(help)floating point to binary in verilog

[rajsrikanth]

how to convert floating point into binary verilog

can any one help me to develop code to convert a floating point number to binary and writing code in verilog.
can u sujjest some sites or material or can give any code for that
pls help. i have to develop a FIr filter in verilog and in that i have a need to convert floating point filter coeffecients to binary and hv to develop coding in verilog.
and i have to implement in FPGA.

thanx and regards
raj

 

[rsrinivas]

floating point in verilog

matlab is the best.
try FDATool in Matlab.
u can design a FIR filter quantize it to fixed point generate verilog/VHDL code and do a cosimulation in an FPGA.
Regards

 

[avimit]

verilog floating point

Hi,
If you understand C then look:
http://www.vlsiip.com/c/point.c
If you understand perl then look:
http://www.vlsiip.com/perl/point2bin.txt

Both these programs are doing exactly what you want to do. You will be able to get the logic from these. Any probs, let me know.
Kr,
Avi
http://www.vlsiip.com

 

[simon111]

verilog floating point to integer

i faced this problem before, finally, i wrote a matlab script to do it:

        function bin_str=decim2bin(decimal,decbits);

% check parameter
if nargin==0, error('usage: decim2bin(decimal, decimalbits)'); end
if nargin==1, decbits=32; end

% check wether is a neg number
is_neg=0;

int=fix(decimal);

if decimal <0
	is_neg=1;
	dec=int-decimal;
	int=-int;
	bin_str='1';
else
	dec=decimal-int;	
	bin_str='0';
end


int_str=dec2bin(norm(int));


if is_neg
	int_str=char('0'+'1'-int_str);
	if dec==0
		int_bits=(length(int_str));
		int=power(2,int_bits)-int;
		int_str=dec2bin(int);
        int_len=length(int_str);
        if int_bits>int_len
            int_fill_str=blanks(int_bits-int_len);
            int_fill_str(:)='0';
            int_str=[int_fill_str int_str];
        end
	end
end

bin_str=[bin_str int_str ];

if dec~= 0
    bin_str=[bin_str '.'];
    int_bits=length(int_str);

    if is_neg
        dec=1-dec;
    end

    dec_str=blanks(decbits);
    dec_str(:)='0';
    bin_str=[bin_str  dec_str];

    for n=1:decbits
        dec=dec*2;
        if dec >=1
            bin_str(n+int_bits+2) = '1';
            dec=dec-1;
        end
    end


    
end

 

[echo47]

decimals loating point verilog

I usually multiply all the floating point values by a constant, round them off to integers, and enter those integers into my HDL.