conv -- IPP Convolution

   -1: Perform convolution on float vector.

  Default Parameters:
    -loop_start 5000 // Suggest number of loops for calibration.
    -start 4         // Starting problem size is 2^4 or 16.
    -param 16        // Size of coefficient kernel.

ipp/fft -- IPP Fft (fast fourier transform) benchmark
Single precision
   -1 -- inter-op: interleaved, out-of-place CC fwd fft
   -2 -- inter-ip: interleaved, in-place     CC fwd fft (vrsn >= 5)
  -11 -- split-op: interleaved, out-of-place CC fwd fft
  -12 -- split-ip: interleaved, out-of-place CC fwd fft (vrsn >= 5)

fft_ext -- FFT using Ext_data to call IPP

  -1: FFT on vector complex<float>

mcopy -- float matrix copy using IPP

   -1: rows -> rows   select algorithm
   -2: rows -> cols   select algorithm
   -3: cols -> rows   select algorithm
   -4: cols -> cols   select algorithm
 
  -11: rows -> rows   copy algorithm
  -12: rows -> cols   copy algorithm
  -13: cols -> rows   copy algorithm
  -14: cols -> cols   copy algorithm
 
  -22: rows -> cols   transpose algorithm
  -23: cols -> rows   transpose algorithm

  Default parameters:
    -stop 12        // Largest problem size is 2^12 or 4096

vmul -- IPP vector multiply

    -1: float           VSIPL++ allocation
    -2: complex<float>  VSIPL++ allocation

   -12: complex<float>  IPP allocation (ippsMul_32fc)   (in place)
   -22: complex<float>  IPP allocation (ippsMul_32fc_I) (in place)

  -101: float           IPP allocation (ippsMul_32f)
  -102: complex<float>  IPP allocation (ippsMul_32fc)