First blitz impressions: Senpai 1.0 et al

[Mike S.]

2m+0,5s, i5-3210M/2,5...2,9 GHz
1 thread, ponder off
256 MB hashtables each
Correspondence chess openings
up to move 10, repeated reversed
Arena 3.5, Windows 8 x64

Code: Select all

   Engine           Score   Se Ho Ry Ga Fr Ni Ch Cr   S-B
1: Senpai 1.0       12,5/14 ·· =1 == 11 11 11 11 11   72,25
2: Houdini 1.5a     10,5/14 =0 ·· 1= 01 11 =1 11 11   59,00
3: Rybka 2.3.2a     9,5/14  == 0= ·· =1 == 11 1= 11   53,75
4: Gaviota 1.0      6,5/14  00 10 =0 ·· 0= == 1= 11   34,50
5: Fruit re 2.1     6,0/14  00 00 == 1= ·· =0 =1 =1   31,75
6: Nirvanachess 1.5 4,0/14  00 =0 00 == =1 ·· =0 =0   24,25
7: Cheng  4_0.36c   3,5/14  00 00 0= 0= =0 =1 ·· 0=   18,75
8: Crafty 23.8      3,5/14  00 00 00 00 =0 =1 1= ··   14,25

56 games played / Tournament is finished
Name of the tournament: Engineflut_2014

Impressive But relative to the intermediate result from the current IPON test (2000+ games already) Senpai seems to have overperformed here. - Maybe the reversed is true for Fruit reloaded. Future will tell. I thank all programmers for their great effort.

[IanO]

Fabien has mentioned Senpai is on par with Protector in strength. It appears to have a better understanding of minor piece usage in the endgame, often obtaining a win with minor and pawns vs. rook. In my testing, Senpai has a slight edge over Protector 1.5 at blitz controls (+27=33-19 so far). The search appears to be more efficient as well, obtaining three ply more depth despite three times slower nps.

[Mike S.]

In my testing, Senpai has a slight edge over Protector 1.5 at blitz controls (+27=33-19 so far).

That tendency corresponds with the intermediate head-to-head result at IPON:

Code: Select all

Senpai 1.0 - Protector 1.5.0 (2834)		74.5	-	70.5		51.38%		Perf=2843

Promising.

[ernest]

In my testing, Senpai has a slight edge over Protector 1.5 at blitz controls (+27=33-19 so far).

That tendency ...

Code: Select all

Senpai 1.0 - Protector 1.5.0 (2834)		74.5	-	70.5		51.38%		Perf=2843

Again, with less than 150 games, this only means that both engines are on par (as Fabien said).

There is no tendency (other than giving hasty interpretation to partial results )...

[Sedat Canbaz]

2m+0,5s, i5-3210M/2,5...2,9 GHz
1 thread, ponder off
256 MB hashtables each
Correspondence chess openings
up to move 10, repeated reversed
Arena 3.5, Windows 8 x64

Code: Select all

   Engine           Score   Se Ho Ry Ga Fr Ni Ch Cr   S-B
1: Senpai 1.0       12,5/14 ·· =1 == 11 11 11 11 11   72,25
2: Houdini 1.5a     10,5/14 =0 ·· 1= 01 11 =1 11 11   59,00
3: Rybka 2.3.2a     9,5/14  == 0= ·· =1 == 11 1= 11   53,75
4: Gaviota 1.0      6,5/14  00 10 =0 ·· 0= == 1= 11   34,50
5: Fruit re 2.1     6,0/14  00 00 == 1= ·· =0 =1 =1   31,75
6: Nirvanachess 1.5 4,0/14  00 =0 00 == =1 ·· =0 =0   24,25
7: Cheng  4_0.36c   3,5/14  00 00 0= 0= =0 =1 ·· 0=   18,75
8: Crafty 23.8      3,5/14  00 00 00 00 =0 =1 1= ··   14,25

56 games played / Tournament is finished
Name of the tournament: Engineflut_2014

Impressive But relative to the intermediate result from the current IPON test (2000+ games already) Senpai seems to have overperformed here. - Maybe the reversed is true for Fruit reloaded. Future will tell. I thank all programmers for their great effort.

