Page 2 of 2
Posted: Tue Oct 21, 2003 12:15 pm
by Psychophanta
Situations like mikew's one probably will be more and more common.
Road Runner said:
The FPU does everything in extended precision anyway, there is no speed difference.
And it's truth, so why not double precision and double extended, inclusively 128bit if choosing MMX or 3DNow compilation modes?
Even some programmers don't need, it would never be a surplus in PB.
And of course JoRo, i personally don't use EditorGadgets now.
If PB had able to manage unlimited string lenghts some problems would go away.
Abolish single precison floats?
Posted: Tue Oct 21, 2003 9:04 pm
by mikew
From my point of view, I would welcome the abolishment of single precision floats in favor of the double floats - to keep it simple... Does anyone need single precision?
Posted: Tue Oct 21, 2003 9:09 pm
by Karbon
YES!
I mean, NO!
I mean, do I?
Stop confusing me!
Posted: Tue Oct 21, 2003 11:18 pm
by Psychophanta
Karbon, if i remember finely, you made a PB-lib or a dll for big length float numbers...
So, come on, don't be shy, don't be confused and give your sincere opinion...
Posted: Wed Oct 22, 2003 11:53 am
by Psychophanta
Sooorry, I was wrong
It was an app by Siegfried Rings, called BigNum (including a dll: bignum.dll) which i found at Paul's site, and which allows to manage big floats numbers.
I don't know why, but I confused Karbon with Siegfried Rings.
Posted: Wed Oct 22, 2003 2:06 pm
by Rings
Psychophanta wrote:I don't know why, but I confused Karbon with Siegfried Rings.
maybe i like eating Carbonara ?
also some good news on the BigNum. next week i got some free time and the lib will be updated and become
free of use for the purebasic-users.watch out ......
Posted: Wed Oct 22, 2003 2:34 pm
by Psychophanta
Thanx, and it will be welcomed.
Please tell us where we can get it.
At least I will use it, i need it for my projects.
Posted: Wed Oct 22, 2003 3:53 pm
by blueznl
double ints? yippie
Posted: Thu Oct 23, 2003 10:29 pm
by geoff
Karbon wrote:
I don't think anyone is arguing that they're not needed - just that they're not needed for money calculations!
Yes they are.
How would you calculate compound interest at 7% pa on a sum of 1000 Euros for 217 days?
Or calculate annual equivalent rate on a 10 year loan with fixed repayment?
Or predict discounted cash flows for a business plan?
Many serious finance calculations use transcendental functions.
mikew wrote:
From my point of view, I would welcome the abolishment of single precision floats in favor of the double floats - to keep it simple... Does anyone need single precision?
Yes, I do.
Actually, I made your suggestion a while ago but then changed my mind.
I use single precision float arrays for image processing applications. These images are very large, so doubling the array sizes with 8 byte variables would be a significant problem for me.
Posted: Fri Oct 24, 2003 12:04 pm
by Froggerprogger
And another one:
If a person has 20.000.000 Dollars at your bank, that uses single-precision-floats and pays in 3 Dollars, he will have 20.000.004 Dollars immediately. Damn. But don't care, because when he will have then 70.000.000 dollars and pays in 3 dollars again, he will still have only 70.000.000 dollars.
Code: Select all
startmoney.f = 20000000
endmoney.f = startmoney + 3.0
Debug startmoney.f
Debug endmoney
startmoney.f = 70000000
endmoney.f = startmoney + 3.0
Debug startmoney.f
Debug endmoney
edit: The problem then would be that he could withdrawn 3 dollars from his 70.000.000 Dollar-account and will still have 70.000.000 . :roll:
Posted: Fri Oct 24, 2003 1:01 pm
by freedimension
Froggerprogger wrote:edit: The problem then would be that he could withdrawn 3 dollars from his 70.000.000 Dollar-account and will still have 70.000.000 . :roll:
phew, what a money-machine
Posted: Fri Oct 24, 2003 2:14 pm
by Psychophanta
Until now, i've noticed that PB in all its versions lacks in matter of floats.
Lets hope good news about it for next version, and now we can use the Rings bignum.dll.