6 piece scorpio bases
6 piece scorpio bases
Does anyone have a status on attempts to make 6 piece scorpio bases?
-
Denis P. Mendoza
- Posts: 476
- Joined: Wed Jun 20, 2007 2:09 am
- Sign-up code: 0
- Location: Philippines
Re: 6 piece scorpio bases
Sorry for the late response, but I guess Daniel Shawul is the right person to ask.
I'm as interested as you are since the bitbases of Scorpio are the best in this area - and fast too. I have tried generating the 3-4-5 men before using the old code (v1.0) but stopped as the 5-men were released. Since my programming skills are limited, I would gladly help anyone try to develop a 6-men generator, since the egbb_src (v3.1) was already released.
I'm as interested as you are since the bitbases of Scorpio are the best in this area - and fast too. I have tried generating the 3-4-5 men before using the old code (v1.0) but stopped as the 5-men were released. Since my programming skills are limited, I would gladly help anyone try to develop a 6-men generator, since the egbb_src (v3.1) was already released.
Re: 6 piece scorpio bases
If we had code to generate 6 piece scorpio bases and are usable by Scorpio and various other engines that have implemented the Scorpio BB's I would happily devote some CPU time to the generation...
Shaun
Shaun
Check out Blitz 40/4 at http://computerchess.org.uk/ccrl/404/
-
Denis P. Mendoza
- Posts: 476
- Joined: Wed Jun 20, 2007 2:09 am
- Sign-up code: 0
- Location: Philippines
Re: 6 piece scorpio bases
Many thanks Shaun. I'll upload a link for the old (v1.0) bitbase generator bundled with the new one (v3.1). In this way, some genius might find a way to make our dreams come true.ShaunBrewer wrote:If we had code to generate 6 piece scorpio bases and are usable by Scorpio and various other engines that have implemented the Scorpio BB's I would happily devote some CPU time to the generation...
Shaun
Thank Daniel Shawul for sharing the source for his egbb.dll probe. You can now build egbb.dll for your OS (32/64 bit). Just request and I'll compile one for any Windows platform. For Linux, somebody from the WB forum who also requested this to DS might help.
http://dshawul.googlepages.com/egbbcomp_src.zip
http://dshawul.googlepages.com/egbbdll_src.zip
This is the original from Dann Corbit's old/defunct site. I bundled the complete set (egbb.dll, generator and compressor, binaries and src) before, in case the zipped files were related:
http://mihd.net/v4qr27
Re: 6 piece scorpio bases
Hi Denis,Denis P. Mendoza wrote:Thank Daniel Shawul for sharing the source for his egbb.dll probe. You can now build egbb.dll for your OS (32/64 bit). Just request and I'll compile one for any Windows platform.
if you have the time a 64-bit compile of the egbb.dll and Scorpio 2 would be fantastic - we can then see what sort of speed-up 64-bit gives the latest Scorpio and can run it without loosing egbb access...
Shaun
P.S. I have just finished a gauntlet with Scorpio 2.0 4CPU with 5 men egbbs I am going to repeat without the egbbs to see how much they give Scorpio...
Check out Blitz 40/4 at http://computerchess.org.uk/ccrl/404/
-
Denis P. Mendoza
- Posts: 476
- Joined: Wed Jun 20, 2007 2:09 am
- Sign-up code: 0
- Location: Philippines
Re: 6 piece scorpio bases
I already mentioned this to Jim Ablett as he is one of the official compilers of Scorpio, But I'll build one for you nuch the the same as Jim's compiles. The 64-bit egbb.dll goes with it too as prerequisite -tried and tested.
-
Denis P. Mendoza
- Posts: 476
- Joined: Wed Jun 20, 2007 2:09 am
- Sign-up code: 0
- Location: Philippines
Re: 6 piece scorpio bases
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------ShaunBrewer wrote:Hi Denis,Denis P. Mendoza wrote:Thank Daniel Shawul for sharing the source for his egbb.dll probe. You can now build egbb.dll for your OS (32/64 bit). Just request and I'll compile one for any Windows platform.
if you have the time a 64-bit compile of the egbb.dll and Scorpio 2 would be fantastic - we can then see what sort of speed-up 64-bit gives the latest Scorpio and can run it without loosing egbb access...
Shaun
P.S. I have just finished a gauntlet with Scorpio 2.0 4CPU with 5 men egbbs I am going to repeat without the egbbs to see how much they give Scorpio...
Sorry for the late response Shaun. I've been having problems with Scorpio 64 bit. It seems it follows a different scorpio.ini parameter (for the hashes) than 32-bit setup. I didn't know that as I'm not an occasional 64-bit user of Scorpio. I see Scorpio too crashes without the scorpio.ini and personality folder. I haven't tested these compiles long enough to check if it is still buggy as others say. I built a MSVC8, MSVC9, Intelx64, and Intelx64 with Openmp and the 64-bit egbb.dll. Please choose whichever fits you most. I'm sure there will be one best compile for you.
http://mihd.net/yprthz
Merry Xmas and a happy new year to all our members here as well as CCRL!
-
Denis P. Mendoza
- Posts: 476
- Joined: Wed Jun 20, 2007 2:09 am
- Sign-up code: 0
- Location: Philippines
Re: 6 piece scorpio bases
Take note also that Scorpio takes a very long time to complete its initialization during each start-up. I tested a 64-bit Toga 1.3.4 with egbb, and it initializes faster using the same 64-bit egmprobe dll. This maybe the reason why Scorpio takes a long time to start a game. You'll have to wait till the "Egbbs loaded" displays!
-
Denis P. Mendoza
- Posts: 476
- Joined: Wed Jun 20, 2007 2:09 am
- Sign-up code: 0
- Location: Philippines
Re: 6 piece scorpio bases
I spoke too soon. Jim already made a much faster compile than mine based on news at CCC. Master Jim does it better though! Just informing.
I better look for some ways to improve mine
and probably look for a way to tweak the code so that the dlls/bitbases of 32 and 64 bit will not be transferrred /duplicated.
I better look for some ways to improve mine
and probably look for a way to tweak the code so that the dlls/bitbases of 32 and 64 bit will not be transferrred /duplicated.Re: 6 piece scorpio bases
Hi Denis,
I had found JA 64-bit compile and I have a test running... ***
Will let you know how it performs...
MERRY CHRISTMAS and a HAPPY NEW YEAR!!!
Shaun
***
First finished 100 games the 32-bit version was ahead... running another 100
I had found JA 64-bit compile and I have a test running... ***
Will let you know how it performs...
MERRY CHRISTMAS and a HAPPY NEW YEAR!!!
Shaun
***
First finished 100 games the 32-bit version was ahead... running another 100
New game - Scorpio 2.0 32-bit, Blitz:4'+2" Brighton
rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1
