Re: [oc] "1/x" function

["Tony Burch" <tony@burched.com.au>]

Hi Rudi,

> 
> Anybody have an efficient algorithm for a "1/x" function for
> hardware implementations ? Or any pointer to white papers, etc ?

I did a quick check of Reto's work.  Some division there,
but not a fast and small reciprocal.  His site is at
http://www.iis.ee.ethz.ch/~zimmi/arith_lib.html

Is "x" a fixed point number?  If so, how many bits?
If you can't use a lookup table approach (because
x is too many bits) with either on-chip or external
SRAM, then perhaps some iterative
technique can be pipelined to calculate reciprocal.
Maybe the Newton-Raphson method for computing
reciprocal.

> I need something that is fast and small, and can be pipelined.
> 
> Any pointers appreciated !
> 
> Thanks,
> -- 
> rudi
> ------------------------------------------------
> www.asics.ws   - Solutions for your ASIC needs -
> NEW ! 3 New Free IP Cores this months (so far :*)
> FREE IP Cores  -->   http://www.asics.ws/  <---
> -----  ALL SPAM forwarded to: UCE@FTC.GOV  -----
> 

HTH:)

Best regards
Tony Burch
http://www.BurchED.biz
FPGA boards for System-On-Chip prototyping and education


--
To unsubscribe from cores mailing list please visit http://www.opencores.org/mailinglists.shtml