Takehiro Tominaga wrote:
>
> >>>>> "E" == Erik de Castro Lopo <[EMAIL PROTECTED]> writes:
>
> E> How much double precision floating point code does LAME use?
>
> IMHO, double precision is over spec for MP3 encoder/decoder. For
> example, mpg123 uses only float(single precision).
I agree completely that double precision is overkill for MP3 encoding.
The only reason to choose between double and float is for reasons of
speed,
> An input file (like .wav file) has only 16 bit precision and dynamic
> range(this value is selected from human hearing ability). And the
> IEEE768 says single precision has 24 bit precision.
>
> Current release of lame is mixed use of FLOAT(=float) and double.
> I think this is not good because it takes type conversion.
>
> See my version of lame. I changed all "double" into "FLOAT".
> And we can select FLOAT from double and float depending on your
> architechture. I think some architechtuer, like ALPHA chip, we will
> get faster conversion defining FLOAT as double. On my linuxbox with
> celeron 464MHz, I got about 10% speedup from defining FLOAT as float.
Thats interesting! I was under the impression that on Pentium processors
that double was usually faster than float. When you did your comparison,
did you use -malign-double to force the double to be aligned to eight
bytes? This is very important for achieving speed with doubles.
Secondly whats the difference with regards to floating point performance
between Celeron and pentium. It would be interesting if we got a
small double/float benchmark organised on this list and got a comparison
of performance across the full spectrum of processors that LAME runs on.
Erik
--
+-------------------------------------------------+
Erik de Castro Lopo [EMAIL PROTECTED]
+-------------------------------------------------+
Will the last person leaving Redmond please turn off all the lights.
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
