OpenEuphoria

2. Re: RNG Diehard results for Eu 'rand'

• Posted by Joel Crook <joel at MAIL.K-A.COM> Feb 14, 2000
• 612 views

Your code is MUCH better than mine --- then again I just started learning
the language... seems my data file fails after the third test.

Your results indicate the rand() function does indeed fail a number of the
tests



At 06:27 PM 02/14/2000 -0500, you wrote:
>I generated 3 sets of results from the numbers generated by the following
>code:
>
>----------------------------------------
>integer noshift, shifted, concat
>atom rn, rn2
>
>
>-- random #'s not shifted --
>noshift = open( "noshift.txt", "w" )
>
>for i = 1 to 400000 do
>  for j = 1 to 10 do
>    -- random # from 1 to #3FFFFFFF
>    rn = rand( #3FFFFFFF )
>    printf( noshift, "%08x", { rn } )
>  end for
>  puts( noshift, "\n" )
>end for
>
>close( noshift )
>
>----------------------------------------
>
>-- random #'s shifted --
>shifted = open( "shifted.txt", "w" )
>
>for i = 1 to 400000 do
>  for j = 1 to 10 do
>    -- random # from 4 to #FFFFFFFC
>    rn = rand( #3FFFFFFF ) * 4
>    printf( shifted, "%08x", { rn } )
>  end for
>  puts( shifted, "\n" )
>end for
>
>close( shifted )
>
>----------------------------------------
>
>-- 2 16-bit random #'s concatenated --
>concat = open( "concat.txt", "w" )
>
>for i = 1 to 400000 do
>  for j = 1 to 10 do
>    -- 2 16-bit #s from #0000 to #FFFF
>    rn =  rand( #10000 ) - 1
>    rn2 = rand( #10000 ) - 1
>    printf( concat, "%04x%04x", { rn, rn2 } )
>  end for
>  puts( concat, "\n" )
>end for
>
>close( concat )
>----------------------------------------
>
>I do not have time to analyze the results at the moment (I didn't even read
>all of the documentation...) but you can download them from
>http://cnw.com/~bkb/results.zip
>
>The testing on the unshifted results ended prematurely for whatever reason
>and the overall results look the ugliest for these numbers...
>
>-- Brian