Thanks for the update dear Mike!

Btw,
What kind of this version is Senpai 1.0 et al ?

For example,
In Turkish means et al = take meat


Best,
Sedat

[michiguel]

2m+0,5s, i5-3210M/2,5...2,9 GHz
1 thread, ponder off
256 MB hashtables each
Correspondence chess openings
up to move 10, repeated reversed
Arena 3.5, Windows 8 x64

Code: Select all

   Engine           Score   Se Ho Ry Ga Fr Ni Ch Cr   S-B
1: Senpai 1.0       12,5/14 ·· =1 == 11 11 11 11 11   72,25
2: Houdini 1.5a     10,5/14 =0 ·· 1= 01 11 =1 11 11   59,00
3: Rybka 2.3.2a     9,5/14  == 0= ·· =1 == 11 1= 11   53,75
4: Gaviota 1.0      6,5/14  00 10 =0 ·· 0= == 1= 11   34,50
5: Fruit re 2.1     6,0/14  00 00 == 1= ·· =0 =1 =1   31,75
6: Nirvanachess 1.5 4,0/14  00 =0 00 == =1 ·· =0 =0   24,25
7: Cheng  4_0.36c   3,5/14  00 00 0= 0= =0 =1 ·· 0=   18,75
8: Crafty 23.8      3,5/14  00 00 00 00 =0 =1 1= ··   14,25

56 games played / Tournament is finished
Name of the tournament: Engineflut_2014

Impressive But relative to the intermediate result from the current IPON test (2000+ games already) Senpai seems to have overperformed here. - Maybe the reversed is true for Fruit reloaded. Future will tell. I thank all programmers for their great effort.

Thanks for the update dear Mike!

Btw,
What kind of this version is Senpai 1.0 et al ?

For example,
In Turkish means et al = take meat


Best,
Sedat

et al. (should have a period) is a common abreviation in scientific references. It means, "and others" from et alii (latin).

Miguel

[Sedat Canbaz]

2m+0,5s, i5-3210M/2,5...2,9 GHz
1 thread, ponder off
256 MB hashtables each
Correspondence chess openings
up to move 10, repeated reversed
Arena 3.5, Windows 8 x64

Code: Select all

   Engine           Score   Se Ho Ry Ga Fr Ni Ch Cr   S-B
1: Senpai 1.0       12,5/14 ·· =1 == 11 11 11 11 11   72,25
2: Houdini 1.5a     10,5/14 =0 ·· 1= 01 11 =1 11 11   59,00
3: Rybka 2.3.2a     9,5/14  == 0= ·· =1 == 11 1= 11   53,75
4: Gaviota 1.0      6,5/14  00 10 =0 ·· 0= == 1= 11   34,50
5: Fruit re 2.1     6,0/14  00 00 == 1= ·· =0 =1 =1   31,75
6: Nirvanachess 1.5 4,0/14  00 =0 00 == =1 ·· =0 =0   24,25
7: Cheng  4_0.36c   3,5/14  00 00 0= 0= =0 =1 ·· 0=   18,75
8: Crafty 23.8      3,5/14  00 00 00 00 =0 =1 1= ··   14,25

56 games played / Tournament is finished
Name of the tournament: Engineflut_2014

Impressive But relative to the intermediate result from the current IPON test (2000+ games already) Senpai seems to have overperformed here. - Maybe the reversed is true for Fruit reloaded. Future will tell. I thank all programmers for their great effort.

Thanks for the update dear Mike!

Btw,
What kind of this version is Senpai 1.0 et al ?

For example,
In Turkish means et al = take meat


Best,
Sedat

et al. (should have a period) is a common abreviation in scientific references. It means, "and others" from et alii (latin).

Miguel

Thanks...now I see

I thought en al is a different version than the below one