Analysis by Scorpio 2.0 32-bit:
1.e2-e4 e7-e5
= (0.10) Depth: 2 00:00:00
1.e2-e4 e7-e5
= (0.10) Depth: 2 00:00:00
1.e2-e4 d7-d5 2.e4xd5 Qd8xd5 3.d2-d4
+/= (0.45) Depth: 3 00:00:00
1.e2-e4 d7-d5 2.e4xd5 Qd8xd5 3.d2-d4
+/= (0.45) Depth: 3 00:00:00
1.e2-e4 Ng8-f6 2.Nb1-c3 e7-e5
= (0.24) Depth: 4 00:00:00
1.e2-e4 Ng8-f6 2.Nb1-c3 e7-e5
= (0.24) Depth: 4 00:00:00
1.e2-e4 d7-d5 2.Nb1-c3 Ng8-f6 3.Bf1-b5+ Nb8-c6
+/= (0.28) Depth: 5 00:00:00
1.Ng1-f3 d7-d5 2.d2-d4 Bc8-f5 3.Nf3-e5
+/= (0.36) Depth: 5 00:00:00
1.Ng1-f3 d7-d5 2.d2-d4 Bc8-f5 3.Nf3-e5
+/= (0.36) Depth: 5 00:00:00
1.Ng1-f3 d7-d5 2.d2-d4 Nb8-c6 3.Bc1-f4 Bc8-f5
= (0.09) Depth: 6 00:00:00 19kN
1.e2-e4 d7-d5 2.Nb1-c3 d5-d4 3.Nc3-d5 e7-e5
= (0.25) Depth: 6 00:00:00 34kN
1.e2-e4 d7-d5 2.Nb1-c3 d5-d4 3.Nc3-d5 e7-e5
= (0.25) Depth: 6 00:00:00 38kN
1.e2-e4 Nb8-c6 2.d2-d4 d7-d5 3.e4-e5 Bc8-f5 4.Bf1-b5
= (0.22) Depth: 7 00:00:00 112kN
1.e2-e4 Nb8-c6 2.d2-d4 d7-d5 3.e4-e5 Bc8-f5 4.Bf1-b5
= (0.22) Depth: 7 00:00:00 124kN
1.e2-e4 e7-e6 2.d2-d4 d7-d5 3.e4-e5 Bc8-d7 4.Bc1-d2 Nb8-c6
+/= (0.26) Depth: 8 00:00:00 361kN
1.e2-e4 e7-e6 2.d2-d4 d7-d5 3.e4-e5 Bc8-d7 4.Bc1-d2 Nb8-c6
+/= (0.26) Depth: 8 00:00:00 425kN
1.e2-e4 e7-e6 2.d2-d4 Ng8-f6 3.Nb1-c3 Bf8-b4 4.Bc1-g5 0-0 5.Bf1-c4
= (0.25) Depth: 9 00:00:00 676kN
1.e2-e4 e7-e6 2.d2-d4 Ng8-f6 3.Nb1-c3 Bf8-b4 4.Bc1-g5 0-0 5.Bf1-c4
= (0.25) Depth: 9 00:00:01 883kN
1.e2-e4 e7-e6 2.d2-d4 d7-d5 3.e4-e5 Bc8-d7 4.Ng1-f3 f7-f6 5.Bf1-d3 Bf8-b4+ 6.c2-c3
+/= (0.40) Depth: 10 00:00:02 1989kN
1.e2-e4 e7-e6 2.d2-d4 d7-d5 3.e4-e5 Bc8-d7 4.Ng1-f3 f7-f6 5.Bf1-d3 Bf8-b4+ 6.c2-c3
+/= (0.40) Depth: 10 00:00:02 2084kN
1.e2-e4 e7-e5 2.Nb1-c3 Nb8-c6 3.Ng1-f3 Ng8-f6 4.d2-d4 e5xd4 5.Nf3xd4 Bf8-b4 6.Bc1-g5 0-0 7.Bf1-c4 Nc6xd4 8.Qd1xd4
+/= (0.31) Depth: 11 00:00:05 4961kN
1.e2-e4 e7-e5 2.Nb1-c3 Nb8-c6 3.Ng1-f3 Ng8-f6 4.d2-d4 e5xd4 5.Nf3xd4 Bf8-b4 6.Bc1-g5 0-0 7.Bf1-c4 Nc6xd4 8.Qd1xd4
+/= (0.31) Depth: 11 00:00:07 6021kN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 Qd8-e7 4.d2-d4 d7-d6 5.Ne5-f3 Nf6xe4 6.Bf1-b5+ c7-c6 7.Bb5-c4
= (0.17) Depth: 12 00:00:11 9728kN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 Qd8-e7 4.d2-d4 d7-d6 5.Ne5-f3 Nf6xe4 6.Bf1-b5+ c7-c6 7.Bb5-c4
= (0.17) Depth: 12 00:00:14 12646kN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 d7-d6 4.Ne5-c4 Nf6xe4 5.d2-d3 Ne4-f6 6.Bc1-g5 Qd8-e7+ 7.Bf1-e2 Bc8-g4
+/= (0.38) Depth: 13 00:00:19 16389kN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 d7-d6 4.Ne5-c4 Nf6xe4 5.d2-d3 Ne4-f6 6.Bc1-g5 Qd8-e7+ 7.Bf1-e2 Bc8-g4
