Fix for ticket:214, gr_rotate speedup. We now normalize once every

512 cycles.  We also now compute the return value based on the
preincremented phase value, thus shortening the dependency chain.
This does put the computed result 1 tick ahead of the previous version
of the code, but none of the code in the tree depends on the absolute
phase.  If yours does, sorry.  You can work around it by initializing
the phase to conj(gr_expj(phase_incr)) instead of gr_complex(1,0).
Added QA code to ensure I didn't break anything.



git-svn-id: http://gnuradio.org/svn/gnuradio/trunk@9384 221aa14e-8319-0410-a670-987f0aec2ac5

0 files changed, 0 insertions, 0 deletions