Prev: ps2 mouse protocol
Next: CPLD beginner questions
From: faza on 16 Jun 2008 09:04
Hai,

I am facing problem while synthesis of Fixed point data type.I cannot
change the synthesis tool.Is there any method which can convert fixed
point to integer before hand and perform computation and convert back
to fixed point without affecting the precision.?

In my FIR filter design i will sample the input and perform
computation,produce result with control signal and then i will sample
the next data..ultimately i will waiting for the o/p control signal
before sampling the data..now my question is how this logic will be
implemented as hardware..do i need to store my input samples in ROM or
i should depend on the software to do this?

pls clarify.

regards,
faza



On Jun 14, 7:46 am, glen herrmannsfeldt <g...(a)ugcs.caltech.edu> wrote:
> faza wrote:
> > Can anyone explain how FIR filter implemented in real time
>
> I would vote for systolic array.
>
> > application....Wat is the role of software supporting the hardware???
>
> Initialize the coefficients, and start everything going.
>
> -- glen

From: Mike Lewis on 16 Jun 2008 11:24

"faza" <fazulu.vlsi(a)gmail.com> wrote in message
news:53fad5d8-9989-4089-82be-cf57588c9b34(a)u36g2000prf.googlegroups.com...
Hai,

I am facing problem while synthesis of Fixed point data type.I cannot
change the synthesis tool.Is there any method which can convert fixed
point to integer before hand and perform computation and convert back
to fixed point without affecting the precision.?

In my FIR filter design i will sample the input and perform
computation,produce result with control signal and then i will sample
the next data..ultimately i will waiting for the o/p control signal
before sampling the data..now my question is how this logic will be
implemented as hardware..do i need to store my input samples in ROM or
i should depend on the software to do this?

pls clarify.

regards,
faz



Dude .. give us some background ..

Are you a student? What year? Is this a project or an assignment?

Mike


From: faza on 17 Jun 2008 01:36
Hai,

I am a final year M.tech student.Its my final sem project..

regards,
faza



On Jun 16, 8:24 pm, "Mike Lewis" <some...(a)micrsoft.com> wrote:
> "faza" <fazulu.v...(a)gmail.com> wrote in message
>
> news:53fad5d8-9989-4089-82be-cf57588c9b34(a)u36g2000prf.googlegroups.com...
> Hai,
>
> I am facing problem while synthesis of Fixed point data type.I cannot
> change the synthesis tool.Is there any method which can convert fixed
> point to integer  before hand and perform computation and convert back
> to fixed point without affecting the precision.?
>
> In my FIR filter design i will sample the input and perform
> computation,produce result with control signal and then i will sample
> the next data..ultimately i will waiting for the o/p control signal
> before sampling the data..now my question is how this logic will be
> implemented as hardware..do i need to store my input samples in ROM or
> i should depend on the software to do this?
>
> pls clarify.
>
> regards,
> faz
>
> Dude .. give us some background ..
>
> Are you a student? What year? Is this a project or an assignment?
>
> Mike

From: glen herrmannsfeldt on 17 Jun 2008 03:18
faza wrote:

> I am facing problem while synthesis of Fixed point data type.I cannot
> change the synthesis tool.Is there any method which can convert fixed
> point to integer before hand and perform computation and convert back
> to fixed point without affecting the precision.?

I presume you mean fixed point with the binary point not
immediately to the right of the least significant bit,
sometimes called scaled fixed point.

Adding such numbers with the same scale factor is the same
as adding integers. Multiplying follows the rules you learned
in 3rd grade, the digits (bits) after the binary point of
the product is the sum of the bits after the binary points
of the to operands. Multiply should give you a double length
product, select the appropriate bits.

(snip)

-- glen

From: faza on 17 Jun 2008 10:49
Hai,

I want to know which is the right way of implementing the
hardware(industry standard)..I have referred various FIR
implementations where they are mostly handling filter coefficients as
integer(truncating from fixed or floating point using MATLAB) or
binary.Is it difficult to handle and implement real(fraction) filter
coefficients value directly in the hardware?

for example:

Filter coefficients:

fixed point=0.081207275390625 or
signed integer= 6945 or
fixed point binary =0011011001000010

all the above are equivalent but belongs to different data type..Now i
am confused which to select for implementation in my code..

Note:

Fixed point representation is looking challenging for some synthesis
tool as it not supported.
Signed integer looks simple but less accurate
Fixed point binary looks tedious..

Pls suggest..

regards,
faza




On Jun 17, 12:18 pm, glen herrmannsfeldt <g...(a)ugcs.caltech.edu>
wrote:
> faza wrote:
> > I am facing problem while synthesis of Fixed point data type.I cannot
> > change the synthesis tool.Is there any method which can convert fixed
> > point to integer  before hand and perform computation and convert back
> > to fixed point without affecting the precision.?
>
> I presume you mean fixed point with the binary point not
> immediately to the right of the least significant bit,
> sometimes called scaled fixed point.
>
> Adding such numbers with the same scale factor is the same
> as adding integers.  Multiplying follows the rules you learned
> in 3rd grade, the digits (bits) after the binary point of
> the product is the sum of the bits after the binary points
> of the to operands.  Multiply should give you a double length
> product, select the appropriate bits.
>
> (snip)
>
> -- glen

 | 
Pages: 1
Prev: ps2 mouse protocol
Next: CPLD beginner questions