jens, i assume you meant this to go to the list?
Jens M Andreasen wrote:
On Sun, 2009-09-27 at 14:53 +0200, Jörn Nettingsmeier
wrote:
those who are following the development of the
xiph CELT low latency
codec might find this presentation interesting:
From the overheads:
- expf() is even slower than exp()
Changes exception handling mode of FPU, even if it's
already set correctly, then changes it "back"
This is a showstopper for DAZ and friends. Can somebody explain to those
in power, that this is a very unkind thing to do?
you could hop on
irc.freenode.net, /join #celt, #theora, #speex and/or
#xiph and see if derf (the presenter in the link i posted) or jmspeex
(jean-marc valin, the remote end in the celt presentation) are around,
and discuss it with them directly. alternatively, maybe torben reads
this and picks up the thread?