different results at different speeds and on different...


#1

There is a challenge in the challenge and tournaments section with an issue. The same two fleets, using three different computers and run at 1x and 4x, are having drastically different outcomes. Going from 80/0 to 0/80 and everything in between. Does anyone have any ideas on causes or ways to avoid?

viewtopic.php?f=24&t=3297&p=17709#p17709 check here for the details.


#2

Yeah, I can see how this could be a big thing. Sounds like GSB needs it’s timer code revised to be a bit more robust.


#3

timer code…?

this is almost certainly just a result of what we found when modding weapons outside the envelope of normality … the simulation engine runs in lockstep with the display engine. so if you run at 4x speed, bullets move 4x the distance between each collision detection point… and they can leapfrog ships they’d otherwise have smacked into.

nothing to do with timers :slight_smile:


#4

err, i thought accuracy was calculated not simulated


#5

so, snowdrift, does this mean the issue is being addressed by cliffski? i missed the result of that thread. also, can this be avoided?


#6

one other thing, does the thing you were referring to also take into account that three different computers got wildly different results at the same speed, and also have result difference based upon optional effects usage. (or one of us did at least.)


#7

if a projectile is simulated to impact a ship, an accuracy calculation takes place to determine whether it actually hit or not. if the simulation doesn’t see the projectile hit … accuracy is never calculated.

if a plasma bolt is approaching a ship, this (i believe) is what happens:
1x speed:
__OO
__
OO
__*OO
__*OO
__*OO
___XO (impact)
2x speed:
_OO
___
OO

__*OO
___OX (impact)
4x speed:
OO
_*OO
____OO
(miss)

obviously this is more noticable with faster projectiles/smaller ships … like fighters…


#8

i’m sure he’s aware of it. i don’t know that it’s been publicly acknowledged:
viewtopic.php?f=23&t=2735&start=15
viewtopic.php?f=19&t=3087&start=60


#9

Sadly this cannot be fixed without a major rewriting of a huge chunk of the game. I’ve spent all morning getting my head around making the games deterministic. I got as far as 9 seconds of truly reproducible playback, but then a single thing threw it off. In this case, it was a missile that launched its decoys on one playback just fractionally before it did on another one, which throws the entire simulation off.
Although I understand and appreciate the desire for 100% deterministic battles, I don’t think its likely to happen, due to the major re-engineering required.
In retrospect the game should have been designed this way from scratch, but it wasn’t :frowning:


#10

End result: those doing tournaments will need to specify what speed the games will be played at.

It seems from the descriptions above that speeding things up gives the advantage to faster firing weapons, is this correct?


#11

I don’t know how deep is random in your code but why not emulate probability with counts (I’ll try to explain, sorry for bad English).

Suppose a turret has a 20% probability of a hit this means that 1/5 shots should hit its target. Instead of making random before each shot, make a counter that trace when last hit occurred. It will be something like miss, miss, hit, miss, miss, miss, miss, hit, miss… etc.

Of course is not so simple as it seams, but I hope you get the idea.

:slight_smile:


#12

Are these two fleets pretty well matched? In that case I would expect wins to be back and forth on successive tries even on the same machine with the same settings.

Now if two completely mismatched fleets… where one is built poorly and the other is built very solidly specifically to counter the first… if THOSE two had highly varying results… then I would be worried. However even in that scenario, it’s possible for a few lucky hits in a row to carry the day for the underdog. Anyone who has played a table top dice game can surely understand that.


#13

uhh… how’d’ya work that?

the reload time of a weapons is irrelevent. the only relevant factors are the speed of the projectile, and the speed of the target. and the higher those get, the more likely the projectile is going to overstep-miss the target.

no weapon gets an advantage from speeding time up - but some will be more disadvantaged than others.


#14

Oh, well. Save it for the sequel I guess. :stuck_out_tongue:

I am going to assume yes.


#15

A better question might be which weapons are not disadvantaged?


#16

It sounds like, if the mechanism is correctly identified, that it gives the advantage to slower traveling weapons. Aka missiles, rockets, maybe plasma. Is this the case?


#17

In the challenge that brought this issue up, the major change I saw between running at 4x and running at 1x was how long three squads of fighters survived, and how many cruisers those fighters destroyed. At 1x speed the fighters got swatted out of the air almost immediately, only destroying one cruiser. At 4x speed the fighters destroyed 4 cruisers before being destroyed.

These fighters were fast, but not crazy fast, I think their speed is 2.3 or so. Regardless, their speed and the speed of weapons hitting (rockets) them seems to be the biggest variable.

As far as consistency goes, I ran the challenge 3 times at 4x and got almost identical results each time: 20% Alliance, 6% Empire.

As I understand it, others weren’t getting as consistent results, but were still pulling empire wins every time. At 4x they reported somewhere around 0% Alliance, 55% Empire.

Anyone interested in doing some scientific research on this one? :slight_smile:


#18

that’s what i thought would be the case - fighters. they’re the fastest thing around, and the smallest.

and if being chased by rockets, well - there’s the projectile part of the equation.

i would (personally) bet that if you replace anti-fighter rockets with beam weapons, you’ll get nice repeatable results.

… to answer the other query - beam weapons will be totally unaffected. everything else will be affected relative to the speed of its projectile… though for almost all instances, it’ll make bugger all™ difference.

[all this presumes that the mechanic i hypothesised is correct of course! :)]


#19

I am restarting the tournament. It isn’t getting fixed, so we might as well use what we have. All things considered, it sounds like we should be running the game at 1x… figure I will just have to walk away and do something else while testing. Was going to do 4x, but it sounds like there are flaws at that speed.


#20

Agreed, 1x is likely much more consistent.

Would you mind testing the MDK challenge on your machine at 1x? I got something like 0% alliance to 80% empire. If these match up, then we should be good to go :slight_smile: