|
BGonline.org Forums
41$-44, data
Posted By: neilkaz In Response To: 41$-44, data (Nack Ballard)
Date: Friday, 25 December 2009, at 5:02 a.m.
I extended for another 5k and it will be interesting to see if anything changes. I did a few quick checks of 4 ply quick cube action and didn't come across anything shocking.
My opinion remains that XG 4 ply should be the choice to use when the weaker settings don't agree. When one looks here http://www.extremegammon.com/independentstudies.asp and sees that XG 4 ply has a total error of 10 compared to GNU and Snowie (2 and 3 ply) of about 17, one sees why I like XG 4 ply. These Depreli tests are from all positions where the bots don't agree from a long money session so they run the full gamut of what occurs when playing and obviously the more common errors are going to be included more often.
Of course when one see XGR+ with only a 6.5 error, one wants to use that for RO's but that isn't possible yet and it is 5 times slower than 4 ply. This means that if and when it becomes possible to use XGR+ for RO, I will want an even faster computer system. .. lets hope i9 is out by then or maybe I just will need dual 4 core HT chips and hopefully more motherboards are out to handle this.
|
BGonline.org Forums is maintained by Stick with WebBBS 5.12.