Page 1 of 2

64bit precision floats

Posted: Mon Jun 14, 2004 6:29 pm
by ^OO^
Hi everyone :-)

A year ago I set out to write a novel calculator but could not get the precision I needed because the floating point values are only 32 bits and I need to use variables of at least 64bits, preferably 80bits. So I dropped PB.
Blitz has the same restriction.

The new IBasic Pro, on ther other hand, CAN do doubles.
But, before I commit to learning IBasic, can you tell me whether PB's math precision has been upgraded yet? (my search failed to turn up any mention of such an upgrade)

Thank you for you kind attention.

Julian

Posted: Mon Jun 14, 2004 7:17 pm
by blueznl
nope, not yet

there is a lib that does it, however

Posted: Mon Jun 14, 2004 8:19 pm
by GPI
one thing:

Floats (and doubles) are not exact values. A PC can't convert a .4 in binary-system. But we have the same with problem with 1/3.

When you need really exact result: Don't use floats and doubles!

Posted: Mon Jun 14, 2004 11:21 pm
by Froggerprogger
As a further alternative you might try calculating with fractional digits. See here for the procedures:
viewtopic.php?p=44599#44599

64bit precision floats

Posted: Tue Jun 15, 2004 3:56 am
by ^OO^
Thank you for your responses ^_^

There would be lots of high-value variables to process and wrestling with library bodges does not appeal to me. PureBasic's weakness is in the very area where I need it to be strong.

I imagine that the original motive for an all-low-precision-math language was to make it go fast?

FPUs are now optimized for 64bit FLOPS and these are now faster than 32bit FLOPS. There is no reason to retain 32bit floats except that you can't be bothered to update them.

I put in feature requests for better precision. I read responses from other users pointing out that I don't need precision or, if I do, I should make library files.

I waited a year to see if PB would recognize the NEED to provide 64bit doubles. It hasn't happened. I guess that nobody else lobbied for the upgrade so it got no priority. Does this mean I'm the only user that needs real-world math?

Such a shame: PB seems destined to remain a TOY, not a TOOL. I'll look in again next year with my fingers crossed.

I'm sorry if I seem a little acerbic - it's the disappointment.

So then, it's decided: I'm going with IBASIC.

Good bye.

Julian

Posted: Tue Jun 15, 2004 4:10 am
by Shannara
I knew that avatar looked familiar. Anyways, agreed. IBasic you say? Now I have to look. Thanks.

Posted: Tue Jun 15, 2004 8:30 am
by PolyVector
@^OO^

Purebasic is by no means a "TOY", and to say so based on one missing feature is laughable...

So the only language that meets your expectations came out, when was it? This month? I hope you enjoy posting bugs and requesting features much more in the IBasic forums...

Good Luck! :roll:

Re: 64bit precision floats

Posted: Tue Jun 15, 2004 9:00 am
by PB
> Such a shame: PB seems destined to remain a TOY, not a TOOL

It's planned for the future, possibly with v4.00 from what Fred has hinted
at in the past. If that's too far away for you (which it appears to be) then
by all means use IBASIC instead if it suits your immediate needs. But it
by no means that PureBasic is a "toy" language -- there are dozens of
professional apps and games written in it.

64bit precision floats

Posted: Tue Jun 15, 2004 1:49 pm
by ^OO^
Nice to hear that their is hope for the future.
Press on, Fred - - - 8)

Julian

Posted: Tue Jun 22, 2004 9:39 pm
by geoff
Maybe not a TOY, but very restricted in serious application.

For example, you cannot calculate a GPS position, predict an eclipse or even do many financial calculations. We are not talking rocket science here, just mainstream applications.

I could go on, but I have said enough on this topic already.

I cannot understand why Fred has given this such a low priority. He could make the language suitable for the majority of applications with this single addition.

Posted: Tue Jun 22, 2004 11:11 pm
by Kale
I cannot understand why Fred has given this such a low priority. He could make the language suitable for the majority of applications with this single addition.
I would very much like to see this too as soon as poss, it would even make calculating disk sizes (yes even that simple task) so much easier.
Image

Posted: Wed Jun 23, 2004 8:50 am
by LarsG
^00^: dude.. you need to change your avatar.. I'm barely able to read the posts!!! 8O

Posted: Wed Jun 23, 2004 12:40 pm
by ^OO^
LarsG wrote:^00^: dude.. you need to change your avatar.. I'm barely able to read the posts!!! 8O
OK - Done!

I hope you copied it first - - - :oops:

Posted: Wed Jun 23, 2004 1:10 pm
by LarsG
^OO^ wrote:
LarsG wrote:^00^: dude.. you need to change your avatar.. I'm barely able to read the posts!!! 8O
OK - Done!

I hope you copied it first - - - :oops:
:lol:
darn, I knew I forgot something... :wink:

Posted: Wed Jun 23, 2004 2:27 pm
by Rings
LarsG wrote:
^OO^ wrote:
LarsG wrote:^00^: dude.. you need to change your avatar.. I'm barely able to read the posts!!! 8O
OK - Done!

I hope you copied it first - - - :oops:
:lol:
darn, I knew I forgot something... :wink:
of course, i saved it to use later for the lesbian amazon chatroom :)