FLOATING POINT DIVISION

[sunwij]

Following is a portion of the code which calculates a rate of rain
fall at a remote location and executes a specfic task if the rate
exceeds a given threshold. Synopsis throws following error: "
STANDARD.STD" package not found

Am I missing any specific pkg. Tried http://www.eda.org/fphdl/
read the whole page. None found. As a work around I am using a
function to do division which uses lot of hardware.

library ieee;
use ieee.std_logic_1164.all;
use ieee.numeric_std.all;
use ieee.math_real.all;
entity real_division is port ( rain_fall : in real;
time_elapsed: in real;
rain_rate: out real}
end real_division;
architecture myreal_division of real_division is

Signal a1,b1 :real;

begin

a1<= rain_fall;
b1<= time_elapsed;
process (a1,b1)
begin
rain_rate <=a1/b1;

end process;

end real_division;

 

[David Bishop]

sunwij wrote:

Following is a portion of the code which calculates a rate of rain
fall at a remote location and executes a specfic task if the rate
exceeds a given threshold. Synopsis throws following error: "
STANDARD.STD" package not found

Am I missing any specific pkg. Tried http://www.eda.org/fphdl/
read the whole page. None found. As a work around I am using a
function to do division which uses lot of hardware.

First off, "real" is not synthesizable, since it is implementation
dependent (I know that that is fixed in vhdl-2000). In the floating
point packages, I created a type called "fp32".
Download "fphdl_base_pkg.vhd", "fphdl_base_pkg_body.vhd", and
"fphdl32_pkg.vhd".

library ieee;
use ieee.std_logic_1164.all;
use ieee.numeric_std.all;
use ieee.math_real.all;

replace this line with:
use work.fphdl32_pkg.all;

entity real_division is port ( rain_fall : in real;
time_elapsed: in real;
rain_rate: out real}

Change "real" to "fp32" (32 bit floating point number).

end real_division;
architecture myreal_division of real_division is

Signal a1,b1 :real;

Once again, "fp32".

begin

a1<= rain_fall;
b1<= time_elapsed;
process (a1,b1)
begin
rain_rate <=a1/b1;

end process;

end real_division;

Now this still won't work in Synopsys. Why? Synopsys can't do a
divide that isn't a divide by 2.
Not to worry. In "fphdl_base_pkg_body.vhd" you will find a synthesizable
unsigned divide routine that will work fine with Synopsys.

The downside here is that the divider will not be very fast. I have one
that will work at 100 MHz in a "Xilinx" part if you want it.

--
NAME: David W. Bishop INTERNET: dbishop@vhdl.org

 

[sunwij]

David Bishop <dbishop@vhdl.org> wrote in message news:<DnEHb.84463$JW3.31474@twister.nyroc.rr.com>...

The downside here is that the divider will not be very fast. I have one
that will work at 100 MHz in a "Xilinx" part if you want it.

Thanks so much for your valuble advise.

Sunwij

 

[sunwij]

David Bishop <dbishop@vhdl.org> wrote in message news:<DnEHb.84463$JW3.31474@twister.nyroc.rr.com>...

The downside here is that the divider will not be very fast. I have one
that will work at 100 MHz in a "Xilinx" part if you want it.

Is this "Xlinix" part is hardware or a VHDL code ? . Please give
little more details. Thanks