+/= (0.38) Depth: 13 00:00:21 18241kN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 d7-d6 4.Ne5-f3 Nf6xe4 5.d2-d3 Ne4-c5 6.d3-d4 Nc5-e4 7.Nb1-d2 f7-f5 8.Nd2xe4 f5xe4
+/= (0.34) Depth: 14 00:00:33 28258kN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 d7-d6 4.Ne5-f3 Nf6xe4 5.d2-d3 Ne4-c5 6.d3-d4 Nc5-e4 7.Nb1-d2 f7-f5 8.Nd2xe4 f5xe4
+/= (0.34) Depth: 14 00:00:38 32700kN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 d7-d6 4.Ne5-f3 Nf6xe4 5.d2-d3 Ne4-f6 6.Bf1-e2 Bf8-e7 7.0-0 0-0 8.Nf3-d4
+/= (0.32) Depth: 15 00:00:59 50746kN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 d7-d6 4.Ne5-f3 Nf6xe4 5.d2-d3 Ne4-f6 6.Bf1-e2 Bf8-e7 7.0-0 0-0 8.Nf3-d4
+/= (0.32) Depth: 15 00:01:13 63028kN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 d7-d6 4.Ne5-f3 Nf6xe4 5.Qd1-e2 Qd8-e7 6.d2-d3 Ne4-c5 7.d3-d4 Nc5-e4 8.Nb1-d2 Ne4xd2 9.Qe2xe7+ Bf8xe7 10.Bc1xd2
= (0.21) Depth: 16 00:02:10 113mN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 d7-d6 4.Ne5-f3 Nf6xe4 5.Qd1-e2 Qd8-e7 6.d2-d3 Ne4-c5 7.d3-d4 Nc5-e4 8.Nb1-d2 Ne4xd2 9.Qe2xe7+ Bf8xe7 10.Bc1xd2
= (0.21) Depth: 16 00:02:56 154mN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 d7-d6 4.Ne5-f3 Nf6xe4 5.d2-d3 Ne4-f6 6.Bf1-e2 Bf8-e7 7.0-0 0-0 8.Nb1-c3 Bc8-f5 9.Bc1-f4
= (0.24) Depth: 17 00:04:05 214mN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 d7-d6 4.Ne5-f3 Nf6xe4 5.d2-d3 Ne4-f6 6.Bf1-e2 Bf8-e7 7.0-0 0-0 8.Nb1-c3 Bc8-f5 9.Bc1-f4
= (0.24) Depth: 17 00:05:16 275mN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.d2-d4 Nf6xe4 4.Bf1-d3 d7-d5 5.Nf3xe5 Bf8-d6 6.0-0 0-0 7.Qd1-h5 Ne4-f6 8.Qh5-e2 Rf8-e8 9.Bc1-g5 Bc8-g4
= (0.24) Depth: 18 00:08:17 438mN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.d2-d4 Nf6xe4 4.Bf1-d3 d7-d5 5.Nf3xe5 Bf8-d6 6.0-0 0-0 7.Qd1-h5 Ne4-f6 8.Qh5-e2 Rf8-e8 9.Bc1-g5 Bc8-g4
= (0.24) Depth: 18 00:11:23 603mN
1.Ng1-f3 d7-d5 2.e2-e3 Ng8-f6 3.d2-d4 e7-e6 4.Bf1-d3 Bf8-d6 5.0-0 0-0 6.c2-c4 d5xc4 7.Bd3xc4 Nb8-c6 8.Nb1-c3 e6-e5 9.Nc3-b5 Bc8-f5 10.d4-d5 Nc6-b4 11.Nb5xd6 Qd8xd6
= (0.14) Depth: 19 00:15:40 831mN
(Brewer, Brighton 23.12.2007)
New game - Scorpio 2.0 32-bit, Blitz:4'+2" Brighton
rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1
Analysis by Scorpio 2.0 64-bit:
1.e2-e4 e7-e5
= (0.10) Depth: 2 00:00:00
1.e2-e4 e7-e5
= (0.10) Depth: 2 00:00:00
1.e2-e4 d7-d5 2.e4xd5 Qd8xd5 3.d2-d4
+/= (0.45) Depth: 3 00:00:00
1.e2-e4 d7-d5 2.e4xd5 Qd8xd5 3.d2-d4
+/= (0.45) Depth: 3 00:00:00
1.e2-e4 Ng8-f6 2.Nb1-c3 e7-e5
= (0.24) Depth: 4 00:00:00
1.e2-e4 Ng8-f6 2.Nb1-c3 e7-e5
= (0.24) Depth: 4 00:00:00
1.e2-e4 d7-d5 2.Nb1-c3 Ng8-f6 3.Bf1-b5+ Nb8-c6
+/= (0.28) Depth: 5 00:00:00
1.Ng1-f3 d7-d5 2.d2-d4 Bc8-f5 3.Nf3-e5
+/= (0.36) Depth: 5 00:00:00
1.Ng1-f3 d7-d5 2.d2-d4 Bc8-f5 3.Nf3-e5
+/= (0.36) Depth: 5 00:00:00
1.Ng1-f3 d7-d5 2.d2-d4 Nb8-c6 3.Bc1-f4 Bc8-f5
= (0.09) Depth: 6 00:00:00 19kN
1.e2-e4 d7-d5 2.Nb1-c3 d5-d4 3.Nc3-d5 e7-e5
= (0.25) Depth: 6 00:00:00 34kN
1.e2-e4 d7-d5 2.Nb1-c3 d5-d4 3.Nc3-d5 e7-e5
= (0.25) Depth: 6 00:00:00 38kN
1.e2-e4 Nb8-c6 2.d2-d4 d7-d5 3.e4-e5 Bc8-f5 4.Bf1-b5
= (0.22) Depth: 7 00:00:00 112kN
1.e2-e4 Nb8-c6 2.d2-d4 d7-d5 3.e4-e5 Bc8-f5 4.Bf1-b5
= (0.22) Depth: 7 00:00:00 124kN
1.e2-e4 e7-e6 2.d2-d4 d7-d5 3.e4-e5 Bc8-d7 4.Bc1-d2 Nb8-c6
+/= (0.26) Depth: 8 00:00:00 361kN
1.e2-e4 e7-e6 2.d2-d4 d7-d5 3.e4-e5 Bc8-d7 4.Bc1-d2 Nb8-c6
+/= (0.26) Depth: 8 00:00:00 425kN
1.e2-e4 e7-e6 2.d2-d4 Ng8-f6 3.Nb1-c3 Bf8-b4 4.Bc1-g5 0-0 5.Bf1-c4
= (0.25) Depth: 9 00:00:00 676kN
1.e2-e4 e7-e6 2.d2-d4 Ng8-f6 3.Nb1-c3 Bf8-b4 4.Bc1-g5 0-0 5.Bf1-c4
= (0.25) Depth: 9 00:00:01 883kN
1.e2-e4 e7-e6 2.d2-d4 d7-d5 3.e4-e5 Bc8-d7 4.Ng1-f3 f7-f6 5.Bf1-d3 Bf8-b4+ 6.c2-c3
+/= (0.40) Depth: 10 00:00:02 1989kN
1.e2-e4 e7-e6 2.d2-d4 d7-d5 3.e4-e5 Bc8-d7 4.Ng1-f3 f7-f6 5.Bf1-d3 Bf8-b4+ 6.c2-c3
+/= (0.40) Depth: 10 00:00:02 2084kN
1.e2-e4 e7-e5 2.Nb1-c3 Nb8-c6 3.Ng1-f3 Ng8-f6 4.d2-d4 e5xd4 5.Nf3xd4 Bf8-b4 6.Bc1-g5 0-0 7.Bf1-c4 Nc6xd4 8.Qd1xd4
+/= (0.31) Depth: 11 00:00:05 4961kN
1.e2-e4 e7-e5 2.Nb1-c3 Nb8-c6 3.Ng1-f3 Ng8-f6 4.d2-d4 e5xd4 5.Nf3xd4 Bf8-b4 6.Bc1-g5 0-0 7.Bf1-c4 Nc6xd4 8.Qd1xd4
+/= (0.31) Depth: 11 00:00:06 6021kN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 Qd8-e7 4.d2-d4 d7-d6 5.Ne5-f3 Nf6xe4 6.Bf1-b5+ c7-c6 7.Bb5-c4
= (0.17) Depth: 12 00:00:10 9728kN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 Qd8-e7 4.d2-d4 d7-d6 5.Ne5-f3 Nf6xe4 6.Bf1-b5+ c7-c6 7.Bb5-c4
= (0.17) Depth: 12 00:00:13 12646kN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 d7-d6 4.Ne5-c4 Nf6xe4 5.d2-d3 Ne4-f6 6.Bc1-g5 Qd8-e7+ 7.Bf1-e2 Bc8-g4
+/= (0.38) Depth: 13 00:00:17 16389kN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 d7-d6 4.Ne5-c4 Nf6xe4 5.d2-d3 Ne4-f6 6.Bc1-g5 Qd8-e7+ 7.Bf1-e2 Bc8-g4
+/= (0.38) Depth: 13 00:00:19 18241kN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 d7-d6 4.Ne5-f3 Nf6xe4 5.d2-d3 Ne4-c5 6.d3-d4 Nc5-e4 7.Nb1-d2 f7-f5 8.Nd2xe4 f5xe4
+/= (0.34) Depth: 14 00:00:30 28258kN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 d7-d6 4.Ne5-f3 Nf6xe4 5.d2-d3 Ne4-c5 6.d3-d4 Nc5-e4 7.Nb1-d2 f7-f5 8.Nd2xe4 f5xe4
+/= (0.34) Depth: 14 00:00:35 32700kN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 d7-d6 4.Ne5-f3 Nf6xe4 5.d2-d3 Ne4-f6 6.Bf1-e2 Bf8-e7 7.0-0 0-0 8.Nf3-d4
+/= (0.32) Depth: 15 00:00:54 50746kN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 d7-d6 4.Ne5-f3 Nf6xe4 5.d2-d3 Ne4-f6 6.Bf1-e2 Bf8-e7 7.0-0 0-0 8.Nf3-d4
+/= (0.32) Depth: 15 00:01:07 63028kN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 d7-d6 4.Ne5-f3 Nf6xe4 5.Qd1-e2 Qd8-e7 6.d2-d3 Ne4-c5 7.d3-d4 Nc5-e4 8.Nb1-d2 Ne4xd2 9.Qe2xe7+ Bf8xe7 10.Bc1xd2
= (0.21) Depth: 16 00:01:59 113mN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 d7-d6 4.Ne5-f3 Nf6xe4 5.Qd1-e2 Qd8-e7 6.d2-d3 Ne4-c5 7.d3-d4 Nc5-e4 8.Nb1-d2 Ne4xd2 9.Qe2xe7+ Bf8xe7 10.Bc1xd2
= (0.21) Depth: 16 00:02:41 154mN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 d7-d6 4.Ne5-f3 Nf6xe4 5.d2-d3 Ne4-f6 6.Bf1-e2 Bf8-e7 7.0-0 0-0 8.Nb1-c3 Bc8-f5 9.Bc1-f4
= (0.24) Depth: 17 00:03:44 214mN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.Nf3xe5 d7-d6 4.Ne5-f3 Nf6xe4 5.d2-d3 Ne4-f6 6.Bf1-e2 Bf8-e7 7.0-0 0-0 8.Nb1-c3 Bc8-f5 9.Bc1-f4
= (0.24) Depth: 17 00:04:48 275mN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.d2-d4 Nf6xe4 4.Bf1-d3 d7-d5 5.Nf3xe5 Bf8-d6 6.0-0 0-0 7.Qd1-h5 Ne4-f6 8.Qh5-e2 Rf8-e8 9.Bc1-g5 Bc8-g4
= (0.24) Depth: 18 00:07:33 438mN
1.e2-e4 e7-e5 2.Ng1-f3 Ng8-f6 3.d2-d4 Nf6xe4 4.Bf1-d3 d7-d5 5.Nf3xe5 Bf8-d6 6.0-0 0-0 7.Qd1-h5 Ne4-f6 8.Qh5-e2 Rf8-e8 9.Bc1-g5 Bc8-g4
= (0.24) Depth: 18 00:10:22 603mN
1.Ng1-f3 d7-d5 2.e2-e3 Ng8-f6 3.d2-d4 e7-e6 4.Bf1-d3 Bf8-d6 5.0-0 0-0 6.c2-c4 d5xc4 7.Bd3xc4 Nb8-c6 8.Nb1-c3 e6-e5 9.Nc3-b5 Bc8-f5 10.d4-d5 Nc6-b4 11.Nb5xd6 Qd8xd6
= (0.14) Depth: 19 00:14:17 831mN
(Brewer, Brighton 23.12.2007)
so 64-bit is fasterCheck out Blitz 40/4 at http://computerchess.org.uk/ccrl/404/
Re: 6 piece scorpio bases
Werner posted some results on CCC - 64-bit is 10% faster, not very much
-
Denis P. Mendoza
- Posts: 476
- Joined: Wed Jun 20, 2007 2:09 am
- Sign-up code: 0
- Location: Philippines
Re: 6 piece scorpio bases
Yes, you are absolutely right. When the Scorpio 2.0 src was released, I already tried compiling a 64-bit version and was still unsure of myself because there was not much improvement in speed. Since 64-bit is new to me, I'd just waited for Jim to buy a new 64-bit hardware (as he hasn't one yet then). Now, I was sure of these findings. The Intel compiler for 32-bit has more compiler options to speed-up compiles at present, so 64-bit coders must manually optimized their code to benefit on the 64-bit platform during compiling.Ray Banks wrote:Werner posted some results on CCC - 64-bit is 10% faster, not very much
