Fwd: Fwd: Mlucas + GMP

[Win C]

Oh sorry to mention that this is the time of squaring a n-bit number by ( n - 2 ) times. Therefore it takes so long...


-------- Original message --------
From: Torbjörn Granlund <tg at gmplib.org>
Date: 07/02/2018 21:43 (GMT+08:00)
To: Win C <winsto003 at hotmail.com>
Cc: gmp-discuss at gmplib.org
Subject: Re: Fwd: Mlucas + GMP

Win C <winsto003 at hotmail.com> writes:

  For 2^700000:

  Mlucas: 10 min
  GMP-6.1.2: 26 min

  For 2^86000:
  GMP-Development: 8.25s
  GMP-6.1.2: 8.8s

  Now it is clearer, right? The difference is not that small between GMP
  and mlucas.

OK. so for at least one operation on one computer, mlucas is 62% faster
than GMP.  That's a very signiicant difference.

But it is not clear what you're measuring.  When I multiply numbers of
around 700000 bits using GMP on a 5 year old Intel Ivy Bridge, it takes
0.0028s, a far cry from your 26 minutes.

But perhaps you really intend to say that you measure 700000 *word*
operands?  Then GMP needs 0.36s.

--
Torbjörn
Please encrypt, key id 0xC8601622