Niels Möller nisse at lysator.liu.se
Thu Sep 5 04:55:18 UTC 2019

nisse at lysator.liu.se (Niels Möller) writes:

> Excellent. I just pushed tuning support, with HGCD2_METHOD == 1 using
> plain division for div1, and HGCD2_METHOD == 2 (default) using the old
> method with the div1 function.

I had a quick look at the machines that completed tuning this night.
These three seem to prefer the old code (HGCD2_METHOD == 2):

armcortexa7neon-unknown-linux-gnueabihf  pi2.gmplib.org-stat
armcortexa8neon-unknown-linux-gnueabihf  beagle.gmplib.org-stat
armcortexa12neon-unknown-linux-gnueabihf  tinker.gmplib.org-stat

The others want HGCD_METHOD == 1, i.e., replacing div1 with plain


Niels Möller. PGP-encrypted email is preferred. Keyid 368C6677.
Internet email is subject to wholesale government surveillance.

More information about the gmp-devel mailing list