Hi Joshua!
> I see in the Makefile that __NO_MATH_INLINES was added to fix a bug in gcc
> 2.8+, what bug is this, what does this do and why do you have it defined?
> And what is RH_AMP?
The math inlines in GLIBC libraries before 2.1.3 seem to be broken. So if
you have some older GLIBC libraries installed you will have to define
__NO_MATH_INLINES at compile time. I have 2.1.3 installed and math inlines
seem to work, so I removed this definition (which enables math inlines).
RH_AMP enables some alternate code. I mentioned it, because if someone will
do the same test on those files, he can compare his results with mine.
By the way, many of the test files are available at sulaco.org.
> Another question would be, what version of gcc and binutils is being used
> for DJGPP. I personally am using pgcc 2.95.3 and binutils 2.10, both of
> which I compiled. That could make a difference. I'm still trying to figure
> out why MMX_choose_table does not work (at least for me) with DJGPP. I found
Hm, I had never good results with pentium optimized gcc. Every new version
of it gave me different results when doing floating point calcs.
So I decided for me not to use it.
Ciao Robert
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
