Performance analysis of Mul algorithms in the same range

lcmeng lingchuanmeng at
Sun Jan 4 06:29:47 CET 2009

Searched first and didn't see any previous discussion. 

Hi all,

I'm new to GMP and trying to measure the performance of the four integer
multiplication algorithms. I tried manually modifying the
MUL_FFT_THRESHOLD in the 'gmp-mparam.h' as follows:

#define MUL_KARATSUBA_THRESHOLD          1
#define MUL_TOOM3_THRESHOLD              5100
#define MUL_FFT_THRESHOLD              5100

Then, I re-ran the configure and make. I expect it to force using Karatsuba
when the sizes of operands are in [1, 5100) in terms of limb. However, the
outcome doesn't change accordingly. It's more like that the Karatsuba hasn't
been forced to be used throughout the testing range.

Any suggestion is appreciated. Happy belated New Year!
Best regards,

Lingchuan Meng
View this message in context:
Sent from the Gnu - GMP - Discuss mailing list archive at

More information about the gmp-discuss mailing list