Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / Most important 6-man tablebase files for Rybka 4?
1 2 3 Previous Next  
- - By hal9000 (**) [no] Date 2010-06-06 15:09
I've copied around 70 GB of 6-man tablebase files over to my SSD (and all 3-4-5 men tablebases). I have 200 GB of 6-man tablebase files on my HDD and my plan is to copy all the files over to a larger SSD later.

But given a 70 GB space restriction, which tablebase files do you feel is most important for Rybka 4? It seems to me that rook endgames are important, since Rybka's rook endgame evaluations tend to be less accurate.

Here's a list of the endgame files I'm currently storing on my SSD (the numbers indicate the rank of the given endgame in computer vs computer matches based on frequency):

1  krpkrp
2  krppkr
3  kppkpp
4  kbpknp
5  kbpkpp
9  kbpkbp
10  krbkrp
11  knpkpp
13  krpkbp
15  kqpkqp
17  krnkrp
18  knpknp
21  krpknp
23  kqpkrp
26  kpppkr
28  kqppkq
40  knppkr
Parent - - By Lukas Cimiotti (Bronze) [de] Date 2010-06-06 15:33
Here's my recommendation of tablebases files for a 64 GB SSD

09.08.2005  11:33       794.919.029 kbppkb.0.nbb.emd
09.08.2005  13:17       707.851.030 kbppkb.0.nbw.emd
12.08.2005  00:12       700.194.384 kbppkb.1.nbb.emd
09.08.2005  16:21       716.465.785 kbppkb.1.nbw.emd
09.08.2005  17:52       653.326.314 kbppkb.2.nbb.emd
09.08.2005  21:46       667.369.514 kbppkb.2.nbw.emd
09.08.2005  22:15       203.770.753 kbppkb.3.nbb.emd
10.08.2005  00:17       894.749.438 kbppkn.0.nbb.emd
10.08.2005  02:10       805.454.283 kbppkn.0.nbw.emd
10.08.2005  04:10       859.556.065 kbppkn.1.nbb.emd
10.08.2005  06:01       788.088.523 kbppkn.1.nbw.emd
10.08.2005  08:02       858.733.063 kbppkn.2.nbb.emd
10.08.2005  09:43       713.278.857 kbppkn.2.nbw.emd
10.08.2005  10:07       166.585.128 kbppkn.3.nbb.emd
12.08.2005  06:41       409.022.202 kbppkr.0.nbb.emd
12.08.2005  08:01       367.435.106 kbppkr.0.nbw.emd
12.08.2005  08:26       395.336.141 kbppkr.1.nbb.emd
12.08.2005  09:31       376.393.099 kbppkr.1.nbw.emd
12.08.2005  10:19       385.117.574 kbppkr.2.nbb.emd
12.08.2005  11:16       396.487.077 kbppkr.2.nbw.emd
12.08.2005  11:52       349.416.786 kbppkr.3.nbb.emd
12.08.2005  12:40       396.248.018 kbppkr.3.nbw.emd
14.08.2005  23:02       304.115.210 kbppkr.4.nbb.emd
12.08.2005  13:34       404.021.403 kbppkr.4.nbw.emd
12.08.2005  14:14       300.958.113 kbppkr.5.nbb.emd
12.08.2005  15:00       342.302.299 kbppkr.5.nbw.emd
12.08.2005  15:21       157.025.582 kbppkr.6.nbb.emd
15.08.2005  02:56       854.396.412 knppkn.0.nbb.emd
15.08.2005  04:52       838.654.773 knppkn.0.nbw.emd
15.08.2005  06:35       759.357.078 knppkn.1.nbb.emd
15.08.2005  08:28       835.911.558 knppkn.1.nbw.emd
15.08.2005  10:14       768.145.087 knppkn.2.nbb.emd
15.08.2005  11:45       663.728.181 knppkn.2.nbw.emd
15.08.2005  12:09       151.663.104 knppkn.3.nbb.emd
26.06.2005  20:55       524.960.256 knppkr.0.nbb.emd
26.06.2005  20:54       624.717.281 knppkr.0.nbw.emd
26.06.2005  20:53       542.862.739 knppkr.1.nbb.emd
26.06.2005  20:52       634.375.208 knppkr.1.nbw.emd
26.06.2005  20:50       435.589.159 knppkr.2.nbb.emd
26.06.2005  20:49       531.165.780 knppkr.2.nbw.emd
26.06.2005  20:48       110.606.990 knppkr.3.nbb.emd
25.12.2006  12:32     1.041.614.209 kppkpp.nbw.emd
