Re: Tuning Karatsuba in LibTomMath

From: Tom St Denis (tomstdenis_at_yahoo.com)
Date: 04/27/03


Date: 27 Apr 2003 04:29:21 -0700

Tim Josling <tej_at_melbpc.org.au_rubbish@nospam.com> wrote in message news:<3EAB6471.9050601@nospam.com>...
> Tom St Denis wrote:
> > Tim Josling <tej_at_melbpc.org.au_rubbish@nospam.com> wrote in message news:<3EA92F89.9050909@nospam.com>...
> >
> >>...
> > It's "Karatsuba" named after the guy who invented the technique and
> > co-authored a paper on it. According to
> >
> > http://www.swox.com/gmp/manual/Nomenclature-and-Types.html#Nomenclature%20and%20Types
> >
> > A "limb" is a single digit. So when they say "as little as 8 limbs"
> > they do mean 8 "digits" [by my fancy terminology].
> >
>
> In gmp a limb is one 'word' e.g. on x86 it is 32 bits.

What is your point? In LTM I call it a "digit" which is more natural
and in GMP they call it a limb. They're both the same thing.

Tom



Relevant Pages

  • Re: Long division algorithm
    ... >> Tom no in university. ... Recommending texts you have to purchase is just another ... I'd say either GMP or LTM would be an excellent place to start [I'd say LTM ...
    (sci.crypt)
  • Re: Feeler for GMP support
    ... > Tom St Denis wrote: ... > I would not use a cryptographic library that envisaged the use of GMP ... > since I would worry that its author had not understood security objectives. ... Though to be fair all around LTM and TFM are not dynamically secure ...
    (sci.crypt)
  • Re: Voters registration
    ... Thanks Tom - let me clarify - there is only one table, ... Within the table there could be multiple people living at the ... labels for the same address. ... then a 5 digit zip? ...
    (microsoft.public.access.queries)
  • Re: Poly1305-AES
    ... > Tom St Denis wrote: ... > People writing cryptographic software for an embedded CPU generally ... > networking tool, it's not safe to rely on the OS to have ... Including OpenSSL and GMP as part of the tool is quite ...
    (sci.crypt)
  • Re: Jim Steuerts Avalanche Cipher
    ... > Wow, Tom, isn't it funny that you haven't rebutted a single ... All 8 of my LTM users depend on me! ... approach lacks technical merit ... you have a Rivest seal of approval you cannot design an algorithm. ...
    (sci.crypt)