On Tue, 7 Sep 1999, Alfred Weyers wrote:

> > takehiro's speedups (mentioned in a previous post) are at
> > >       http://www.isoternet.org/~tominaga/lame-0906.tar.bz2
> > 
> > 16% speedup.  May be some quality issues though - preliminary listen
> indicated
> 
> yes, confirmed,
> on my box a "lame -b 160" encoding is about 20% faster. The sound isn't
> right, though. The MDCT does not look quite correct. I poked a bit in the
> MDCT-code and now it sounds better. I cannot say whether I got it right.

hi,

ok, this is the first time i'm writing here, because i think that i won't
have anything intelligent to tell you. i'm amazed.

for the records: the takehiro-version does not run when compiled with pgcc
2.95.1. it gives me a "memory allocation error". processor is a K6, it
does not matter which compiler options i turn off. 

btw: it seems like in pgcc 2.95.1 -ffast-math is broken. using it when
compiling the standard lame 2.25 gives me a (nearly) constant result of 32
kbit/s with options -V0 -d -h. the result is of course awful.

should we write this into the makefile?

ciao,

olaf 

--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )

Reply via email to