Re: Atlas proposal
> I agree on this. What we still don't agree on is whether you can build
> an optimized package at all, since Atlas will optimize it for the
> machine where it got built, and the optimizations it does will
> potentially make performance worse on another machine...
How does atlas cope with cgroup and cpuset ? In some hpc environnement
some user have a restricted number of node, wheras other have the full
machine. And this kind of stuff is getting more and more common. So
the hand compilation of atlas is not a solution. Atlas should use run
time switching optimisation.