Repeat loop using floating point number

grasshopper

I am using a repeat loop that counts from 1 to 250 but 250 is in a floating point data type. I can say that it runs fine now, but will this cause problems on various propellers or any future problems that I may get hung up on?

Thanks

cgracey

The REPEAT loop works with integers only. The value 250 encoded as floating point is $437A0000, or 1,132,068,864 -- which would be a lot of loops.

It sounds like you just assigned the integer value 250 to a variable and used that for·the REPEAT, which would work fine.

To specify a floating point value, you must use a decimal point·or 'e' for exponent. For example, the floating-point value 250 would have to be specified as 250.0, or 2.5e2, or 25e1.

Such floating point values can only be manipulated at run-time by using routines designed to handle that data type, such as those in floatmath.spin and floatstring.spin.


▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


Chip Gracey
Parallax, Inc.

Mike Green

What do you mean by "floating point data type". There is no floating point data type in Spin or Propeller assembly language. You can write floating point constants which are converted to IEEE754 floating point representation, but these are treated as 32 bit integer values outside of constant expressions. You can write a repeat loop that uses the floating point function library to do its arithmetic, but a "for x = 1.0 to 250.0" doesn't work as you might expect.

The intention is for existing software to work the same (allbeit faster) on the Prop II as much as possible. Since the Prop II design is still in flux, noone can say for sure whether there will be changes required to the floating point library other than improvements using the new features (like hardware multiply).

grasshopper

Well I mean "Floating Point" as in below

   Nub1 := math.Ffloat(250)                             
   Temp1 := 0
   Temp2 := 0
   Temp3 := 0 
 
     
  Repeat X from 1 to Nub1                              ' Get temp 250 times loop // here i float the number 250
   
         /Getting data in here for Temp1
        Temp1 := Temp1+ Temp2                    ' Add it up
         
    Temp2 := Math.Ffloat(Temp1)                            
    Temp3 := Math.Fdiv(Temp2, Nub1)

Mike Green

Like I said, "Repeat X from 1 to Nub1" won't do what you expect it to do. math.Ffloat(250) converts the 250 (which is $FA in hex) to IEEE754 format which comes out to something like $43FD0000 in hex. The REPEAT works out to "Repeat X from 1 to $43FD0000" since that's all done using integer arithmetic.

If you're interested in how floating point works, read en.wikipedia.org/wiki/IEEE_754-2008 and www.validlab.com/goldberg/paper.ps