DSPRelated.com
Forums

Integer FFT

Started by Sangamanath Kalashetty January 5, 2012
Hi all,

I am new to the DSP area. Currently I am trying to understand the
integer FFT algorithm. But after going through the document I am not
clear about the uplifting and quantization of the twiddle co-
efficient's by using which the multiplication operation is converted
into the shift and addition operations.


Eagerly waiting for reply.
Thank you in advance.
On 1/5/12 6:15 AM, Sangamanath Kalashetty wrote:
> > I am new to the DSP area. Currently I am trying to understand the > integer FFT algorithm. But after going through the document I am not > clear about the uplifting and quantization of the twiddle co- > efficient's by using which the multiplication operation is converted > into the shift and addition operations. >
in other words, you don't understand how a floating-point algorithm is converted to a fixed-point algorithm. right? -- r b-j rbj@audioimagination.com "Imagination is more important than knowledge."
On Thu, 5 Jan 2012 03:15:22 -0800 (PST), Sangamanath Kalashetty
<pintu2727@gmail.com> wrote:

>Hi all, > >I am new to the DSP area. Currently I am trying to understand the >integer FFT algorithm. But after going through the document I am not >clear about the uplifting and quantization of the twiddle co- >efficient's by using which the multiplication operation is converted >into the shift and addition operations. > > >Eagerly waiting for reply. >Thank you in advance.
It might help to identify or share a link to whatever document you're reading, as there are many that cover integer implementations of the FFT. Eric Jacobsen Anchor Hill Communications www.anchorhill.com
On 1/5/2012 6:15 AM, Sangamanath Kalashetty wrote:
> Hi all, > > I am new to the DSP area. Currently I am trying to understand the > integer FFT algorithm. But after going through the document I am not > clear about the uplifting and quantization of the twiddle co- > efficient's by using which the multiplication operation is converted > into the shift and addition operations. > > > Eagerly waiting for reply. > Thank you in advance.
I suggest that you get an understanding og integer operations on their own before applying that knowledge to any particular algorithm. For example, shift-and-add multiplication makes sense on its own. Binary numbers add powers of two. The one bits tell us which powers to include in the sum and the zeros, which powers to ignore. Shifting a binary number left multiplies it by two. Shifting N places multiplies it by 2^N. Multiply by hand (in binary) 13*3 and then 13*5. You will see that you learned the basic method (in decimal) in elementary school. Jerry -- Engineering is the art of making what you want from things you can get. &#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;
On Jan 5, 5:22&#4294967295;pm, robert bristow-johnson <r...@audioimagination.com>
wrote:
> On 1/5/12 6:15 AM, Sangamanath Kalashetty wrote: > > > > > I am new to the DSP area. Currently I am trying to understand the > > integer FFT algorithm. But after going through the document I am not > > clear about the uplifting and quantization of the twiddle co- > > efficient's by using which the multiplication operation is converted > > into the shift and addition operations. > > in other words, you don't understand how a floating-point algorithm is > converted to a fixed-point algorithm. &#4294967295;right? > > -- > > r b-j &#4294967295; &#4294967295; &#4294967295; &#4294967295; &#4294967295; &#4294967295; &#4294967295; &#4294967295; &#4294967295;r...@audioimagination.com > > "Imagination is more important than knowledge."
Buddy If you know about floating point let me know.