GMP results for Woodcrest, Conroe, anyone?

Torbjorn Granlund tege at
Fri Jul 28 20:27:05 CEST 2006

Karl Hasselström <kha at> writes:

  On 2006-07-28 13:21:29 +0200, Torbjorn Granlund wrote:
  > It would be interesting to know how well it performs with GMP. While
  > the new P6 incarnation clearly has great 32-bit performance, for GMP
  > we almost only care about its 64-bit multiply throughput.
  Unless the 64-bit multiply throughput is more than 4 times as low as
  the 32-bit multiply throughput. But they wouldn't do that, would they?
Well, they've already done that...

The Prescott P4 (and later) can sustain one 32x32->64 multiply
every 2nd cycle, and one 64x64->128 multiply every 10th cycle.

If Net rumours are to be believed, there are issues with 64-bit
arithmetic performance also with the new P6.


More information about the gmp-discuss mailing list