The following example data files are provided as part of the
software constructed to perform benchmarks on the new Hybrid
and other implementations of the nrm2 routines.

AllJustAcuracy.dat
  Runs both single and double precision versions on random
  (0,1) data for accuracy only.

AllTestsAsc.dat
  Runs both single and double precision versions on data sorted
  into ascending order for both timing and accuracy.

AllTestsDesc.dat
  Runs both single and double precision versions on data sorted
  into descending order for both timing and accuracy.

AllTestsRand.dat
  The default settings.

dpKahanComp.dat, spKahanComp.dat
  Data for running a comparison between the original
  and one-pass versions of Kahan's method for single
  (spKahanComp.dat) and double (dpKahanComp.dat) precision and
  both real and complex (to test out both include files). Both
  accuracy and timing are reported.

dpHighScaling.dat
  Runs the double precision, real routine (dnrm2) on data vectors
  which will cause scaling to occur for all elements with Blue,
  Kahan and the original Algorithm 539 routines but not with
  the hybrid version. All elements are selected in the low
  range only.

spLowScaling.dat
  Runs the single precision, real routine (snrm2) on data vectors
  which will cause scaling to occur for all elements with Blue,
  Kahan and the original Algorithm 539 routines but not with
  the hybrid version. All elements are selected in the high
  range only.

dpBothScaling.dat
  Runs the double precision, real routine (dnrm2) on data vectors
  which will cause scaling to occur for all elements with Blue,
  Kahan and the original Algorithm 539 routines but not with
  the hybrid version. All elements are selected in either the low
  or high ranges only.

spKahanComp.dat
  Runs the single precision real and complex routines on data
  vectors that are worse case as far as the original version
  of the Kahan method is concerned.

dpKahanComp.dat
  Runs the double precision real and complex routines on data
  vectors that are worse case as far as the original version
  of the Kahan method is concerned.
