fi.sum increases bitlength although fimath is set to KeepMSB

Hi there,

I was wondering why the sum function increases the length of my fixed point variable.

Here is a little example:

fimath('SumMode','KeepMSB')
x = fi(1:100,true,8,0)
.
.
.
y = sum(x)
y = 5050

            DataTypeMode: Fixed-point: binary point scaling
              Signedness: Signed
              WordLength: 15
          FractionLength: 0

I actually expected y to be 39, which is the decimal representation of 00100111 (the first 8 bit of y). It seams that the sum function just works with FullPrecision anyways?! Am I doing something wrong here?

Any help is highly appreciated.

Regards,

Len

Answer 1

[Walter Roberson]

You did not specify SumWordLength in your fimath() object.

[Lennart]

Allright thanks for that I sort of overread that I guess. Unfortunately it still does not work. Here is what I did:

rounding = 'nearest';  
overflowAction = 'Saturate';
sumMode  = 'KeepMSB'; 
productMode = 'KeepMSB';
wordLength = 8;
fimath('RoundingMethod', rounding, 'OverflowAction', overflowAction, 'SumMode', sumMode, 'SumWordLength', wordLength, 'ProductMode', productMode, 'ProductWordLength', wordLength );

but if I type fimath in the command window it returns the default (sumMode = FullPrecission etc) Am I assigning fimath wrong?

Regards,

Len

[Walter Roberson]

I think you might need to assign the fimath object to a variable to get it to work, so that the object exists in the workspace. When you do not assign it, the object would get discarded at the end of the line.