MULTIPLIER Inpots

[Bar Nash]

Hi all

When a multipler is designed like those used in FIRS will the nature of the
multiplicands influence the structure of it ?

I mean in case those are both fractional numbers <1 and in case both are
whole numbers .

I hope I described clear enough the problem .

Thanks in advance
EC

 

[KJ]

On Oct 8, 8:34 am, "Bar Nash" <waves...@netvision.net.il> wrote:

Hi all

When a multipler is designed like those used in FIRS will the nature of the
multiplicands influence the structure of  it ?

I mean in case those are both fractional numbers <1 and in case both are
whole numbers .

It shouldn't. I'm assuming your question is targetted towards
multipliers meant to be synthesized, that being the case the first
step would be to define how you plan to represent numbers in general.
Depending on the particular needs of your FIR filters the numbers will
likely be either fixed point or floating point. Fixed point
representation will consume less logic than floating point but either
can be synthesized.

If you use the relatively newly standardized fixed or floating point
packages as the basis for representing 'numbers' then the answer to
your question is 'no, the structure of the multiplier will not be
influenced by the nature of the multiplicands'.

For more info on these packages, root around at vhdl.org. The link
below takes you to the user's guide for the fixed point package.

http://www.vhdl.org/vhdl-200x/vhdl-200x-ft/packages/Fixed_ug.pdf

Kevin Jennings

 

[Bar Nash]

Thanks
EC

"KJ" <kkjennings@sbcglobal.net> ???
??????:7cae9b36-74b0-4e05-a6ba-55aa34bfd598@f63g2000hsf.googlegroups.com...
On Oct 8, 8:34 am, "Bar Nash" <waves...@netvision.net.il> wrote:

Hi all

When a multipler is designed like those used in FIRS will the nature of
the
multiplicands influence the structure of it ?

I mean in case those are both fractional numbers <1 and in case both are
whole numbers .

It shouldn't. I'm assuming your question is targetted towards
multipliers meant to be synthesized, that being the case the first
step would be to define how you plan to represent numbers in general.
Depending on the particular needs of your FIR filters the numbers will
likely be either fixed point or floating point. Fixed point
representation will consume less logic than floating point but either
can be synthesized.

If you use the relatively newly standardized fixed or floating point
packages as the basis for representing 'numbers' then the answer to
your question is 'no, the structure of the multiplier will not be
influenced by the nature of the multiplicands'.

For more info on these packages, root around at vhdl.org. The link
below takes you to the user's guide for the fixed point package.

http://www.vhdl.org/vhdl-200x/vhdl-200x-ft/packages/Fixed_ug.pdf

Kevin Jennings