Quadruple Precision Float Point Emulator

[coder0xff]

Several weeks ago I promised to release this. It's a library for doing high precision mathematics. I'm releasing it under the GPL license. See the readme.txt file.

http://dl.dropbox.com/u/5105767/QPFloat.rar

[Gary]

Can you tell us what exactly this allows us to do? Is it something to do with huge objects(planets)?

[Rasky_26]

This is frickin' awesome! Seriously coder, the things you do blow me away.

@Gary: The more precision that is utilized, the more accurate the results. This is because computers usually only go to X decimal points and round off when they hit that limit. For standard uses this poses little issue; however, this rounding becomes more apparent when you are using something like a weather model and it is cycling through these numbers many million times. A small error becomes more pronounced the further along something propagates and leads to a greater error distribution. Something with quadruple precision point would greatly lessen such errors and would essentially allow a model to run much longer under a greatly lessened error curve.

[Gary]

Ah, I see. Thanks Rasky.

Good work Coder, though I cannot use this, I am sure it will help others.

[omnicoder]

Is this designed in a way that I could fairly easily change it to give sextuplet precision etc or is it hardcoded to quadruple?

[coder0xff]

It is somewhat flexible. The underlying mathematical operations and block copies can work on any size. I wouldn't go so far as to say it's trivial though. At that point, you may just want to go with an arbitrary precision library.

@Gary
Yes, it has a lot to do with huge objects(planets or even the largest stars or bigger)

I've update the download to 0.2 beta just now. I added Tan, ATan, ASin, ACos, and ATan2 at the request of my father. Also filled in some missing function definitions. The download also includes the debug and release binaries now for convenience. If someone would like to follow the ****** USING ****** part of the README.txt and try some tests, I'd greatly appreciate it. (I dare you to divide by zero )

Some example C# code

Code: Select all

Quadruple myVariable = 5;
Quadruple anotherVar = Quadruple.Pow(myVariable, 1000);
String whatIsThat = anotherVar.ToString();
int tryTruncation = (int)anotherVar;


[BattleGrunt]

Excuse my ignorance, but what kind of applications to modding would this emulator help? You mentioned the weather model earlier.

[no00dylan]

Coder, if the world was full of coding nerds you would be crowned king. But nope, god fearin' folks like me just sit here and go duhhhhhhh make something explode.

[coder0xff]

you guys make me lol.

@BattleGrunt
This has no usefulness for modding purposes. I've created it as a component of my own engine.

@dylan Thanks for the vote of confidence. <3

[BattleGrunt]

@coder0xff

Ahh, cool

[zombie@computer]

Just interested: what was your father needing it for?

[MELVIn]

Doomsday device!

[coder0xff]

Just interested: what was your father needing it for?

A toy, lol. He doesn't need it, he just thought it would be fun to use. For his application he needs the "haversine formula" which uses an arcsin. He also convinced me that it would make the library more impressive.

[Deathbagel]

I came here because the title sounded cool. I only wish I understood the things that you did enough to appreciate them coder, one day, one day........