Using multiple LPMs of same type but different characteristics

Altera_Forum · ‎11-16-2010

Hi all,

I was trying to find a way to use the sam LPM Megafunction but with different settings. For example, the lpm_add_sub function, I would like to have both an adder and subtractor but can't figure it out.

- Here's an example

COMPONENT lpm_sub

GENERIC ( lpm_direction : STRING := "ADD";

lpm_hint : STRING := "ONE_INPUT_IS_CONSTANT=NO,CIN_USED=NO";

lpm_representation : STRING := "SIGNED";

lpm_type : STRING := "LPM_ADD_SUB";

lpm_width : NATURAL := 32

);

PORT ( dataa : IN STD_LOGIC_VECTOR(31 DOWNTO 0);

datab : IN STD_LOGIC_VECTOR(31 DOWNTO 0);

result : OUT STD_LOGIC_VECTOR(31 DOWNTO 0)

);

END COMPONENT;

- and then instantiate it with:

lpm_sub_component : lpm_sub

PORT MAP ( a, c, result_sub );

- I get

Error: Node instance "lpm_sub_component" instantiates undefined entity "lpm_sub"

So it appears the component name has to be that of the lpm function. What is the use of the LPM_TYPE characteristic then?

Thanks

Altera_Forum · ‎11-16-2010

At some point the compiler will search for the entity "lpm_sub", which you probably haven't created as you want to invoke an existing LPM function. Hence the error.

LPM_TYPE is to be seen as a constant rather than a parameter. LPM functions go back a long time, at some point it was an industry standard.

You could do it like this


library lpm ;
  use lpm.lpm_components.all ;
- - -
lpm_sub_component1 : lpm_add_sub
  generic map (
    lpm_direction : STRING := "ADD";
    . . . 
    )
  port map ( 
     . . .
     ) ;
lpm_sub_component2 : lpm_add_sub
  generic map (
    lpm_direction : STRING := "SUB";
    . . . 
    )
  port map ( 
     . . .
     ) ;

Altera_Forum · ‎11-16-2010

That works, thanks!

Altera_Forum · ‎11-16-2010

even simpler, use behavioural VHDL:

c <= a + b;

f <= d - e;

etc.

Altera_Forum · ‎11-16-2010

--- Quote Start ---

even simpler, use behavioural VHDL:

c <= a + b;

f <= d - e;

etc.

--- Quote End ---

yes it is a lot less typing and I do it that way almost always, but that wasn't the question ... necsup was contemplating LPM-functionality in general