FPGA Central

what's the best way to transform decimal numbers to 32bits wide ieee
754 single precision floating point numbers? possibly using perl or
verilog. i did some searching, there are converters like this
http://babbage.cs.qc.edu/IEEE-754/Decimal.html, but not for large
amount of data. thanks, -j

[email protected] wrote:
> what's the best way to transform decimal numbers to 32bits wide ieee
> 754 single precision floating point numbers? possibly using perl or
> verilog. i did some searching, there are converters like this
> http://babbage.cs.qc.edu/IEEE-754/Decimal.html, but not for large
> amount of data. thanks, -j

I've never had cause to use it, but there are Verilog functions
$bitstoreal and $realtobits which convert reals to the IEEE-754 format
and vice versa. (Note: these use the 64-bit double format, not
single-precision.) One main use of these is to allow a user to pass
reals as vectors through 64-bit ports (since a port can't have a real
format). If you have Matlab, you can also use "format hex".
-Kevin

On Sep 25, 5:32*pm, Kevin Neilson
<[email protected]> wrote:
> [email protected] wrote:
> > what's the best way to transform decimal numbers to 32bits wide ieee
> > 754 single precision floating point numbers? possibly using perl or
> > verilog. i did some searching, there are converters like this
> >http://babbage.cs.qc.edu/IEEE-754/Decimal.html, but not for large
> > amount of data. thanks, -j
>
> I've never had cause to use it, but there are Verilog functions
> $bitstoreal and $realtobits which convert reals to the IEEE-754 format
> and vice versa. *(Note: these use the 64-bit double format, not
> single-precision.) *One main use of these is to allow a user to pass
> reals as vectors through 64-bit ports (since a port can't have a real
> format). *If you have Matlab, you can also use "format hex".
> -Kevin

If you have a C compiler you can write a simple program that
includes your data file as a comma-delimited list like:

float foo[] = {
#include numberlist.csv
};

and then write some simple code to coerce the values to unsigned int
and print them out.

Just a thought....