Power of X

[Jim P]

LabVIEW returns the wrong value when using the Power of X function.

x=2, y=63: result should be bit 63 should be 1, bits 0-62 should be 0.  Labview returns the wrong result when compared to Microsoft calculator

[altenbach]

Well, you need a data representation with at least 63+ bits (in the mantissa). Try U64. 😉

For powers of two with integers, use the "other" tool. It does simple bit shifts and is thus very cheap.

"Power of X" works internally with floating point numbers. Fortunately, it produces the correct result in this special case. Just display it in a U64 indicator.

Can you show an example where it fails?

Message Edited by altenbach on 01-26-2007 11:35 AM

[Jim P]

I'm using LabVIEW 7.1.  Was U64 introduced in a newer version.

[altenbach]

Yes, you need at last LabVIEW 8.0 for 64 bit integers.

(Of course you could write your own "infinite precision" tools, e.g. by representing each number as an arrray of shorter integers).

As an example, have a look at the results of the factorial coding challenge (http://zone.ni.com/devzone/cda/tut/p/id/5299). For example the result of 10000! is a number that needed to be correct with all 35660 decimal digits. We had entries that could calculate all in uder 100ms. 😉

[TCPlomp]

That's not too strange, the datatype (dbl) uses a mantissa of 53 if i'm correct. So only the 53 MSB are correct the rest is rounded off...
If you use the following code with U64 you'll get the correct answers.



Ton

Message Edited by TonP on 01-26-2007 08:54 PM