On 17 August 2014 10:01, Nikolay Sivov <nsivov at codeweavers.com> wrote: > So I guess it means test is incorrect, and implementation too. Yeah, most likely, although I haven't verified those results on real hardware.