New thresholds in table
nisse at lysator.liu.se
Mon Nov 14 12:02:30 CET 2011
Torbjorn Granlund <tg at gmplib.org> writes:
> I suppose 10 is still too high then. Or doesn't the code support lower
I think it should work with a threshold as low as 4 or so.
> The spreading of measured values for k10/k8 makes me suspect that
> something might not be right.
> All k10 and k8 machines using 64-bit limbs get values less then 30,
> but one is far off that:
> gcc12.fsffrance.org 171
> These machines have two nehalem generation CPUs:
> biko-solaris64.gmplib.org 169
> gcc20.fsffrance.org 12
Hmm. I think the threshold should be in the same ballpark as
HGCD_THRESHOLD. Which makes small values for all x86_64, 64-bit ABI
builds suspicious. The smallest value for builds in that class is 78,
One would need to look at sime time/size graphs for those machines to
figure out what's right.
>From comments, it seems the reason min_size was set in tuneup for the
HGCD_THRESHOLD was some problem with bogus small results.
> Most of the GMP testing is nowadays done on emulators or under
> virtualisers; running tuneup there makes limited sense.
Then it would be good to somehow mark those which do make some sense, on
the threshold report pages...
Niels Möller. PGP-encrypted email is preferred. Keyid C0B98E26.
Internet email is subject to wholesale government surveillance.
More information about the gmp-devel