Stockfish 11 is coming out very soon!

Today, Stockfish' boss Stephane Nicholet announced that they are going to release Stockfish 11.

Stockfish is best! The best things in life are free.

It is free cos a group of people have been donating for human resources, hardwares and electricity bills for 10 years.

Just for electricity bills alone people might have donated for US$ 200,000-300,000.

Meanwhile, Leela project might have spent US$ 300,000- 500,000. ( electricity and server rental alone)

Wow! be aware lc0; beast is coming

Ok. SF 11 , will come out imminently.  (just waiting the release note from the boss)

I hope the Stockfish people are happy with their progress, but it really doesn't matter to me if their machine improves another hundred rating points. It already outrates me by way over a thousand.

Out of topic. While we amateurs are studying GM database, are the Professionals specially super GMs relying on  the most updated Engine games database?

Mostly people use engines for analysis. Human competition against engines was already over since decades ago.

What is the difference between Stockfish 10 vs Stockfish 11?

Stockfish 11 is + 50 elo stronger than Stockfish 10. As every +50 elo save doubling in computing power, "1 min of analysis by SF 11 is equivalent to 2 min of analysis by SF 10". 

In comparision with a decade old engine SF 3,  " 1 min of SF 11 can perform same quality of play vs 1 to 2  hours of SF 3 analysis in 10 years ago"

This is tremendous time/ power saving in improvements.

Stockfish 11 is officially released today.

http://abrok.eu/stockfish/

Note.. It will take a few days to appear in official web page.

Droidfish already has Stockfish 11 now. Peter Osterlund probably compiled from the source code himself.

As it is already known +50 elo.( based on 40,000) sample size games, which is like 57% expected score from Stockfish 11.

But if you run only 10 games, it could be 60% or 50%  scores, or even Stockfish 10 can beat SF 11 due to very small sample size( high statistical odds).

100 games tournment is generally enough to  show Likelyhood Of Superiority by SF 11.

Oops, I thought 40,000 games as usual , but they tested for 60,000 games.

Considering +35 elo vs +53 elo in two tests, the two results doesnt contradict.

In your test, low performance of SF 11 may be

1. Longer time control

2. Statistical variation.

https://github.com/glinscott/fishtest/wiki/Regression-Tests