home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!pacific.mps.ohio-state.edu!linac!att!ucbvax!virtualnews.nyu.edu!brnstnd
- From: brnstnd@nyu.edu (Dan Bernstein)
- Newsgroups: sci.crypt
- Subject: New record non-networked factorization of difficult number
- Message-ID: <26250.Jul2400.28.0492@virtualnews.nyu.edu>
- Date: 24 Jul 92 00:28:04 GMT
- Organization: IR
- Lines: 29
-
- Arjen Lenstra and I are pleased to announce the record non-networked
- factorization of a difficult number (no prime factors under 40 digits).
- The number is (2^488 + 1)/257. Its prime factors are p_49 and p_97,
- where
-
- p_49 = 1035817877926014488587133818491976759389034764353
-
- and
-
- p_97 = 3002073757428777382273857922385512797763792723266417656025\
- 021527116989779952950182556537541850817.
-
- We carried out the bulk of the computation in under ten days on a MasPar
- SIMD computer with 16384 processors at Bellcore. We also used under a
- day of CPU time on a DEC workstation. The entire factorization took
- approximately a month and a half real time.
-
- This should be compared with the overall record factorization of a
- difficult number, namely 2^512 + 1, which took four months real time on
- approximately 700 workstations around the world, together with a
- 65536-processor Connection Machine in one stage.
-
- This computation can be seen as a substantial step towards a practical
- implementation of the general number field sieve (GNFS): it depended on
- character columns and a square root inside the algebraic number field
- rather than explicit generators of the unit group and of prime ideals of
- the field. We are investigating the practical effectiveness of GNFS.
-
- ---Dan
-