wacnew.epd & single search improvements (extensions

[Stefan Knappe]

Geschrieben von: / Posted by: Stefan Knappe at 19 June 2003 09:11:42:

Hi,
short description of the test:

last week I started to research the effect of isolated improvements
of my program (Matador) on the playing strength of the program.
At first I compiled a version of Matador without any improvements.
This version worked without any kind of sorting, extensions, pruning,
hash … The evalution based exclusively on material.

I finished the next tests. Matador had to search all 300 positions with pure alpha-beta algo with zero window, quiescence search and some kinds of check extensions.
The results of pure alpha-beta algo:
//Test: wacnew.epd / Time: 5s
total solve time : 952.17 s / solved: 115/300
avrg. depth: 5.5 / avrg. max depth: 21.6
avrg. QS: 77.8% / avrg. MO: 61.7%
The results with some extensions:
//Test: wacnew.epd / Time: 5s
//with CheckExtensions
total solve time : 678.16 s / solved: 173/300
avrg. depth: 5.4 / avrg. max depth: 21.6
avrg. QS: 72.6% / avrg. MO: 63.3%
Matador extends the search by a full ply, if one side is in check.
I think, it isn’t a big surprise that this extension works perfektly Matador could solve approx. 50% more positions. The program needed only 71% of the reference time.
//Test: wacnew.epd / Time: 5s
//with allCheckExtensions
total solve time : 628.2 s / solved: 182/300
avrg. depth: 4.9 / avrg. max depth: 24.1
avrg. QS: 65.5% / avrg. MO: 64.3%
In this test Matador could extend until 7/4 plys without limiting of extensions! I activated all kinds of check extensions (1 ply – CheckExtensions; 1/4 ply – 2 or more checks in a line; 1/4 ply – Check by other piece (not by the moved piece); 1/4 ply – Check by two pieces).
In the normal configuration of Matador these extreme extensions are not allowed.
Best regards,
Stefan

[Uri Blass]

Geschrieben von: / Posted by: Uri Blass at 19 June 2003 09:59:15:
Als Antwort auf: / In reply to: wacnew.epd & single search improvements (extensions II) geschrieben von: / posted by: Stefan Knappe at 19 June 2003 09:11:42:

Hi,
short description of the test:

last week I started to research the effect of isolated improvements
of my program (Matador) on the playing strength of the program.
At first I compiled a version of Matador without any improvements.
This version worked without any kind of sorting, extensions, pruning,
hash … The evalution based exclusively on material.

I finished the next tests. Matador had to search all 300 positions with pure alpha-beta algo with zero window, quiescence search and some kinds of check extensions.
The results of pure alpha-beta algo:
//Test: wacnew.epd / Time: 5s
total solve time : 952.17 s / solved: 115/300
avrg. depth: 5.5 / avrg. max depth: 21.6
avrg. QS: 77.8% / avrg. MO: 61.7%
The results with some extensions:
//Test: wacnew.epd / Time: 5s
//with CheckExtensions
total solve time : 678.16 s / solved: 173/300
avrg. depth: 5.4 / avrg. max depth: 21.6
avrg. QS: 72.6% / avrg. MO: 63.3%

I am surprised that the check extension costed you only 0.1 ply.
I expected more than 0.1 ply difference because of check extension espacially when I expect the mates in 2 to be solved at smaller depth so I expect the program to get smaller depth in all the positions when there is a mate in 2 because I assume that you stop to search after finding mate in 2 because it is clear that you are not going to find mate in 1
if you did not find it at depth 1 or 2.
Uri

[Stefan Knappe]

Geschrieben von: / Posted by: Stefan Knappe at 19 June 2003 10:06:33:
Als Antwort auf: / In reply to: Re: wacnew.epd & single search improvements (extensions II) geschrieben von: / posted by: Uri Blass at 19 June 2003 09:59:15:

Hi Uri,

I am surprised that the check extension costed you only 0.1 ply.
I expected more than 0.1 ply difference because of check extension espacially when I expect the mates in 2 to be solved at smaller depth so I expect the program to get smaller depth in all the positions when there is a mate in 2 because I assume that you stop to search after finding mate in 2 because it is clear that you are not going to find mate in 1

before I started the test, I decided to continue the search after finding a mate. I hope, that due to continueing of the search the average depth is more comparable!
Best regards,
Stefan

[Stefan Knappe]

Geschrieben von: / Posted by: Stefan Knappe at 19 June 2003 10:16:10:
Als Antwort auf: / In reply to: wacnew.epd & single search improvements (extensions II) geschrieben von: / posted by: Stefan Knappe at 19 June 2003 09:11:42:

Hi,
before I start to recheck all with move ordering, I checked a combination of check extensions and oneReply extension without limiting! It worked pretty well!
For comparison the old results:

The results of pure alpha-beta algo:
//Test: wacnew.epd / Time: 5s
total solve time : 952.17 s / solved: 115/300
avrg. depth: 5.5 / avrg. max depth: 21.6
avrg. QS: 77.8% / avrg. MO: 61.7%
//Test: wacnew.epd / Time: 5s
//with allCheckExtensions
total solve time : 628.2 s / solved: 182/300
avrg. depth: 4.9 / avrg. max depth: 24.1
avrg. QS: 65.5% / avrg. MO: 64.3%

//Test: wacnew.epd / Time: 5s
//with allCheckExtensions + OneReplayExtensions
total solve time : 565.29 s / solved: 193/300
avrg. depth: 4.5 / avrg. max depth: 25.1
avrg. QS: 60.8% / avrg. MO: 66.0%
The number of solved positions is increased, the needed time is decreased. I will try this combination in later test too.
Best regards,
Stefan