17.08.2005  05:47       513.909.389 kpppkb.nbb.emd
17.08.2005  06:46       443.300.446 kpppkb.nbw.emd
17.08.2005  08:02       571.554.566 kpppkn.nbb.emd
17.08.2005  09:04       445.255.296 kpppkn.nbw.emd
17.08.2005  10:26       619.243.462 kpppkp.nbb.emd
17.08.2005  11:22       406.608.583 kpppkp.nbw.emd
17.08.2005  15:44       745.886.026 kpppkr.nbb.emd
17.08.2005  16:55       518.947.969 kpppkr.nbw.emd
23.08.2005  16:46       560.832.373 kqppkq.0.nbb.emd
23.08.2005  17:02       472.941.969 kqppkq.0.nbw.emd
24.08.2005  08:14       531.906.904 kqppkq.1.nbb.emd
24.08.2005  08:27       478.863.172 kqppkq.1.nbw.emd
24.08.2005  10:31       516.688.163 kqppkq.2.nbb.emd
24.08.2005  10:37       492.483.741 kqppkq.2.nbw.emd
24.08.2005  12:42       494.148.779 kqppkq.3.nbb.emd
24.08.2005  12:47       486.225.197 kqppkq.3.nbw.emd
24.08.2005  14:48       478.534.276 kqppkq.4.nbb.emd
24.08.2005  14:53       476.362.852 kqppkq.4.nbw.emd
24.08.2005  16:54       484.502.785 kqppkq.5.nbb.emd
24.08.2005  16:01       245.921.895 kqppkq.5.nbw.emd
24.08.2005  16:30        99.401.531 kqppkq.6.nbb.emd
19.01.2007  23:46       574.419.973 krpknp.0.nbb.emd
06.01.2007  00:38       466.210.391 krpknp.0.nbw.emd
09.02.2007  09:08       568.441.596 krpknp.1.nbb.emd
05.01.2007  01:55       473.891.122 krpknp.1.nbw.emd
01.02.2007  10:00       552.666.191 krpknp.2.nbb.emd
12.02.2007  14:48       468.773.389 krpknp.2.nbw.emd
08.02.2007  10:40       547.235.349 krpknp.3.nbb.emd
04.01.2007  20:38       470.564.123 krpknp.3.nbw.emd
12.02.2007  05:31       542.747.044 krpknp.4.nbb.emd
19.01.2007  14:12       482.811.388 krpknp.4.nbw.emd
13.02.2007  02:14       531.810.958 krpknp.5.nbb.emd
06.01.2007  03:55       488.340.798 krpknp.5.nbw.emd
06.02.2007  02:04       531.707.635 krpknp.6.nbb.emd
15.01.2007  04:05       480.655.320 krpknp.6.nbw.emd
19.01.2007  05:52       531.496.904 krpknp.7.nbb.emd
19.01.2007  20:03       479.728.988 krpknp.7.nbw.emd
18.01.2007  00:16       522.362.925 krpknp.8.nbb.emd
28.01.2007  19:48       464.956.570 krpknp.8.nbw.emd
11.02.2007  05:00       529.129.233 krpknp.9.nbb.emd
19.01.2007  03:47       459.984.790 krpknp.9.nbw.emd
05.02.2007  17:13       535.534.475 krpknp.a.nbb.emd
13.01.2007  02:01       456.197.988 krpknp.a.nbw.emd
14.02.2007  11:33       535.673.934 krpknp.b.nbb.emd
19.01.2007  10:42       456.732.980 krpknp.b.nbw.emd
26.12.2006  15:45        31.780.219 krpknp.c.nbb.emd
26.12.2006  15:35       202.821.880 krpknp.c.nbw.emd
07.06.2004  18:47       494.198.577 krpkrp.0.nbw.emd
07.06.2004  18:59       526.651.061 krpkrp.1.nbw.emd
07.06.2004  19:11       580.854.991 krpkrp.2.nbw.emd
07.06.2004  19:23       610.649.837 krpkrp.3.nbw.emd
07.06.2004  19:35       617.928.806 krpkrp.4.nbw.emd
07.06.2004  19:48       626.745.956 krpkrp.5.nbw.emd
07.06.2004  19:51       136.075.525 krpkrp.6.nbw.emd
26.06.2005  21:40       880.049.887 krppkr.0.nbb.emd
26.06.2005  21:39       653.662.668 krppkr.0.nbw.emd
26.06.2005  21:38       862.187.960 krppkr.1.nbb.emd
26.06.2005  21:33       643.882.060 krppkr.1.nbw.emd
26.06.2005  21:29       820.410.149 krppkr.2.nbb.emd
26.06.2005  21:23       581.219.336 krppkr.2.nbw.emd
26.06.2005  21:19       231.093.108 krppkr.3.nbb.emd
             103 File(s) 54.167.774.049 bytes

If you go to 128 GB - it makes sense to add these:

28.12.2006  16:06       405.905.402 kbpkbp.0.nbw.emd
06.01.2007  06:09       426.356.202 kbpkbp.1.nbw.emd
06.01.2007  16:59       457.758.108 kbpkbp.2.nbw.emd
29.12.2006  12:32       480.527.780 kbpkbp.3.nbw.emd
29.12.2006  20:29       481.304.231 kbpkbp.4.nbw.emd
11.01.2007  08:20       480.604.132 kbpkbp.5.nbw.emd
25.12.2006  14:59       128.850.155 kbpkbp.6.nbw.emd
26.12.2006  02:04       313.541.516 kbpknp.0.nbb.emd
26.12.2006  02:08       285.030.634 kbpknp.0.nbw.emd
26.12.2006  02:34       299.807.809 kbpknp.1.nbb.emd
26.12.2006  02:30       277.996.093 kbpknp.1.nbw.emd
26.12.2006  02:56       297.814.206 kbpknp.2.nbb.emd
26.12.2006  03:00       302.346.248 kbpknp.2.nbw.emd
26.12.2006  03:24       326.100.110 kbpknp.3.nbb.emd
26.12.2006  03:28       310.113.303 kbpknp.3.nbw.emd
26.12.2006  03:56       326.921.690 kbpknp.4.nbb.emd
26.12.2006  03:57       324.159.714 kbpknp.4.nbw.emd
26.12.2006  04:29       332.777.992 kbpknp.5.nbb.emd
26.12.2006  04:30       356.960.578 kbpknp.5.nbw.emd
26.12.2006  04:58       365.954.369 kbpknp.6.nbb.emd
26.12.2006  05:02       338.887.465 kbpknp.6.nbw.emd
26.12.2006  05:34       366.846.246 kbpknp.7.nbb.emd
26.12.2006  05:36       372.360.622 kbpknp.7.nbw.emd
26.12.2006  06:03       368.845.625 kbpknp.8.nbb.emd
26.12.2006  06:09       361.565.803 kbpknp.8.nbw.emd
26.12.2006  06:37       381.039.750 kbpknp.9.nbb.emd
26.12.2006  06:39       345.666.486 kbpknp.9.nbw.emd
26.12.2006  07:11       384.883.119 kbpknp.a.nbb.emd
26.12.2006  07:11       360.379.520 kbpknp.a.nbw.emd
26.12.2006  07:44       387.268.335 kbpknp.b.nbb.emd
26.12.2006  07:39       333.711.448 kbpknp.b.nbw.emd
26.12.2006  07:42        25.389.565 kbpknp.c.nbb.emd
26.12.2006  07:55       194.668.920 kbpknp.c.nbw.emd
22.01.2007  00:22       502.019.614 kbpkpp.0.nbb.emd
20.01.2007  03:42       541.090.410 kbpkpp.0.nbw.emd
21.01.2007  19:19       495.068.447 kbpkpp.1.nbb.emd
22.01.2007  09:13       545.109.205 kbpkpp.1.nbw.emd
21.01.2007  10:36       482.362.015 kbpkpp.2.nbb.emd
20.01.2007  02:55       557.648.328 kbpkpp.2.nbw.emd
21.01.2007  12:58       482.739.617 kbpkpp.3.nbb.emd
21.01.2007  00:39       545.262.524 kbpkpp.3.nbw.emd
19.01.2007  02:15       304.308.715 kbpkpp.4.nbb.emd
20.01.2007  17:11       498.964.495 kbpkpp.4.nbw.emd
13.01.2007  07:33       505.618.069 knpknp.0.nbw.emd
11.02.2007  01:00       534.148.854 knpknp.1.nbw.emd
06.01.2007  19:54       578.016.797 knpknp.2.nbw.emd
19.01.2007  09:38       634.607.864 knpknp.3.nbw.emd
11.02.2007  19:23       652.984.994 knpknp.4.nbw.emd
22.01.2007  07:52       652.966.696 knpknp.5.nbw.emd
26.12.2006  20:54        26.844.356 knpknp.6.nbw.emd
26.12.2006  05:32       506.251.319 knpkpp.0.nbb.emd
26.12.2006  05:02       605.711.473 knpkpp.0.nbw.emd
25.12.2006  11:29       491.241.649 knpkpp.1.nbb.emd
28.12.2006  04:25       624.064.318 knpkpp.1.nbw.emd
27.12.2006  23:17       488.972.543 knpkpp.2.nbb.emd
26.12.2006  11:15       631.724.537 knpkpp.2.nbw.emd
29.12.2006  19:15       493.875.110 knpkpp.3.nbb.emd
26.12.2006  10:23       610.725.945 knpkpp.3.nbw.emd
30.12.2006  01:13       314.417.617 knpkpp.4.nbb.emd
29.12.2006  00:51       451.944.304 knpkpp.4.nbw.emd
14.08.2005  17:08       761.379.475 knppkb.0.nbb.emd
14.08.2005  19:05       848.012.034 knppkb.0.nbw.emd
14.08.2005  20:34       647.697.554 knppkb.1.nbb.emd
14.08.2005  22:28       821.921.602 knppkb.1.nbw.emd
16.08.2005  12:06       634.151.856 knppkb.2.nbb.emd
15.08.2005  00:31       672.256.357 knppkb.2.nbw.emd
15.08.2005  01:02       213.242.885 knppkb.3.nbb.emd
23.01.2007  18:39       334.844.676 kqpkqp.0.nbw.emd
22.01.2007  12:48       324.351.418 kqpkqp.1.nbw.emd
22.01.2007  02:46       315.052.379 kqpkqp.2.nbw.emd
23.01.2007  19:17       338.004.832 kqpkqp.3.nbw.emd
23.01.2007  02:58       362.012.523 kqpkqp.4.nbw.emd
23.01.2007  15:53       374.410.174 kqpkqp.5.nbw.emd
23.01.2007  05:02       384.706.946 kqpkqp.6.nbw.emd
22.01.2007  15:20       394.047.828 kqpkqp.7.nbw.emd
21.01.2007  23:16       405.831.106 kqpkqp.8.nbw.emd
20.01.2007  23:00       404.409.194 kqpkqp.9.nbw.emd
23.01.2007  06:17       404.130.030 kqpkqp.a.nbw.emd
22.01.2007  09:45       329.965.083 kqpkqp.b.nbw.emd
15.01.2007  04:48       869.109.727 kqpkrp.0.nbb.emd
26.01.2007  19:56       669.834.365 kqpkrp.0.nbw.emd
16.01.2007  05:28       887.861.171 kqpkrp.1.nbb.emd
03.02.2007  03:29       645.780.988 kqpkrp.1.nbw.emd
17.01.2007  13:10       919.189.256 kqpkrp.2.nbb.emd
17.01.2007  04:47       671.928.279 kqpkrp.2.nbw.emd
21.01.2007  17:52       913.809.323 kqpkrp.3.nbb.emd
25.01.2007  21:12       672.213.367 kqpkrp.3.nbw.emd
19.01.2007  11:43       886.500.764 kqpkrp.4.nbb.emd
04.02.2007  15:53       663.736.298 kqpkrp.4.nbw.emd
03.02.2007  14:31       874.813.773 kqpkrp.5.nbb.emd
03.02.2007  17:49       617.849.425 kqpkrp.5.nbw.emd
17.01.2007  22:49       195.610.294 kqpkrp.6.nbb.emd
13.05.2004  18:44       407.977.939 krbkrp.0.nbb.emd
13.05.2004  18:53       358.256.227 krbkrp.0.nbw.emd
13.05.2004  19:03       433.963.979 krbkrp.1.nbb.emd
13.05.2004  19:13       357.110.011 krbkrp.1.nbw.emd
13.05.2004  19:24       458.385.503 krbkrp.2.nbb.emd
13.05.2004  19:33       367.160.684 krbkrp.2.nbw.emd
13.05.2004  19:44       468.898.043 krbkrp.3.nbb.emd
13.05.2004  19:54       381.641.057 krbkrp.3.nbw.emd
13.05.2004  20:05       469.061.119 krbkrp.4.nbb.emd
13.05.2004  20:15       381.010.947 krbkrp.4.nbw.emd
13.05.2004  20:25       459.282.447 krbkrp.5.nbb.emd
13.05.2004  20:35       380.954.920 krbkrp.5.nbw.emd
13.05.2004  20:45       440.200.677 krbkrp.6.nbb.emd
13.05.2004  20:55       380.181.989 krbkrp.6.nbw.emd
13.05.2004  21:04       372.784.282 krbkrp.7.nbb.emd
13.05.2004  21:14       381.423.098 krbkrp.7.nbw.emd
13.05.2004  21:16        85.320.850 krbkrp.8.nbw.emd
04.01.2007  15:08       537.478.956 krpkbp.0.nbb.emd
03.01.2007  10:51       488.626.022 krpkbp.0.nbw.emd
08.01.2007  16:07       533.713.055 krpkbp.1.nbb.emd
02.01.2007  03:44       496.756.971 krpkbp.1.nbw.emd
02.01.2007  13:57       509.877.752 krpkbp.2.nbb.emd
07.01.2007  23:55       492.819.584 krpkbp.2.nbw.emd
04.01.2007  14:12       502.297.923 krpkbp.3.nbb.emd
07.01.2007  15:59       492.058.481 krpkbp.3.nbw.emd
08.01.2007  15:42       498.030.607 krpkbp.4.nbb.emd
31.12.2006  21:42       505.779.626 krpkbp.4.nbw.emd
31.12.2006  13:21       482.447.636 krpkbp.5.nbb.emd
08.01.2007  16:00       504.999.931 krpkbp.5.nbw.emd
07.01.2007  18:08       491.282.598 krpkbp.6.nbb.emd
07.01.2007  13:09       497.935.905 krpkbp.6.nbw.emd
06.01.2007  22:29       473.802.457 krpkbp.7.nbb.emd
07.01.2007  13:36       490.900.554 krpkbp.7.nbw.emd
06.01.2007  18:57       468.598.244 krpkbp.8.nbb.emd
02.01.2007  02:23       475.210.717 krpkbp.8.nbw.emd
05.01.2007  18:53       473.925.124 krpkbp.9.nbb.emd
06.01.2007  18:56       473.311.546 krpkbp.9.nbw.emd
06.01.2007  20:56       466.333.890 krpkbp.a.nbb.emd
06.01.2007  18:29       464.411.690 krpkbp.a.nbw.emd
04.01.2007  06:28       486.060.058 krpkbp.b.nbb.emd
06.01.2007  16:48       471.924.162 krpkbp.b.nbw.emd
27.12.2006  02:30       243.222.014 krpkbp.c.nbb.emd
27.12.2006  09:37       209.177.439 krpkbp.c.nbw.emd
             135 File(s) 62.234.294.687 bytes

That's the set I'm using for the cluster (+5 man)
Parent - - By hal9000 (**) [no] Date 2010-06-06 21:40
Thanks, I'll take a closer look at your list and compare it with my own notes.

BTW, I've got a 160 GB SSD, but there's only 90 GB free space. I might purchase a 256 GB SSD later.
Parent - - By RFK (Gold) Date 2010-06-06 23:00
In your estimation for a 250gb  is a  120mb/s  MLC fast enough or too slow.
Parent - By hal9000 (**) [no] Date 2010-06-06 23:59 Edited 2010-06-07 00:05
I believe random access performance is more important than raw sequential transfer rates in this case. I'm using the Intel X-25 M G2 which is one of the faster SSDs on the market in terms of random transfers and IO performance. The X-25 also supports the TRIM command, which prevents performance degradation over time.

Just keep in mind that some SSDs suffer from bad firmware. There have been problems with certain drives from Crucial and OCZ. I even think the first generation X-25 had some firmware problems as well. So if you want to purchase the Intel drive, make sure you purchase the G2 version.

As for performance, the SSD is much faster than my old 7200 RPM HDD. Whenever the engine started using the TBs, the search speed dropped substantially (and the HDD would make a lot of noise...). The SSD on the other hand, is silent of course, and although I haven't performed any formal measurements, the search speed doesn't seem to drop much, if at all, when using the TBs on the SSD.

Generally speaking, purchasing a SSD is one of the best computer purchases I've ever made: They're expensive, yes (but will get cheaper), but it's so wonderful being able to boot Windows in under 30 seconds (before it took over 5 minutes!) and start programs in less than 1 second :) Launching Rybka with 6-man TBs from SSD is also much faster than before.

EDIT: I'll do some performance tests later this week and compare the search speeds when using SSDs and HDDs.
Parent - - By adamsfbay (*) [us] Date 2010-06-08 22:46
I'm doing the same thing, albeit with 240Gb of SSD space. Could you recommend additional 6-men that you would add to fill the additional 100 Gb or so?
Parent - - By Lukas Cimiotti (Bronze) [de] Date 2010-06-09 06:31
Not really. It took me some work to put the stuff above together. I looked at the frequency of occurence and kicked out the ones that are unnecessary because the ending is won anyway - like kqrpkp.
Parent - - By adamsfbay (*) [us] Date 2010-06-09 16:22
I appreciate the work you put in, and sharing it with us. Thank you!

In Rybka 4, when you use TB, do you set NalimovUse to "rarely" (default) or "normally" ??? I've currently set it to "rarely" with 256 cache.
Parent - - By Lukas Cimiotti (Bronze) [de] Date 2010-06-09 19:12
I'm still uncertain which setting to use - either Normally or Frequently.
Parent - - By Mark (****) [us] Date 2010-06-10 00:24
I wonder why "rarely" is the default?
Parent - - By Moz (****) Date 2010-06-10 01:02

> I wonder why "rarely" is the default?


Probably because the other settings negatively impact performance in games with the fast time controls.
Parent - By Mark (****) [us] Date 2010-06-10 01:20
Well, that makes sense.  I wonder at what the time control (say for a standard quad core) "Normally" or "Frequently" outperforms "rarely?"
Parent - By Lukas Cimiotti (Bronze) [de] Date 2010-06-10 04:55
Rarely is good for normal harddisks - I only use SSDs.
Parent - By Vasik Rajlich (Silver) [pl] Date 2010-06-10 08:20

> I wonder why "rarely" is the default?


Mainly to be safe, in case some user has a slow hard drive, etc. Those will be the users who are least likely to tinker with settings.

Vas
Parent - - By hal9000 (**) [no] Date 2010-06-11 21:28
In my first post I listed some 6-piece endgames sorted by frequency. Here's the complete Top 100 list (ordered by frequency. When choosing tablebase-files you should focus on the more difficult endgames, and not only the statistical frequency. As mentioned before, many Rybka endgames involve rooks):
  Endgame  Frequency  Percentage
1  krpkrp  1148  11.566
2  krppkr  690  6.951
3  kppkpp  460  4.634
4  kbpknp  299  3.012
5  kbpkpp  289  2.912
6  krpkpp  276  2.781
7  kqppkp  265  2.670
8  kpppkp  252  2.539
9  kbpkbp  238  2.398
10  krbkrp  219  2.206
11  knpkpp  207  2.085
12  krppkp  199  2.005
13  krpkbp  199  2.005
14  kbppkp  189  1.904
15  kqpkqp  186  1.874
16  kqpkpp  186  1.874
17  krnkrp  178  1.793
18  knpknp  156  1.572
19  krbpkr  132  1.330
20  knppkp  131  1.320
21  krpknp  120  1.209
22  kbppkb  116  1.169
23  kqpkrp  108  1.088
24  krnpkr  106  1.068
25  knppkb  103  1.038
26  kpppkr  100  1.007
27  kqrpkr  96  0.967
28  kqppkq  91  0.917
29  kbppkr  91  0.917
30  knppkn  84  0.846
31  kqrkrp  84  0.846
32  kbppkn  83  0.836
33  kqbpkp  80  0.806
34  kqrpkp  79  0.796
35  krppkb  77  0.776
36  kqqpkp  70  0.705
37  kqpkbp  65  0.655
38  krbpkp  59  0.594
39  kqppkr  58  0.584
40  knppkr  58  0.584
41  krbkrn  56  0.564
42  krppkn  55  0.554
43  krppkq  53  0.534
44  kpppkn  47  0.474
45  kpppkb  45  0.453
46  kqqppk  44  0.443
47  krrkrp  44  0.443
48  kqbppk  43  0.433
49  kqpppk  43  0.433
50  kqnpkp  42  0.423
51  kqrppk  40  0.403
52  krpppk  40  0.403
53  kqpknp  39  0.393
54  kbpppk  32  0.322
55  krpkbn  32  0.322
56  krbpkq  31  0.312
57  kpppkq  31  0.312
58  kqppkb  30  0.302
59  krbpkb  30  0.302
60  krbkbp  30  0.302
61  krnpkp  28  0.282
62  krbkrb  28  0.282
63  krrpkr  27  0.272
64  kbnpkp  26  0.262
65  kqbkpp  26  0.262
66  krbkpp  26  0.262
67  kqbkqp  25  0.252
68  kbppkq  23  0.232
69  kqqkpp  22  0.222
70  kqpkrr  22  0.222
71  krnkbp  22  0.222
72  kqqbpk  21  0.212
73  kqrkpp  21  0.212
74  kqnkqp  21  0.212
75  krbppk  20  0.201
76  krnpkq  20  0.201
77  knppkq  20  0.201
78  krnkrn  20  0.201
79  krpkbb  20  0.201
80  kqnppk  19  0.191
81  kqbpkr  19  0.191
82  kbnknp  19  0.191
83  kqqkqp  18  0.181
84  kqrkqp  18  0.181
85  krbknp  18  0.181
86  kbnkbp  18  0.181
87  kbnkpp  18  0.181
88  kqbpkb  17  0.171
89  kqnpkb  17  0.171
90  kbbkbp  17  0.171
91  kqrbkr  16  0.161
92  kqrbkp  16  0.161
93  kqrnkr  16  0.161
94  kqrpkb  16  0.161
95  kbnpkb  16  0.161
96  kqqpkq  15  0.151
97  kbbpkb  15  0.151
98  krnknp  15  0.151
99  krnkpp  15  0.151
100  kqbpkq  14  0.141
Parent - - By Mark (****) [us] Date 2010-06-12 02:11
This is the frequency of 6-man endings actually reached in games.  Has there been any work on which tablebases are accessed the most?  In other words, there may be an important tablebases that are accessed in the search but never show up on the board.

Also, this looks like the list from computer-computer games.  If anyone is interested in the list from human games, I have that.  (Basically the same list, though.)
Parent - - By hal9000 (**) [no] Date 2010-06-12 14:05

> Has there been any work on which tablebases are accessed the most?


None that I'm aware of. Such a list would not be easy to compile, because the engine output only shows you the number of hits, not the frequency distribution.

> Also, this looks like the list from computer-computer games.


Yes, that's correct. I actually mentioned this in my first post :)
Parent - - By Mark (****) [us] Date 2010-06-12 14:27
One of the programmers would have to add some code in his program to count hits on particular tablebases.  This would be purely an academic exercise, since it wouldn't help at all to increase elo.  Maybe a computer science major would like to do this as an interesting project?  (My younger son is in computer science at Carnegie Mellon and has already written a chess program, although it didn't use tablebases, and my older son is going back to WVU for computer science as a second degree.  I'll talk to them about it.)
Parent - By hal9000 (**) [no] Date 2010-06-12 15:00

> One of the programmers would have to add some code in his program to count hits on particular tablebases.


Yes. Perhaps some programs (but none that I'm aware of) do support this using special (undocumented) parameters?

On the other hand, it shouldn't be too difficult for a CS student to modify an open source chess program (or use his own chess program for that matter) so that hit distributions are logged. But since the search algorithms of chess programs aren't (necessarily) identical, a given position would produce (slightly) different distribution data (the search determines which positions are being examined, and consequently it affects the tablebase usage).
Parent - - By cma6 (****) Date 2010-09-16 16:24
Lukas,
   Have you a recommendation as to where and how to get your recommended set of 6-man TBs?
                           Thanks, CMA
Parent - - By Felix Kling (Gold) [de] Date 2010-09-16 16:33
http://kirill-kryukov.com/chess/tablebases-online/

Otherwise you can buy some here http://chessok.com/shop/index.php?main_page=product_info&cPath=7_2&products_id=24

I think Lukas got them by Convekta, but it was a package of all 6 men tablebases just created for him iirc :)
Parent - - By cma6 (****) Date 2010-09-16 16:51
Felix, thanks for the link. It seems that on the kryukove site, one has to install a very complicated piece of software. I think I will buy them from ChessOK. Do you know if the Chess OK set has them listed individually, so that I can choose the ones listed in Lukas' post?
Parent - - By Felix Kling (Gold) [de] Date 2010-09-16 16:59 Edited 2010-09-16 17:02
No, they just have that DVD package I linked to afaik.

The software you need to install is not complicated, you just need Emule and use the download links or even easier use the ftps listed there, like Sesse's server http://tablebase.sesse.net/. There you can just download the files.
Parent - - By cma6 (****) Date 2010-09-16 17:18
Felix,
Thanks for the info. The Sesse site does not work. When one clicks on a link, instead of starting a download, you just get a page with odd symbols.
   They other 6-man TB page looks good but eMule is something I don't have time for.
  I am hopeful that someone who has bought the ChessOK DVD can tell me if I would be able to select individual 6-man TBs from the DVD or from the files copied from the DVD.
                                     Thanks, CMA
Parent - - By Felix Kling (Gold) [de] Date 2010-09-16 17:47
right click and "save as" for the files on sesse's server.

oh, now I understand the question with the DVDs. Sure, you can select the files you want to install. Basically those DVDs just contain the big files you would download. But note that those DVDs never contain all the bases, the complete set requires a huge amount of DVDs :)
Parent - - By cma6 (****) Date 2010-09-16 18:19
Felix:
I will now buy the DVD from ChessOK. Ring that up to another sale for Felix! (Victor, I hope you notice!)
     BTW, the ChessOK website refers to "double-side" or "double-density" DVDs. Do you think that an oridinary DVD drive could read these ChessOK DVDs?
                                 CMA
Parent - By Felix Kling (Gold) [de] Date 2010-09-16 20:07 Edited 2010-09-16 20:15
yes, I think so. this just means that it's the bigger dvd (8,5 GB)
Parent - - By cma6 (****) Date 2010-09-16 19:27
The Sesse site is now working very well for me. (Cancel that ChessOK sale!)
  However, it occurred to me that with the enormous amount of downloads I have started, one day I will come in and find that Time Warner has shut off, or severely limited, my internet use because of "excessive" downloading. Could that be a problem?
Parent - - By Sesse (****) [no] Date 2010-09-17 00:23
I don't know about Time Warner, but _I_ will certainly shut your connection (to me) off if you make several downloads at a time despite the clear warning at the very top of the download page. Actually, there was a system in place for doing this, but since an upgrade it's broken. I'll make sure it works again shortly :-)

/* Steinar */
Parent - - By cma6 (****) Date 2010-09-17 00:47
I don't "make several downloads at a time".
Parent - By Sesse (****) [no] Date 2010-09-17 01:06
Then it's fine, don't worry for my part.

/* Steinar */
Parent - - By Mark (****) [us] Date 2010-09-17 00:42
It sure could be a problem!  I almost lost my Comcast service for a year for excessive downloads.  Never knew they had a limit until I downloaded 500GB in one month.  I got the other 700GB to complete the 6-man files at about 50 or so GB/month.
Parent - - By Fulcrum2000 (****) [nl] Date 2010-09-17 09:11
Just change provider.
Download limits are so pre-2008.
Parent - - By Mark (****) [us] Date 2010-09-17 10:37
Comcast (in the US) finally has a stated maximum download limit of 250 GB/month, which isn't really too bad.  There aren't a lot of options for cable internet service where I live, although last week Verison was laying fiber optic cable in the neighborhood, so I should have another option soon.

What was interesting with Comcast before was that they wouldn't tell you what the maximum download limit was!  They would issue warnings to the top .1% or so of users each month and threaten to cancel service unless downloads were decreased.  No second warnings, and they would not say what the limit was!  They also questioned me on the phone, in a very accusatory manner, about what I was downloading that was 500GB.  Needless to say, the woman that was questioning me wasn't familiar with 6-man tablebases for chess!  Only in America...
Parent - - By cma6 (****) Date 2010-09-17 12:48
Mark, this is very useful information about Comcast. I don't really want to have to deal with the possible loss of internet service because of downloads from Sesse.
My first choice all along was to buy the DVDs from Chess OK, even though I know I won't get all the 6-man TBs. With the DVDs, hopefully I can install the key 6-man TBs as recommended in two lists by Lukas. Then as I approach the endgame in my correspondence games, I can download individual TBs as needed.
I am still trying to determine if the Chess OK "Nalimov Gold" TBs will work with Aquarium. I got an contradictory email from Chess OK, which told me the TBs could be copied from the DVDs and linked with an engine. On the other hand, the email said that "In aquarium you may use only on-line tablebases."
    So I do not yet know if the "Nalimov Gold" TBs will work at all with Aquarium/Rybka.
Parent - By Dadi Jonsson (Silver) [is] Date 2010-09-17 12:51

> I am still trying to determine if the Chess OK "Nalimov Gold" TBs will work with Aquarium.


It doesn't matter where the Nalimov tablebases come from. They are always the same and work with Aquarium.
Parent - By Mark (****) [us] Date 2010-09-17 13:47
Downloading one at a time from Sesse works ok, if you're patient.  That's where I got the last 700GB myself.  Too bad there's not an easier way to share them. 

I actually only use a small subset of the 6-man files anyway because they load so slowly.  I don't even think the full set would load with only 4GB of ram.
Parent - - By Fulcrum2000 (****) [nl] Date 2010-09-17 17:56

> Comcast (in the US) finally has a stated maximum download limit of 250 GB/month, which isn't really too bad.  ... What was interesting with Comcast before was that they wouldn't tell you what the maximum download limit was!  They would issue warnings to the top .1% or so of users each month and threaten to cancel service unless downloads were decreased.  No second warnings, and they would not say what the limit was! 


Yeah, reminds me of the old days. My provider in that time (Chello, Netherlands) had a "fair Use Policy", not a fixed limit. If you would surpass this (soft and unknown) limit your speed was capped to something like 64 KB/s till the end of the month. The next month you would have the full speed again. But even then, I had months with several hundreds GB of download and nothing happened, not even a mail from the provider.

People just have to make a statement and leave a provider using download limits. If enough people are walking away, they will change the rules fast enough :grin:
Parent - - By cma6 (****) Date 2010-09-17 18:07
The Comcast limit is about 10 GB/day (weekdays), so one could download the 6-man TBs (1000 GB?) in 4 months.
Parent - By Mark (****) [us] Date 2010-09-17 18:50
I never heard of a daily limit for Comcast, but they have published the 250GM/month limit.  If you devoted say 200GB/month for tablebases, it'd still take about 6 months to get the full 1.2TB set.  Because of the harsh penalty for exceeding their limit, I wouldn't want to cut it that close, though. I'd suggest trying for 150GB/month for 8 months instead.
Parent - By Mark (****) [us] Date 2010-09-17 18:52
I could live with that penalty a lot easier than cutting off service for a year.  At least you could get your email...
Parent - - By cma6 (****) Date 2010-09-17 15:37
"The most complete collection of Nalimov tablebases on 12 DVDs of double density (100 GB in total)."
I checked with Sony about my DVD drive. Sony told me that my drive, like all Sony DVD drives, will will read dual layer but not double density DVDs. Sony told me that the double density format is rare and that few DVD drives can read it.
                                         CMA
Parent - - By Felix Kling (Gold) [de] Date 2010-09-17 22:43
I think the 12 DVDs are 8,47 GB discs. Any DVD drive will read them. those should be dual layer discs. I guess the information there is not correct.
Parent - - By cma6 (****) Date 2010-09-18 00:54
Thanks, Felix. Has anyone on this board actually bought the Gold Nalimov and can tell me what kind of DVDs they are? I won't be able to return them to ChessOK if they don't work.
                                                       CMA
Parent - - By Felix Kling (Gold) [de] Date 2010-09-18 07:20
just forget about double layer, double density and so on. If your drive can read 8,5 GB DVDs (double layer) there's no problem. You can calculate it by dividing 100 GB by 12 - so it's just a 8,5 GB DVD and not double density or something. Dadi Jonsson should have tested them I assume, he should be able to confirm that.
Parent - By Dadi Jonsson (Silver) [is] Date 2010-09-18 07:57
Yes, that was my experience a few years back. I had no problems reading them.
Parent - - By Shaun (****) [gb] Date 2010-06-06 22:23
I am currently running some TB tests....

Tests run on T8100 laptop, TBs on fast external USB flash drive.

Selected 6 men = 5 men  + krppkr and krpkrp

Time control 40/1

Rank Name                                   Elo    +    - games score oppo. draws
   1 Rybka 4 x64 RC3 1CPU (5men)             27   15   15  1500   63%   -54   39%
   2 Rybka 4 x64 RC3 1CPU (Selected 6men)    26   15   15  1500   62%   -54   38%
   3 Stockfish 1.7.1 JA 64bit 1CPU          -54   10   10  3000   38%    27   38%

   1 Rybka 4 x64 RC3 1CPU (5men)             27 1500.0 (939.0 : 561.0)
   2 Rybka 4 x64 RC3 1CPU (Selected 6men)    26 1500.0 (936.0 : 564.0)


Disapointing, I supose that the difference should be minimal as reaching the end game where the additional 6 men TBs is decisive should be rare. In end game test suites the result may be very different.

Shaun
Parent - By hal9000 (**) [no] Date 2010-06-07 00:53
I'm aware that 6-man tablebases might not help that much in practical play, but it doesn't hurt using them with fast SSDs or flash drives, and it can be handy when playing games with short time controls. I've witnessed a few games in which Rybka was very low on time and would probably have lost on time if it weren't for the 6-man tablebases.

I also use TBs for game analysis (human games), and this has turned out to be useful on several occasions.
Parent - - By Moz (****) Date 2010-06-10 03:17
Have you tested Rybka TB performance with different usage settings?  The default is Rarely but I'm wondering if increasing it to Normal or Frequent would change the test results at all.
Parent - By Shaun (****) [gb] Date 2010-06-10 08:29
on my list - I am currently running a test with 'no' TBs

Shaun
Up Topic Rybka Support & Discussion / Rybka Discussion / Most important 6-man tablebase files for Rybka 4?
1 2 3 Previous Next  

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill