how to round down rather than automatic round up at .5

[k-waris]

I really have lost my brain cells to work out suitable formula to achieve this!

I need at all cost be able to round down on the result of an equation. Obviously Labview compliant with IEEE would always round up at/above .5. Is there a way to achieve what I am after?

Many thanks

K Waris

[crossrulz · Accepted solution]

Does Round Toward -Infinity do the job?

[mikeporter]

Define .5. What about .4999999999999.

Up or down?

There is no such thing as exactly .5.

Mike...

[Bob_Schor]

---

mikeporter wrote:
There is no such thing as exactly .5.

---

Sorry, but that's not correct.  Dividing any integer by 2 gives a floating representation of that integer times 2^(-1)..  To see the "exact" representation of, say, 9/2, typecast it to an array of U8 (bytes), and you'll see 64 18 0 0, exponent and mantissa.

BS

[billko]

---

@k-waris wrote:

I really have lost my brain cells to work out suitable formula to achieve this!

I need at all cost be able to round down on the result of an equation. Obviously Labview compliant with IEEE would always round up at/above .5. Is there a way to achieve what I am after?

Many thanks

K Waris

---

Actually, the statement in bold is untrue.  IEEE standard for "round to nearest" is "round to even integer", which is the rule that LabVIEW follows.

http://zone.ni.com/reference/en-XX/help/371361H-01/glang/round_to_nearest/

[k-waris]

That solved my problem. Thank you

[tst]

Note that this method might not work like you want for negative numbers. If you want -3.2 to round to -3, you will need to check the sign and if negative round to +inf.