Re: Please review my float package

Jerker Hammarberg (DST) wrote:

> I put together a generic floating point package that performs addition,
> subtraction, multiplication and comparison. Contrary fphdl
> (www.eda.org/fphdl), it is synthesizable with Xilinx XST - that's why
> I made my own. Anyway, I present it here in the hope that anyone may
> wants to look through it for
> * possible optimizations, particularly for implementation size,
> * language misuse or simplifications,
> * bugs,
> * or actually, any possible improvements.

Consider:

1. A simple testbench with an assertion for each function.
2. numeric_std:

-- use ieee.std_logic_arith.all;
use ieee.numeric_std.all;

m := "01" & shift_left((m(size-3 downto 0))
to_integer(to_unsigned(leftshift,EXPSIZE)));

-- m := "01" & shl(m(size-3 downto 0),
-- conv_unsigned(leftshift, EXPSIZE));

-- rsres := shr(rsarg1, rsarg2);
rsres := shift_right(rsarg1, to_integer(rsarg2));

-- return conv_std_logic_vector(result, TOTALSIZE);
return std_logic_vector(result);


-- return conv_std_logic_vector(result, TOTALSIZE);
return std_logic_vector(result);



-- Mike Treseler

Mike Treseler

missed a comma \

m := "01" & shift_left((m(size-3 downto 0)) ,
to_integer(to_unsigned(leftshift,EXPSIZE))) ;

-- Mike Treseler

Mike Treseler

"Jerker Hammarberg (DST)" wrote:
>
> > 2. numeric_std:
>
> This is interesting... what is the difference between numeric_std and
> std_logic_arith? I can see the former is slightly more readable. Is it also
> more efficient?

I don't know for sure about efficiency but I doubt there is any
appreciable difference since efficiency is determined primarily by the
tool, not the package, unless you have optimization turned off.

The most important difference is that numeric_std is an IEEE standard
while std_logic_arith actually has several different (and incompatible)
versions, depending on whose tool you're using. If you want your code to
be portable, use numeric_std.
--
Tim Hubberstey, P.Eng. . . . . . Hardware/Software Consulting Engineer
Marmot Engineering . . . . . . . VHDL, ASICs, FPGAs, embedded systems
Vancouver, BC, Canada . . . . . . . . . . . http://www.marmot-eng.com

Tim Hubberstey