Integer or STD_LOGIC_VECTOR

[jk]

Hi,

I am sort of new to VHDL and still trying to get around the optimisation
considerations associated with the language as opposed to using C.

A have written a function which uses a variable 'ppos' to hold the current
position in a vector. I have defined it as an INTEGER but I am beginning to
think that it might be more efficient implemented in binary. Is this a real
consideration in VHDL coding? Would using a binary vector in place of an
integer wherever possible resort in a more efficient implementation in the
long run?

cheers

- Kingsley

 

[Nicolas Matringe]

jk a écrit:

Hi,

I am sort of new to VHDL and still trying to get around the optimisation
considerations associated with the language as opposed to using C.

A have written a function which uses a variable 'ppos' to hold the current
position in a vector. I have defined it as an INTEGER but I am beginning to
think that it might be more efficient implemented in binary. Is this a real
consideration in VHDL coding? Would using a binary vector in place of an
integer wherever possible resort in a more efficient implementation in the
long run?

Hi
Integers always end up synthesized to binary numbers so you don't have
much to worry about that. The only way you can improve the synthesis
efficiency is to limit the possible values of your integer.
Since "ppos" represents a position in a vector, it is a positive
integer, a.k.a. natural. Your vector is not of infinite length, I don't
think it is longer than 128 bits (which is already quite long) so you
can define ppos like this:
signal ppos : natural range 0 to 127;
This way you tell the synthesizer that it won't need more than 7 bits to
represent ppos.


--
____ _ __ ___
| _ \_)/ _|/ _ \ Adresse de retour invalide: retirez le -
| | | | | (_| |_| | Invalid return address: remove the -
|_| |_|_|\__|\___/

 

[Kwaj]

Nicholas,

beautiful response - it was exactly what I was looking for given my
unfamiliarity with the language. thanks a lot

- Kwaj


"Nicolas Matringe" <matringe.nicolas@numeri-cable.fr> wrote in message
news:4003B18B.6090709@numeri-cable.fr...

jk a écrit:
Hi,

I am sort of new to VHDL and still trying to get around the optimisation
considerations associated with the language as opposed to using C.

A have written a function which uses a variable 'ppos' to hold the
current
position in a vector. I have defined it as an INTEGER but I am beginning
to
think that it might be more efficient implemented in binary. Is this a
real
consideration in VHDL coding? Would using a binary vector in place of an
integer wherever possible resort in a more efficient implementation in
the
long run?

Hi
Integers always end up synthesized to binary numbers so you don't have
much to worry about that. The only way you can improve the synthesis
efficiency is to limit the possible values of your integer.
Since "ppos" represents a position in a vector, it is a positive
integer, a.k.a. natural. Your vector is not of infinite length, I don't
think it is longer than 128 bits (which is already quite long) so you
can define ppos like this:
signal ppos : natural range 0 to 127;
This way you tell the synthesizer that it won't need more than 7 bits to
represent ppos.


--
____ _ __ ___
| _ \_)/ _|/ _ \ Adresse de retour invalide: retirez le -
| | | | | (_| |_| | Invalid return address: remove the -
|_| |_|_|\__|\___/

 

[Amontec Team, Laurent Gau]

Please DONT forget to use RANGE casting of your integer.

Please goto http://www.amontec.com/fix/vhdl_memo/index.html to know the
how to use RANGE VHDL keyword.

ex : If you do not use range for a 4 bits counter based on integer
signal, you will synthesize 'in reality' a 32 bits counter -> too many
flip-flops and a very bad speed of your 4 bits counter.

So, we ( Amontec Team ) never use integer for RTL description, but we
use unsigned type (look lib ieee.std_numeric.all).

Also, do not use std_logic_vector but std_ulogic_vector.

Laurent
www.amontec.com

If you are describing RTL synthesizable VHDL code

Nicolas Matringe wrote:

jk a écrit:

Hi,

I am sort of new to VHDL and still trying to get around the optimisation
considerations associated with the language as opposed to using C.

A have written a function which uses a variable 'ppos' to hold the
current
position in a vector. I have defined it as an INTEGER but I am
beginning to
think that it might be more efficient implemented in binary. Is this a
real
consideration in VHDL coding? Would using a binary vector in place of an
integer wherever possible resort in a more efficient implementation in
the
long run?


Hi
Integers always end up synthesized to binary numbers so you don't have
much to worry about that. The only way you can improve the synthesis
efficiency is to limit the possible values of your integer.
Since "ppos" represents a position in a vector, it is a positive
integer, a.k.a. natural. Your vector is not of infinite length, I don't
think it is longer than 128 bits (which is already quite long) so you
can define ppos like this:
signal ppos : natural range 0 to 127;
This way you tell the synthesizer that it won't need more than 7 bits to
represent ppos.