erikd@coltrane > tests/benchmark 
Benchmarking libsndfile-0.0.28
------------------------------
Each test takes a little over 5 seconds.

    Raw write PCM_16  :   30660117 samples per sec
    Raw read  PCM_16  :   62788982 samples per sec

Native endian I/O :
    Write short   to  PCM_16 :  83.37% of raw write
    Read  short  from PCM_16 :  83.17% of raw read
    Write int     to  PCM_24 :  30.78% of raw write
    Read  int    from PCM_24 :  32.96% of raw read
    Write int     to  PCM_32 :  42.05% of raw write
    Read  int    from PCM_32 :  41.11% of raw read
    Write float   to  PCM_16 :  17.75% of raw write
    Read  float  from PCM_16 :  43.27% of raw read
    Write float   to  PCM_24 :  15.30% of raw write
    Read  float  from PCM_24 :  28.09% of raw read
    Write float   to  PCM_32 :  14.55% of raw write
    Read  float  from PCM_32 :  34.65% of raw read
    Write float   to  FLOAT  :  28.98% of raw write
    Read  float  from FLOAT  :  56.71% of raw read

Endian swapped I/O :
    Write short   to  PCM_16 :  43.39% of raw write
    Read  short  from PCM_16 :  49.12% of raw read
    Write int     to  PCM_24 :  29.65% of raw write
    Read  int    from PCM_24 :  33.66% of raw read
    Write int     to  PCM_32 :  19.62% of raw write
    Read  int    from PCM_32 :  21.97% of raw read
    Write float   to  PCM_16 :  17.63% of raw write
    Read  float  from PCM_16 :  31.43% of raw read
    Write float   to  PCM_24 :  14.91% of raw write
    Read  float  from PCM_24 :  27.99% of raw read
    Write float   to  PCM_32 :  13.69% of raw write
    Read  float  from PCM_32 :  22.23% of raw read
    Write float   to  FLOAT  :  19.25% of raw write
    Read  float  from FLOAT  :  25.66% of raw read