As a French person I feel like it's my duty to explain strikes to you. - AdrienIer

Create an account  

 
Test games played

Ok, 9 chimera (+219 combat skill, 9 chaos books, 1 life book, 2 sorcery books) vs 6 spiders and 3 stone giants (neutral)

1 chimera:
4 figures : Melee 10, Fire Breath 6, +1 To Hit, Defense 5, Resistance 8, Moves 3, 8 health
Melee Offense: (10-2)*4*20*1.32 + (6-2)*4*20*1.32 = 1267
Defense: 4*4*8*(5+2) = 896
9 chimera: 11407 melee offense, 8064 defense
But the chimera are damaged; between the 9 units they've taken 34 points of damage. So multiply both offense and defense by (288-34)/288.
9 damaged chimera: 10116 melee offense, 7112 defense

1 spider:
4 figures : Melee 6, +1 To Hit, Defense 5, Resistance 9, Moves 5, 5 health, Web spell, Poison 4.
Melee Offense: (6-2)*4*20*1.32 + 60*4 = 662
Defense: 4*4*5*(5+2) = 560
6 spiders: 3972 melee offense, 3360 defense

1 stone giant:
1 figure : Melee 20, Ranged 20 (2 ammo), +2 To Hit, Defense 12, Resistance 9, Moves 3, 25 health, Stoning Immunity, Poison Immunity, Wall Crusher, Mountaineer.
Melee Offense: (20-2)*1*20*1.65 = 594
Ranged Offense: (20-2)*1*20*1.65 = 594
Defense: 4*1*25*(12+2) = 1400
3 stone giants: 1782 melee offense, 1782 ranged offense, 4200 defense

Total Nature Lair: 5754 melee, 1782 ranged, 4200 defense


In 10 rounds of combat, the nature lair can expect to average 980 damage. However, for the purposes of this example, we'll assume they do max damage every ranged round, and actually do 1782 damage.

We ignore spellcasting again, and we see that the chimera start the melee phase with 10116*(7112-1782)/7112 melee offense, and (7112-1782) defense, or 7581 melee offense and 5330 defense.

Compare to the nature lair (5754 melee offense and 4200 defense), we can see that WITHOUT any spellcasting, AND assuming the stone giants roll max damage for all 10 rounds of ranged combat, the chimera should handily win this. They won't win every time (if the nature lair manages to average roughly 65% better rolls than the chimera, the nature lair will win; that means if the nature lair rolls ~2 higher on EVERY SINGLE ROLL, then the fight will be even; if they roll ~3 higher on every single roll, then the nature lair should win. Not impossible, but certainly improbable.)

I've only won 5/7 attempts so far for this fight.

However, there IS spellcasting. Note that the spellcasting for me at this point is pretty obscene:
9 Chaos books: 9*150 Attack*(219/10/5) = an extra 5913 attack first the first 10 rounds of combat
1 Life book: 30 PercentATK*(219/10/100) = +0, as there are no ranged units for the percentage to increase.
1 Life book: 40 Defense*(219/10/25) = an extra 35 defense added for the first 10 rounds of combat
2 Sorcery books: 2*30 Attack*(219/10/5) = 263 extra attack first 10 rounds of combat
2 Sorcery books: 2*30 PercdntATK*(219/10/100) = +0 as there are no ranged units for the percentage to increase.
2 Sorcery books: 2*30 Defense*(219/10/25) = 53 extra defense added for the first 10 rounds of combat

So with the modified defense, the chimera now have 880 more defense added (total: 7992)
The chimera also have a ranged attack strength of 6176 to fight back.

So during the ranged phase, this is now:
Chimera ranged strength: 6176 offense, 7992 defense.
Nature lair ranged strength: 1792 offense, 4200 defense.

Assuming the nature lair rolls MAX and the chimera roll MIN every round, the ranged phase ends with SOMETHING like:

The nature lair has 3696 defense left, and therefore, 5064 melee strength.
The chimera has 6394 defense left, and therefore, 8092 melee strength.

Now, the nature lair CAN still win this; but they now need to roll ~3 times higher than the chimera every round. (If Chimera roll 1, nature lair needs 3; if chimera roll 2, nature lair need 6; if chimera roll 3, nature lair need 9). However, recall, in order to even do THIS, we assumed that the nature lair rolls a 10 EVERY ranged turn, and the chimera roll a 1 EVERY ranged turn. That's.. ridiculous (1/100^10). If the ranged rounds were actually average, then the nature lair CANNOT win the melee phase.


