Example of random number generator

The JIFFY one only works if you're using the complete standard interrupt handler and have interrupts enabled all the time.

Always the best random generator is using the human interface with a LD A,R or any other (described above) method. Reading the joystick status directly from the PSG ports or some other user-influenceable input is very hard to plan and therefore one of the best methods to implement.

The JIFFY one only works if you're using the complete standard interrupt handler and have interrupts enabled all the time.

The JIFFY value could be read in the game start routine once, before the interrupt handler be installed. After all, you only need to set SEED once too.

You could create a time variable in your interrupt handle, and use that variable to initiate the seed value.

Always the best random generator is using the human interface with a LD A,R or any other (described above) method. Reading the joystick status directly from the PSG ports or some other user-influenceable input is very hard to plan and therefore one of the best methods to implement.

A good idea to think about. LD A,R will generate numbers almost in sequence if used almost sequentially. This behavior is not useful to use this value without any other treatment. Do you agree?

SEED = 1

20000 trials: here you have the histogram (in txt :-) )

• 0 85
  1 74
  2 75
  3 87
  4 62
  5 82
  6 73
  7 84
  8 74
  9 73
  10 83
  11 83
  12 71
  13 78
  14 79
  15 82
  16 81
  17 91
  18 81
  19 81
  20 74
  21 72
  22 77
  23 77
  24 76
  25 77
  26 92
  27 85
  28 80
  29 78
  30 80
  31 83
  32 77
  33 79
  34 91
  35 65
  36 79
  37 86
  38 78
  39 72
  40 76
  41 77
  42 72
  43 78
  44 85
  45 81
  46 83
  47 74
  48 80
  49 75
  50 76
  51 84
  52 85
  53 79
  54 76
  55 72
  56 78
  57 76
  58 80
  59 87
  60 78
  61 80
  62 84
  63 74
  64 83
  65 71
  66 85
  67 79
  68 89
  69 78
  70 71
  71 75
  72 76
  73 85
  74 80
  75 81
  76 86
  77 76
  78 79
  79 78
  80 73
  81 85
  82 73
  83 88
  84 73
  85 77
  86 82
  87 84
  88 82
  89 71
  90 73
  91 74
  92 81
  93 68
  94 75
  95 76
  96 78
  97 76
  98 77
  99 76
  100 71
  101 75
  102 75
  103 72
  104 85
  105 83
  106 77
  107 69
  108 78
  109 75
  110 80
  111 78
  112 75
  113 74
  114 75
  115 81
  116 72
  117 87
  118 78
  119 78
  120 78
  121 70
  122 78
  123 81
  124 83
  125 72
  126 77
  127 77
  128 75
  129 77
  130 71
  131 75
  132 80
  133 89
  134 74
  135 84
  136 83
  137 88
  138 70
  139 68
  140 69
  141 89
  142 83
  143 76
  144 67
  145 80
  146 79
  147 76
  148 82
  149 74
  150 85
  151 79
  152 84
  153 75
  154 73
  155 64
  156 74
  157 78
  158 74
  159 82
  160 83
  161 80
  162 84
  163 71
  164 74
  165 79
  166 82
  167 81
  168 77
  169 74
  170 76
  171 80
  172 77
  173 73
  174 78
  175 77
  176 80
  177 76
  178 77
  179 80
  180 77
  181 73
  182 75
  183 80
  184 77
  185 76
  186 68
  187 76
  188 84
  189 81
  190 76
  191 76
  192 79
  193 74
  194 79
  195 77
  196 77
  197 65
  198 80
  199 86
  200 72
  201 85
  202 79
  203 77
  204 83
  205 74
  206 78
  207 74
  208 75
  209 78
  210 73
  211 80
  212 82
  213 76
  214 69
  215 75
  216 82
  217 81
  218 76
  219 80
  220 83
  221 76
  222 83
  223 80
  224 80
  225 74
  226 75
  227 82
  228 82
  229 89
  230 86
  231 82
  232 78
  233 75
  234 89
  235 77
  236 78
  237 72
  238 79
  239 73
  240 81
  241 77
  242 79
  243 80
  244 75
  245 85
  246 78
  247 76
  248 84
  249 83
  250 75
  251 84
  252 82
  253 73
  254 82
  255 77

and the matlab code:

SEED = 1;

T=[];

for n=1:20000

if SEED >= 32768

SEED = bitand(SEED * 2, 2^16-1);

SEED = bitxor(SEED,131);

OUT = bitand(SEED , 255);

else

SEED = SEED * 2;

OUT = bitand(SEED , 255);

end

T = [T; OUT];

end

The data seems more o less uniform

SBC A,A will turn all A bits equal Carry Flag. This will be used to calculate SEED LSB.

;Carry Flag=0 => A=0, Carry Flag=1 => A=FFh
SBC A,A
;A and 83h
AND 83h
;A xor SEED LSB
XOR L

Carry Flag come from the overflow of ADD HL,HL.

Am I correct?

a very real ramdon number is too hard to obtain. Indeed there is books that only are dedicated to this matter.
You can make complex or less complex calculations but with the time and certain ammount of loops it will always back to a previously given pattern.

The secret is how to do a pattern that returns all the values of the range, each value in an equal proportion..

So, to test a ramdon generator you needs to do it

like in 10.000 executions.... how much times it returns 0, how much times it returns 1.... how much times it returns 2....... how much times it returns 255....

then make a graphic, with X axis the VALUE returned.... and Y axis the ammount of time that the value was returned....
if that graphic has a LINE, and not mountains or peaks-up or peaks-down.... you got it!. The perfect one!