Discrepancy in measuring performance?

[Torbjorn Granlund]

[Resend, I accidentally replied to just the sender.]

Some guesses:

1. Your mp_bits_per_limb is 32 when it should be 64.
   This could be caused by picking up the wrong gmp.h.

2. You're using the kernel Linux and your system is loaded without you
   noticing.  This kernel doesn't virtualise TSC, making cycles taken by
   any other process spill into you process' TSC reads.  (To my
   knowledge, all BSD kernels and Solaris gets this right.)

3. You're linking to an outdated version of GMP for you own tests (as
   opposed to the tests with the 'speed' program.

-- 
Torbjörn