From the technical point of view I wonder whether their AES-CTR-MT

https://bugzilla.novell.com/show_bug.cgi?id=412441 https://bugzilla.novell.com/show_bug.cgi?id=412441#c10 --- Comment #10 from Sebastian Krahmer <krahmer@suse.com> 2014-08-12 12:23:03 UTC --- My experiance is that HPC admins used to apply performance patches in a lot of places in order to tune their cluster. Ranging from openssh to kernel-patches or infiniband. So it probably gives little benefit to maintain this intrusive patch from our side when they are going to mod their stuff anyways. patch is correct anyways, since they do create N threads, but they do not bind it to any core. So they rely on the kernel/luck to schedule it in a way that the key bytes are generated truely in parallel. Depending on the cores load all threads may end up on the same core so you have actually worse performance. Would probably be better to use aes-ni? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.