x86_64 generic

[Jason Moxham]

Hi

in the mpn/x86_64/ directory we have the files
add_n.as
addmul_1.as
lshift.as
mul_1.as
rshift.as
sub_n.as
submul_1.as

these are only ever used by x86_64 cpus that are not in our cpuid , and we
have all them with the possible exception of Via Nano and Westmere . They are
only tested in these circumstances , and they are little better than the c
code , which does get tested .
I vote we get rid of them.

Jason

[Bill Hart]

Isn't the idea that if a new chip comes out that we don't recognise
then that would be the fallback? As they are not linked in anyway I'm
not sure I see the point of removing them. It makes our tarball
smaller, but not by much.

Bill.

2009/9/7 Jason Moxham <ja...@njkfrudils.plus.com>:

[Jason Moxham]

Yeah , thats the idea , and as they are little better than C code , I just
thought it would clean up the code a bit by getting rid of them.The C code
has the added advantage that it does get tested when we build with build=none
to do a full test of all the assertions(--enable-assert)