Author Topic: Floating point PRNG  (Read 3300 times)

johnsa

  • Member
  • ****
  • Posts: 807
    • Uasm
Re: Floating point PRNG
« Reply #15 on: September 29, 2018, 01:51:09 AM »
I can't find any claim that it's better than xoroshiro128+, only xorshift or xoroshiro64.. from some other links:

https://nullprogram.com/blog/2017/09/21/
http://xoshiro.di.unimi.it/#shootout
http://www.pcg-random.org/posts/birthday-test.html

I'll stick to it for now anyway either way.

AW

  • Member
  • *****
  • Posts: 2583
  • Let's Make ASM Great Again!
Re: Floating point PRNG
« Reply #16 on: September 29, 2018, 02:03:25 AM »
Quote
For Crypto RNG, rdrand.. yep.. but it's slow as heck
Slow but unpredictable.
Not everybody wants fast food, sometimes it is better to wait a couple of nanoseconds more to eat something based on entropy.

johnsa

  • Member
  • ****
  • Posts: 807
    • Uasm
Re: Floating point PRNG
« Reply #17 on: September 29, 2018, 04:26:43 AM »
Indeed, for the right application it's ideal.. I was planning on using it as a source of entropy/randomness to work on my telekinesis skills.. :) with a nice gui to boot.. haha

jj2007

  • Member
  • *****
  • Posts: 10536
  • Assembler is fun ;-)
    • MasmBasic
Re: Floating point PRNG
« Reply #18 on: September 29, 2018, 11:11:45 AM »
I can't find any claim that it's better than xoroshiro128+, only xorshift or xoroshiro64

Google for "O'Neill" "Vigna" to see an interesting controversy.

P.S., your link to null program:
Quote
August 2018 Update: xoroshiro128+ fails PractRand very badly.
http://www.pcg-random.org/posts/a-quick-look-at-xoshiro256.html is also a good read 8)

johnsa

  • Member
  • ****
  • Posts: 807
    • Uasm
Re: Floating point PRNG
« Reply #19 on: September 29, 2018, 08:24:50 PM »
Ahh good spot and some interesting reading.
I'll give the 256** version a go and see how it fares.