Damage Modifier Complex Light Damage Modifier Complex Light Damage Modifier Complex Ferroscale Plates Complex Ferroscale Plates Complex Ferroscale Plates Complex Armor Repairer Complex Armor Repairer Allotek Plasma Cannon Core Flaylock Pistol Core Locus Grenade Ishukone Nanohive
it uses 92.25 PG out of available 91.875. lolwut
can someone from CCP please look into it? |
Kaze Eyrou
Molon Labe. General Tso's Alliance
1572
|
Posted - 2015.01.07 14:06:00 -
[14] - Quote
Jack:
In order for CCP to fix this problem, we're going to have to nail the problem of where it's at.
From the info you have given me, it seems my hypothesis about the problem being within the Assault suit bonuses seems to be even more apparent. But you only gave the info which I assume is from Protofits while not telling us if that Assault gk.0 was valid or not and the exact amount of PG used.
If possible, I'd like to squad up with you sometime and troubleshoot with you to figure out what's going on with this weird bug.
CB Vet // Logi Bro // @KazeEyrou
Kaze's Helpful Links
|
Jack McReady
DUST University Ivy League
1734
|
Posted - 2015.01.07 14:29:00 -
[15] - Quote
Kaze Eyrou wrote:Jack:
In order for CCP to fix this problem, we're going to have to nail the problem of where it's at.
From the info you have given me, it seems my hypothesis about the problem being within the Assault suit bonuses seems to be even more apparent. But you only gave the info which I assume is from Protofits while not telling us if that Assault gk.0 was valid or not and the exact amount of PG used.
If possible, I'd like to squad up with you sometime and troubleshoot with you to figure out what's going on with this weird bug.
how often do I need to repeat that it obviously is THE ROUNDING and not any skill bonus. if skill bonus would not work the numbers would be far more off. it simply is rounded incorrectly on the minmatars assault CPU available CPU 514.5 rounded is 515. ingame it is 514 for what ever reason. required CPU 514.4625 rounded is 514. ingame it is 515.
result is invalid fit. even if you would double round the required CPU which is not the regular way rounding works it SHOULD still end with 515 out of 515 CPU and be valid, however it doesnt.
the gallente fit I just posted is working ingame and is not just from protofits. when I was fiddling around with my AUR respecs I gave it a try and played it a few matches.
with the gallente assault rounding seems to work fine => available PG 91.875 rounds up to 92. correct! required PG 92.25 rounds down to 92. correct! fitting is valid ingame |
KalOfTheRathi
Nec Tributis
1420
|
Posted - 2015.01.11 03:01:00 -
[16] - Quote
Old story, same results. CCP doesn't round. It truncates. No rounding up happens at all, AFIK.
This is an old MSFT coders hassle. MSFT tools all assume round down or truncate as the default. Many libraries assume standard rounding rules which we assume to mean .5 rounds up while < .5 rounds down. It is a valid option for setting how FP rounds. Options are Up, Down and Truncate. It does give more consistent values in some situations. MSFT has always rounded down at .5 as I recall, it has to be > .5 to round up when Up is the option. Still a valid operation.
Note that each and every step in a formula does this. Regardless if it is written out in a long command or separate statements.
When figuring out what is supposed to happen we post A * B * C * D. That however isn't accurate. The actual process is closer to this, with apologies for those who understand operator precedence:
Temp1 = A * B; Temp2 = C * D; Temp3 = Temp1 * Temp2.
Temp1 is rounded by truncation to level of significance. As is Temp2 **and** Temp3 calculations.
It results in the lowest possible value.
As stated above, it is a perfectly legitimate floating point library option. It is very user unfriendly, though.
Buyer beware.
My favorite tank is a Lightning. Just sayin.
|
Jack McReady
DUST University Ivy League
1744
|
Posted - 2015.01.11 08:52:00 -
[17] - Quote
KalOfTheRathi wrote:Old story, same results. CCP doesn't round. It truncates. No rounding up happens at all, AFIK.
what you are describing is not truncation, it is something made up. Truncation is equivalent to rounding towards zero which means the fit I have would end with 514 out of 514 and be valid. even if the game would round down at .5 the fit would still end up with 514 out of 514.
beside that, it is definately rounded correctly: TEST 2: Logistics mk.0 Enhanced Light Damage Modifier Enhanced Light Damage Modifier
CPU used according to Protofits: 76.5 CPU used according to Dust: 77 |
Orber Gen
Academy Inferno E-R-A
279
|
Posted - 2015.01.12 12:06:00 -
[18] - Quote
IN EVE and DUST, CCP always uses floor() function, where is no math rounding in dust Dust does negative rounding 514.5 => 514 |
Jack McReady
DUST University Ivy League
1746
|
Posted - 2015.01.12 12:16:00 -
[19] - Quote
Orber Gen wrote:IN EVE and DUST, CCP always uses floor() function, where is no math rounding in dust Dust does negative rounding 514.5 => 514 That is why sometimes we have bugs whe u use 60 PWG out of 60 PWG and it's invalid I think that total usage rounding by ceil(), but suit stats + skills rounding by floor()
you are totally wrong and it is obvious that you havent read anything in this thread. all the information is in my first post. the game does in fact round as shown per examples, it only doesnt for the minmatar assault.
also in EVE game takes into account decimals. |
THUNDERGROOVE
Fatal Absolution
1294
|
Posted - 2015.01.13 02:01:00 -
[20] - Quote
Internally, all CPU and PG are stored as integers except when calculations are being done in which case they are floats.
Floating point numbers suffer from rounding errors.
Because the floating point number could be 514.49... due to the inaccuracies, it gets rounded down when cast to an integer.
Dual tanking is for bad players.
Come play a better game.
|
|
Jack McReady
DUST University Ivy League
1774
|
Posted - 2015.01.13 08:23:00 -
[21] - Quote
THUNDERGROOVE wrote:Internally, all CPU and PG are stored as integers except when calculations are being done in which case they are floats.
Floating point numbers suffer from rounding errors.
Because the floating point number could be 514.49... due to the inaccuracies, it gets rounded down when cast to an integer. this theory does not hold because in this case the fit would end as 514 out of 514, I have done the math in this regards too, I just havent posted it because it would make the thread explode.
I just want CCP to look into it, thats all. |
Kaze Eyrou
DUST University Ivy League
1716
|
Posted - 2015.01.13 09:14:00 -
[22] - Quote
Jack:
I tested out another suit. The Sentinel M/1-Series gets bonuses that boost it 283.5. Due to the rounding on the dropsuits, the actual number of CPU Dust gives it is 283.
So your Minmatar suit rounding to 514 is correct unfortunately.
CB Vet // Logi Bro // @KazeEyrou
Kaze's Helpful Links
|
Jack McReady
DUST University Ivy League
1774
|
Posted - 2015.01.13 09:29:00 -
[23] - Quote
Kaze Eyrou wrote:Jack:
I tested out another suit. The Sentinel M/1-Series gets bonuses that boost it 283.5. Due to the rounding on the dropsuits, the actual number of CPU Dust gives it is 283.
So your Minmatar suit rounding to 514 is correct unfortunately. does not really matter because the fit would end with 514 out of 514 instead but it doesnt. |
Jack McReady
DUST University Ivy League
1905
|
Posted - 2015.01.25 13:23:00 -
[24] - Quote
I got some info because I filed a help ticket, it definately is not working correctly.
it works the following way: numbers are rounded first to to 4 significant digits, thus becoming 514.5 and since the fitting window can only accommodate whole numbers it is again rounded up to whole numbers
required CPU 514.4625 is first rounded to 514.5 an then becames 515. all fine.
now the gist is, available CPU is 514.5 and it should became 515 too but it isnt. |
rasputin900000
Subsonic Synthesis RISE of LEGION
27
|
Posted - 2015.01.26 10:41:00 -
[25] - Quote
Jack McReady wrote:broken rounding FTW. and the "fun" begins... I used my AUR for some respecs to check how the rounding works, e.g. full proto PLC assault fit: Assault gk.0 Complex Light Damage Modifier Complex Light Damage Modifier Complex Light Damage Modifier Complex Ferroscale Plates Complex Ferroscale Plates Complex Ferroscale Plates Complex Armor Repairer Complex Armor Repairer Allotek Plasma Cannon Core Flaylock Pistol Core Locus Grenade Ishukone Nanohive it uses 92.25 PG out of available 91.875. lolwut can someone from CCP please look into it and fix my minmatar fit?
Think you may have faltered your math due to stacking penalties |
Jack McReady
DUST University Ivy League
1912
|
Posted - 2015.01.26 10:43:00 -
[26] - Quote
rasputin900000 wrote:Jack McReady wrote:broken rounding FTW. and the "fun" begins... I used my AUR for some respecs to check how the rounding works, e.g. full proto PLC assault fit: Assault gk.0 Complex Light Damage Modifier Complex Light Damage Modifier Complex Light Damage Modifier Complex Ferroscale Plates Complex Ferroscale Plates Complex Ferroscale Plates Complex Armor Repairer Complex Armor Repairer Allotek Plasma Cannon Core Flaylock Pistol Core Locus Grenade Ishukone Nanohive it uses 92.25 PG out of available 91.875. lolwut can someone from CCP please look into it and fix my minmatar fit? Think you may have faltered your math due to stacking penalties your answer does not make sense.
beside that, I already got the answer from a GM and now know how it is done and that it is not rounded correctly in the minmatar assaults case. it should end with 515/515 CPU and be a valid fit. |
|
|
|
Pages: [1] :: one page |
First page | Previous page | Next page | Last page |