[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Bug-gnubg] Simpler PRNG
From: |
Michael Petch |
Subject: |
Re: [Bug-gnubg] Simpler PRNG |
Date: |
Thu, 15 Jan 2015 17:42:41 -0700 |
User-agent: |
Mozilla/5.0 (Windows NT 6.3; WOW64; rv:31.0) Gecko/20100101 Thunderbird/31.4.0 |
On 2015-01-14 7:52 AM, Louis Zulli wrote:
> Saw this recently in the chess program Stockfish. Worth a look? Better
> randomness than MT:
>
> "This patch replaces RKISS by a simpler and faster PRNG, xorshift64* proposed
> by S. Vigna (2014). It is extremely simple, has a large enough period for
> Stockfish's needs (2^64), requires no warming-up (allowing such code to be
> removed), and offers slightly better randomness than MT19937."
>
> Paper:
> http://xorshift.di.unimi.it/
>
Thanks for the information. I am pretty sure I saw Philippe Michel at
one point mention information regarding his findings with profiling. My
understanding that the random number generation (with MT) is much of a
bottleneck within the context of rollouts.
"slightly better" randomness, but from the end user perspective within
the context of backgammon it likely wouldn't be noticeable (if at all)
even over a lot of matches.
I'm in agreement with Ian Shaw that a decrease in the options of random
number generators is probably more ideal. If others wish to weigh in on
this I am more than happy to listen.
Thanks,
--
Michael Petch
GNU Backgammon Maintainer / Developer
OpenPGP FingerPrint=D81C 6A0D 987E 7DA5 3219 6715 466A 2ACE 5CAE 3304
Re: [Bug-gnubg] The future of Windows builds / Https etc., Michael Petch, 2015/01/15