# .font: font.text_2
2021-09-28 21:56 dustin  
# .font: font.text_1
random seed: (170 102 87 198 86) 256  
yes, actually a number out of the 0-255 range! as there's no pattern in the original five seeds, i abused it for a little experiment: in gdash, it's possible to chose any random seed, also beyond 255. i thought that 256 would give the same result as 0 because everything is calculated modulo 256 anyway. indeed the seeds 0 and 256 are very similar but not identic! there are three boulders which would not be there with seed 0 (i marked them as heavy boulders; it's easy to compare with my m/6 version where seed is 0. although m/6 has a higher boulder density, those three boulders are missing!) also, there are five missing rocks which would be there with seed 0.interesting!  
  
\required: (85) 95  
  
cave time: (150 150 110 110 110) 86  
  
btw, this cave is close to insane with the boulders close to the butterfly-entrance and the low timer. i think it fits well with general bd3-higher level-madness! :d

# .font: font.text_2
2021-09-28 22:10 nesdori  
# .font: font.text_1
 dustin , it seems that if in gdash the seed is over 255, it sets seed_1 (which usually is initially 0) with the upper bits. so the effect you are getting is equivalent starting with seed_1 = 1 and seed_2 = 0. if the gdash seed is 512, then it would be equal to seed_1 = 2, seed_2 = 0 and so on.

# .font: font.text_2
2021-09-28 22:57 altermaven  
# .font: font.text_1
Interesting, considering that in typical c64 restraints, if you did somehow go above 255, you'd not only commit a wrap and conflict with the nearest byte, you would also wind up with an rng that is a modulus of what is listed (as the RNG seed is uint!). so if you had a seed of 384, you'd wind up with 01:80 which would just read 0x80 for the seed (128!)

# .font: font.text_2
2021-09-28 23:04 nesdori  
# .font: font.text_1
 altermaven , that's right. but c64 rng has two 8-bit seeds (seed_2 is the one which is chosen in cave data and seed_1 is just 0 initially) and gdash just combines those two into one 16-bit value. a little confusing way to enter seeds for other than coders :)

# .font: font.text_2
2021-09-28 23:41 dustin  
# .font: font.text_1
ah, interesting! but then i still wonder why the whole results are so similar with initial seed pairs (0,0) and (1,0). when we had an intense discussion of the rng in  Arno 's forum,  Arno  mentioned a case where seed pairs of the form (x, y) and (x+2, y) stayed close together for a long time, but i would not intuitively expect the same behaviour here because i remember there were quite some nuances in the formulas that made a huge difference between evenness and oddness...

# .font: font.text_2
2021-09-28 23:55 nesdori  
# .font: font.text_1
 dustin , surprising similar indeed. with seed_1 set to 1 i tried also other seed_2:s. with seed_2=1 there's also just 3-boulder difference. with seed_2=2 there's no difference at all! :O with seed2=3 there are several different boulders and in fact there are just few similar regions. interesting.. :)