Closed tlepoint closed 8 years ago
Good work @tlepoint !! @serge-sans-paille @aguinet is the PR ok with you too ?
(quick answer) Ireally like the approach / concept. It's definitievly the way to go.
Concerning the implementation, what about including
http://www.boost.org/doc/libs/1_58_0/doc/html/align/tutorial.html
instead, it's header only, we could ship a copy and that's probably safer than the version you used (boost inside). I did not checked the dependencies though.
@serge-sans-paille indeed, but there is quite a lot of work to do to integrate all the gmp stuff too.
Can you propose a clean alternative soon ?
If not we can integrate @tlepoint 's proposal, which has the advantage of being ready. And you can open a second PR to replace it when you have the time to do it.
On Tue, Apr 12, 2016 at 01:04:17PM -0700, Carlos Aguilar Melchor wrote:
@serge-sans-paille indeed, but there is quite a lot of work to do to integrate all the gmp stuff too.
Can you propose a clean alternative soon ?
If not we can integrate @tlepoint 's proposal, which has the advantage of being ready. And you can open a second PR to replace it when you have the time to do it.
I agree, let's move on to keep the motivation high! Let me just valide it on my laptop and I'll merge it.
poly_p had no alignment (contrary to our tests) so I adapted a aligned memory allocator from http://jmabille.github.io/blog/2014/12/06/aligned-memory-allocator/ (just changed the construct and fixed the return type of deallocate) and now it works in Travis.