Tuneup using FFT thresholds

Torbjorn Granlund tg at gmplib.org
Sun Nov 29 19:03:28 CET 2009


I am suspicious about the FFT tuneup.  Does post-FFT tunup parameters
really get a properly tuned mul_fft.c?  What happens with
e.g. MUL_FFT_TABLE?

We need to look into this.

For redc_n we will need a MULMOD_BNM1_TO_MUL_FFT_MODF_THRESHOLD, if we
implement FFT-based wrapping multiplication.  We'll use that in lots of
the division code too.

-- 
Torbjörn


More information about the gmp-devel mailing list