Cheat Engine

Noobrzor · Posted: Tue Jan 12, 2016 2:16 pm Post subject: +Inf, NaN, float and double.

Greetings

First and foremost, how can a game deal with high numbers including fractions other than basic types?
Example being one flash game (I know this isn't a flash side of forum, I'm just asking about the specific mechanic of it) where it's plain and simple 4 bytes with a certain 'value offset' and typical flash multiplication of 8.

But with amounst beyond certain point, such as positive over 2147483647 as well as fractions the thing get's weird, and it's neither float nor double.

But main question is what's the difference between float/double +Inf value and NaN? As in how does that work out in most games, at least the ones that you (potential reader) have had the pleasure of interacting with? The reason behind this is NaN is a very crash-happy thing, how different is +Inf?

I noticed for certain values in float, and I'm presuming double as well as I had been dealing with it a little less often, get stuck in a 'range', such as 2147483648 for float. And using up little amounts in game doesn't change, as if the value was inherently frozen.

How does that work out, too? Is it like the bigger non fraction float amount the bigger the range of the 'frozen' constant rounding up?

Best Regards, and thanks a lot for your time!

ZacTheSin · Posted: Tue Jan 12, 2016 6:39 pm Post subject:

Hmm, probably a good example of this is Clicker Heroes, however it does use Double values.

In the end-game you're no longer using it's weird arbitrary value counters showing billions/trilliions etc and it moves to scientific notation.

E.g. 6.445e65

6.445e35 or 6.445×10^35 = 644500000000000000000000000000000000

It's something I was trying to work out with the mega values as well, with little luck due to me derping my save (numbers too big = insta crash on load).

Kind of past the initial block where it starts using scientific notation by default to store the value, I think it must use a secondary address to give the value up to another point.

like in the same way you have EAX/AX/AH/AL

Zanzer · Posted: Tue Jan 12, 2016 6:59 pm Post subject:

Yea, just throw more bytes at it. Feel free to check out Java's implementation.

Also, for +Inf, any normal number "mathed" with +Inf equals +Inf. So you can subtract all you want from it.

NaN is... not a number. So when it is "mathed" an exception is thrown and you crash.

Noobrzor · Posted: Wed Jan 13, 2016 9:18 am Post subject:

Redouane · Posted: Wed Jan 13, 2016 9:55 am Post subject: