Armour GDR

[Marbit]

Have you ever noticed that an armour's GDR is almost four times its AC?
The error of this method is small except for the robe and the crystal plate armour.

[bcadren]

GDR doesn't affect much anything; it's just a guaranteed percentage of a max roll you will get instead of a REALLY HORRIBLE roll. Relatively small importance compared to the normal AC.

[duvessa]

imo 0.40 isn't almost four times 10

[twelwe]

it`s close

[Zooty]

imo 0.40 isn't almost four times 10

Give or take an order of magnitude or two.

[khalil]

imo 0.40 isn't almost four times 10

Give or take an order of magnitude or two.

That's OK in astrophysics, so long as you're within six orders of magnitude you'll generally be alright.

[damiac]

There needs to be another button besides thank for posts like this...

imo 0.40 isn't almost four times 10

FR: "Thanks, smartass" button on the forums.

[and into]

Well I'm not sure what marbit meant by the OP, but if this is just a mnemonic for remembering GDR then I guess it holds fairly well that base AC times 4 gives you the GDR as a percentage. (Or base AC times 4/100 for proper decimal value.) With robes and CPA being the exceptions.

But then the more important fact to remember is that people have weakened their characters, sometimes to the point of contributing to their deaths, by misunderstanding and/or overvaluing GDR due to bad advice, whereas probably no characters have actually benefited significantly from knowing said information and using it to make a decision.

(Theoretically if you have some really strange randart armor drops of different type but otherwise basically identical to choose from, maybe GDR can be the factor that tips the scales in terms of which is technically "optimal," but even in such theoretical cases, you are probably going to be choosing between two pieces of armor that are very good and appropriate for your character, either of which would work fine so it isn't a big deal either way.)

[damiac]

It seems to be true that in 99.9999% of cases, GDR isn't a worthwhile metric for making any choice in the game.

Knowing of its existence might explain some stuff, like why your caster in robes with Ozo's armor and 30AC seems to take more damage in melee than your berserker in plate with 30AC, but what are you going to do with that information?

If you ever find yourself in a situation like crate pointed out, where you can choose between two artifact armors that somehow give the same AC and EV, pick the one with more AC. But of course, even without GDR that's still the right answer.

[Sandman25]

imo 0.40 isn't almost four times 10

?

  Code:

    int gdr = 14 * sqrt(max(body_base_AC - 2, 0));

[duvessa]

imo 0.40 isn't almost four times 10

?

  Code:

    int gdr = 14 * sqrt(max(body_base_AC - 2, 0));

it gets divided by 100 when used; indeed, the function your line comes from is called gdr_perc(), not gdr()!

[Sandman25]

It looks like GDR is 40, not 0.40.

  Code:

int actor::apply_ac(int damage, int max_damage, ac_type ac_rule,
                    int stab_bypass) const
{
    int ac = max(armour_class() - stab_bypass, 0);
    int gdr = gdr_perc();
    int saved = 0;
    switch (ac_rule)
    {
    case AC_NONE:
        return damage; // no GDR, too
    case AC_PROPORTIONAL:
        ASSERT(stab_bypass == 0);
        saved = damage - apply_chunked_AC(damage, ac);
        saved = max(saved, div_rand_round(max_damage * gdr, 100));
        return max(damage - saved, 0);

    case AC_NORMAL:
        saved = random2(1 + ac);
        break;
    case AC_HALF:
        saved = random2(1 + ac) / 2;
        ac /= 2;
        gdr /= 2;
        break;
    case AC_TRIPLE:
        saved = random2(1 + ac) + random2(1 + ac) + random2(1 + ac);
        ac *= 3;
        // apply GDR only twice rather than thrice, that's probably still waaay
        // too good.  50% gives 75% rather than 100%, too.
        gdr = 100 - gdr * gdr / 100;
        break;
    default:
        die("invalid AC rule");
    }

    saved = max(saved, min(gdr * max_damage / 100, ac / 2));
    return max(damage - saved, 0);
}

[twelwe]

okay 40 is probably bigger than 10 we can all agree

[duvessa]

It looks like GDR is 40, not 0.40.

saved = max(saved, div_rand_round(max_damage * gdr, 100));

[Hirsch I]

it is interesting when I cant understand a single thing in the thread. I need to learn how to code.