GMP+MPFR to GPU?

[Torbjorn Granlund]

Torbjorn Granlund <tg at gmplib.org> writes:

  Alas, the comparison presents strangely low GMP scores.  When I time it
  now, I get 4900 ops/s while the slides say the very exact number 2738.9.

No, I don't get that.  I tried to make sense of the numbers in the
slides, and in the end did not measure the right thing.

I don't understand the number 2738.9.  It does not make sense for any
common RSA length for the CPU in question.  Did the guy multiply a
one-core score by 4?  Then we're getting close, I'd get about 3430 for
that CPU (or about 6400 on a current high-end AMD CPU).

-- 
Torbjörn