split a decimal number into 2 numbers

[ThSa]

Simply change the format specifier constant manually 🙂

[LabViewGuruWannabe]

altenbach is correct. The Q&R handles negative numbers in an unexpected way. But this too is easily handled.





There's more than one way to build a cat.


Message Edited by LabViewGuruWannabe on 11-18-2007 03:18 PM

[blaz-kelbl]

I undestand this is an ancient topic, but for reference, this should be a free implementation

 

If we're talking pure bitwise split the snippet works. 

Regarding negative numbers: in the dual complement, negative numbers can be read as "whole part rounded towards -infinity" + (1 - decimal part):

Bitwise split works as intended. Additionally, looking at the split as values: if you add the Whole FXP and Decimal FXP value is equal to FXP value. 

[GerdW]

Hi blaz,

---

@blaz-kelbl wrote:

this should be a free implementation

---

What's "free" in this implementation?

Why go from FXP to bitarray to integer to FXP?

---

@blaz-kelbl wrote:

Bitwise split works as intended.

---

Sure.

But when working with integers you don't need any bitarrays, use plain boolean operations (AND, OR, shift) instead!

To put it in other words: you only need blue wires!

Edit: added the orange way of getting quotient and remainder…

[blaz-kelbl]

I meant free as in uses no resources, but your implementation is actually more elegant and should also use no resources (maybe I missed a post). 

The back conversion was just to illustrate. 

[GerdW]

Hi blaz,

---

@blaz-kelbl wrote:

I meant free as in uses no resources,

---

This only is valid on LV-FPGA.

In other environments (Win, RT) a bitarray uses way more memory than your underlying FXP datatype (or INT or FLOAT)!

[blaz-kelbl]

Correct, I mistakenly assumed the question was in the context of FPGA

(I think I just replied to the incorrect forum post) https://lavag.org/topic/23873-how-to-separate-decimals-and-integers-on-the-fpga-side/

Anyway, your solution is best.