Kingsrow download

[Ed Gilbert]

Kingsrow for 10x10 draughts is now a free download. You can find the link here: http://edgilbert.org/Checkers/KingsRow.htm

-- Ed

[Krzysztof Grzelak]

Thank you Ed.

[ildjarn]

Ed,

Thank you very much for all your work on Kingsrow, it's helped me a lot while composing (I know, not the primary use for a draughts program usually, but it was for me).

[hendrikv]

Thank you Ed,

Really happy with my free copy of one of the best programs in the world !

[Birdy]

Fantastic!
Thank you very much Ed!!

[Alvaro]

Many thanks Ed.
I'm very curious about the single core speed of Kingsrow 10x10 which is very high.
Did you do a simple search? Maybe the game is more tactical and needs a more tactical search, is that it?
My spanish checkers engine only does 2Mns in my old pc, while yours does 30Mns.

best regards,
Alvaro

[Ed Gilbert]

I'm very curious about the single core speed of Kingsrow 10x10 which is very high.
Did you do a simple search? Maybe the game is more tactical and needs a more tactical search, is that it?
My spanish checkers engine only does 2Mns in my old pc, while yours does 30Mns.

I think kingsrow is in the same speed ballpark as other 10x10 engines.
On my Xeon E5-2630 v3 @ 2.4GHz I get only 1/4 the speed you're seeing (on an old PC?). 30M n/s doesn't sound right.

-- Ed

[Rein Halbersma]

I'm very curious about the single core speed of Kingsrow 10x10 which is very high.
Did you do a simple search? Maybe the game is more tactical and needs a more tactical search, is that it?
My spanish checkers engine only does 2Mns in my old pc, while yours does 30Mns.

I think kingsrow is in the same speed ballpark as other 10x10 engines.
On my Xeon E5-2630 v3 @ 2.4GHz I get only 1/4 the speed you're seeing (on an old PC?). 30M n/s doesn't sound right.

-- Ed

I routinely get 38M n/s right out of the opening book on my Xeon @3.6 Ghz

[Ed Gilbert]

I routinely get 38M n/s right out of the opening book on my Xeon @3.6 Ghz

Maybe it's time for me to get a new PC?!

I was recalling the speed from memory. I've been working on the 8x8 version of kingsrow lately, which is a little slower. I just tested the 10x10 version, and it's about 10M n/s, so about 1/3 of what Alvaro sees on his "old" PC. Maybe Alvaro gets a new PC every 6 months? :-)

But anyway, to answer Alvaro's question, it's a pretty standard MTD-f search with hashtable, history, killer moves, LMR, and qsearch. The eval is fairly quick because it's mostly just a few table lookups into tables of pattern weights, but when I had a more traditional manually written and tuned eval a few years ago the search speed was only about 25% slower.

-- Ed

[Alvaro]

I routinely get 38M n/s right out of the opening book on my Xeon @3.6 Ghz

Maybe it's time for me to get a new PC?!

I was recalling the speed from memory. I've been working on the 8x8 version of kingsrow lately, which is a little slower. I just tested the 10x10 version, and it's about 10M n/s, so about 1/3 of what Alvaro sees on his "old" PC. Maybe Alvaro gets a new PC every 6 months?

But anyway, to answer Alvaro's question, it's a pretty standard MTD-f search with hashtable, history, killer moves, LMR, and qsearch. The eval is fairly quick because it's mostly just a few table lookups into tables of pattern weights, but when I had a more traditional manually written and tuned eval a few years ago the search speed was only about 25% slower.

-- Ed

Thanks Ed, for the feedback.
My PC is from 2009, I know people that upgrade every two years, and a few every year, but that is not only out of my reach as it seems irrational to me!
Anyway my slow speed must be because of my slow eval, I'm really doing a lot of stuff in it, which maybe I should get rid off.
I'm taxing the general case with a lot of exceptions in the eval and so it gets slow.
Maybe when I upgrade to a Ryzen 3000 series, I'll have more testing power to find out what I should remove from the eval.
Your engines would scream on these new cpus.
Also I detected I was counting nodes in some places I shouldn't. I only count nodes when I make a move on the board.

best regards,
Alvaro