Therefore, the fact that the chimera lost 2/7 games is .. incredible.

(I did all this so I didn't have to search back for my save game to find the combat that triggered this, and instead just give you the save file of the turn I happen to be on.)

I'll post the save file shortly.
Reply

This should be the right save.  There's 2 islands in the bottom left; the right hand of those 2 islands has a stack of 9 chimera (and a second group of 2 or 3 chimera) next to the nature lair.


Attached Files
.gam   SAVE1.GAM (Size: 151.96 KB / Downloads: 1)
Reply

Ok, let's see the numbers in the game. Before start of the combat :

Your Defense : $2282
Your Ranged : 0
Your Attack : $28D4
Your HP : $F8
Enemy Defense : $23D4
Enemy Ranged : $6F6
Enemy Attack : $167A
Enemy HP : $C3

Your Mana : $DB
Your Skill/Turn : $24
Your Magic Attack : $582
Your Percentage Attack : $4
Your Magic Defense : $14

Enemy Mana Available : 0
Enemy Skill/Turn : $8C 
Enemy Magic Attack : $8C
Enemy Magic Attack Percentage : $94
Enemy Magic Defense : 1
Note that if mana is not available, the other fields are not filled or used so they are random.

These numbers look about right for me, and the two sides are nearly equal.


Okay, I think I have a suspect.

Here : 
Quote:mov     ax, [bp+MagicAttackBonusAttacker]
imul    [bp+SkillperTurnAttacker]
mov     bx, 5
cwd
idiv    bx
mov     [bp+TotalAttackBonusAttacker], ax

That's $24*$582 = $0000C648, stored in dx:ax.
However there is an extra "cwd" there, "convert word to double". What that instruction does is, it treats ax as an integer value and extends it into a 4 byte integer. Basically it turns your $C648 into $FFFFC648, interpreting it as a negative number.
This instruction is entirely unnecessary and breaks every number where the highest bit in the lower word is 1.
It was there in the original, so I guess the source that was complied used integer types - and the compiler without overflow checking on just forced the integer type on it instead of reporting an overflow error.

Lots of thanks for finding this one, it is a quite elusive bug!
Reply

One other thing I was thinking about. Maybe the 50-100% range for item value in treasure it too wide.

When you are supposed to get a 3000 value item, getting 1500 is a major disappointment - the quality difference is ridiculously high.
Reply

I'm not sure, but in certain cases it was turning magic attack into a negative number?

That's definitely elusive all right. I'm kind of concerned about what that has been doing to ai wars.
Reply

All 3 magic categories, not sure if the other two can have a high enough number for it to happen though - probably not. It turned it to negative or cut the highest digit(s) off or both.
Reply

Which explains why I always felt like it was chaos - they get the highest number in one category the fastest. And of course the early game it wouldn't apply (since it's based on combat skill, it would also take much longer depending on your strategy). Neat!

Which actually means, it shouldn't have been a HUGE problem on ai wars, because it would only be late game (as defined by casting skill), and in many cases the AI it might affect would be dead already, or allied with each other. It was certainly still probably causing issues, but at least not huge ones.
Reply

For the magic items; I'm inclined to agree. You really want absolute ranges, not percentage ranges. Like, 0-250, 250-500, 500-750, 750-1000, 1000-1300, 1300-1600, 1600-2000, 2000-2500, 2500-3000
Reply

My surprisingly good chaos high elf start game started stalling because all 3 AI wizards are allied/wizard pact to each other and declaring war on me (is this common? none of them have aura of majesty!!). Hanging on with a city producing top-tier unit, a bunch of myrran settler spamming, and the right coordination of fire-giants, gargoyles and low-tier units protecting towns.

I did manage to conquer 2-3 towns with fire storm and the mid-tier summons and top-tier elf units, but now that the top wizard declared war on me (after the other 2) I'd have to focus even more on the defensive.

Around 1409, 2nd place.

Reply

It's not very common but it still does happen. If the enemy wizards have similar realms, peaceful/lawful personalities, charismatic, or more of these at once, that can result in that. Also higher difficulties generally means it happens more.
Reply



Forum Jump: