Themaister bebe0d78a7 Optimize sinc for AVX as well.
Disable it for now however, as it's slightly slower
than SSE1 for the few taps we're using.

From testing, it's 10-20% faster when number of taps are increased.
The AVX path might need some more tuning, but it's fair to
assume the algorithm is memory bound.
2012-11-06 18:52:09 +01:00
..
2012-10-16 12:54:58 +02:00
2012-11-03 23:03:57 +01:00
2012-09-25 13:33:11 +02:00
2012-04-21 23:31:57 +02:00
2012-10-18 06:06:55 +02:00
2012-10-13 20:22:45 +02:00
2012-09-07 21:56:26 +02:00
2012-06-20 00:44:40 +02:00
2012-04-21 23:31:57 +02:00
2012-10-23 22:27:15 +02:00
2012-09-07 21:56:26 +02:00
2012-09-07 21:56:26 +02:00
2012-04-21 23:31:57 +02:00
2012-04-21 23:31:57 +02:00
2012-11-06 18:52:09 +01:00
2012-11-06 09:26:28 +01:00
2012-11-03 14:15:03 +01:00
2012-04-21 23:31:57 +02:00