At least now, 64-bit engines using Scorpio bitbases could use the tables as an added feature.
Re: 6 piece scorpio bases
Yes, this is very worthwhileDenis P. Mendoza wrote: At least now, 64-bit engines using Scorpio bitbases could use the tables as an added feature.
-
Denis P. Mendoza
- Posts: 476
- Joined: Wed Jun 20, 2007 2:09 am
- Sign-up code: 0
- Location: Philippines
Re: 6 piece scorpio bases
I just gone home and tinkered a bit this new 6 piece bitbase generator code by Daniel Shawul.
https://github.com/dshawul
I can't find yet a specific command parameter to adjust but it works instantly! Just execute in console mode and it generates and compresses automatically.
I'm still on the 5pc. stage so I can't verify yet if 6pc is stable or working properly.
Anyone who has a faster hardware can test my beta compile using MSVC6.
See attachment.
https://github.com/dshawul
I can't find yet a specific command parameter to adjust but it works instantly! Just execute in console mode and it generates and compresses automatically.
I'm still on the 5pc. stage so I can't verify yet if 6pc is stable or working properly.
Anyone who has a faster hardware can test my beta compile using MSVC6.
See attachment.
- Attachments
-
- dshawul-6pc-bitbase-generator.7z
- (103.1 KiB) Downloaded 1037 times
Re: 6 piece scorpio bases
It is working here, but as with you, it is on the 5-men at the moment. It seems pretty slow.
-
Denis P. Mendoza
- Posts: 476
- Joined: Wed Jun 20, 2007 2:09 am
- Sign-up code: 0
- Location: Philippines
Re: 6 piece scorpio bases
Yes, it is slow. I hope my 4 GB memory is enough when the 6pc comes.Ray wrote:It is working here, but as with you, it is on the 5-men at the moment. It seems pretty slow.
I'm still on the 5pc generation stage too since yesterday evening.
Re: 6 piece scorpio bases
Disk space is potentially more of an issue.... How much will be required for the full set ?Denis P. Mendoza wrote:Yes, it is slow. I hope my 4 GB memory is enough when the 6pc comes.Ray wrote:It is working here, but as with you, it is on the 5-men at the moment. It seems pretty slow.
I'm still on the 5pc generation stage too since yesterday evening.
And assuming it finishes, what engines can use the 6-piece set ?
Re: 6 piece scorpio bases
...and (I think it's not the same question) how big will be the 6-men bitbases?Ray wrote:How much will be required for the full set ?
I recall that SMK said that his projected Shredder's 6-men bitbases (shredderbases) would be 40 GB in size (compressed).
-
Denis P. Mendoza
- Posts: 476
- Joined: Wed Jun 20, 2007 2:09 am
- Sign-up code: 0
- Location: Philippines
Re: 6 piece scorpio bases
You're right. I have no idea how big will it be.
Daniel Shawul may need to modify the probing code if ever a 6pc database is available to test.
(BTW, I was not worried by the diskspace as I still have enough but the ram memory.)
Thanks.
Daniel Shawul may need to modify the probing code if ever a 6pc database is available to test.
(BTW, I was not worried by the diskspace as I still have enough but the ram memory.)
Thanks.
-
Denis P. Mendoza
- Posts: 476
- Joined: Wed Jun 20, 2007 2:09 am
- Sign-up code: 0
- Location: Philippines
Re: 6 piece scorpio bases
Just to update,it was only today that I saw the 6 pc bitbase being created. But sad to say it only created kqqqqk, krrrrk, kbbbbk as examples and the rest were empty bases due to to memory allocation failures based on the log file. Some 6pc need more ram and failed as I'm using the pc memory for other applications which may have affected it. I'm not sure if this was the case but will retry again.
I'll just test this quite big ~859mb 3-4-5 men and see the difference with the previous batch.
(I'll probably contact Daniel Shawul concerning stability of the code, and Jim or Dann as they had more experience in dealing with generating/compressing scorpio bitbase generators).
I'll just test this quite big ~859mb 3-4-5 men and see the difference with the previous batch.
(I'll probably contact Daniel Shawul concerning stability of the code, and Jim or Dann as they had more experience in dealing with generating/compressing scorpio bitbase generators).
Re: 6 piece scorpio bases
I stopped mine while it was still on 5-men, becuase I needed the machine for something else. Maybe I'll restart them tonight, although I wonder if it will pick up where it left off or start all over again
Re: 6 piece scorpio bases
I restarted mine. It picked up where it left off
-
Denis P. Mendoza
- Posts: 476
- Joined: Wed Jun 20, 2007 2:09 am
- Sign-up code: 0
- Location: Philippines
Re: 6 piece scorpio bases
It's good to know, thanks.Ray wrote:I restarted mine. It picked up where it left off
I already sent a pm to Daniel Shawul (@ CCC) so that we could test whatever 6pc base files it generates.
Re: 6 piece scorpio bases
Mine is now on it's first 6 piece bitbase. I have 8GB of RAM so it shouldn't fail through lack of RAM (maybe)
-
Denis P. Mendoza
- Posts: 476
- Joined: Wed Jun 20, 2007 2:09 am
- Sign-up code: 0
- Location: Philippines
Re: 6 piece scorpio bases
I had used the new egbbprobe 4.0 included and compiled a 64-bit dll for testing (to see how it responds). Sad to say it can only load the 3-4 men (in 5 seconds) using Scorpio 2.7 (32/64 bit). When I tried 3-4-5 men, Scorpio just displays "Egbb loading" and stays that way. Version 3.3 as an example even loads its 3-4-5 men in 0 seconds - very fast. I think Scorpio is not yet compatible using the new v4.0 eggbbprobe and bitbases though I see Toga 1.4 beta5c responding very fast with the new 3-4-5 men. I'll run some formal endgame epd test on compatible engines and see the difference (plus without the bases for comparison).
Note: Using Arena at the moment, I placed Toga engine in the new bitbase directory and just left the bitbases path empty in the configuration dialogue box! AFAIK, default should be C:\egbb\ (not C:\egbb or same case if placed in other locations on the disk).
I'll attach a test dll for 64-bit.
Note: Using Arena at the moment, I placed Toga engine in the new bitbase directory and just left the bitbases path empty in the configuration dialogue box! AFAIK, default should be C:\egbb\ (not C:\egbb or same case if placed in other locations on the disk).
I'll attach a test dll for 64-bit.
- Attachments
-
- egbbdll64.7z
- egbbprobe 4.0 64-bit test dll
- (19.75 KiB) Downloaded 784 times