Measuring Hash Collisions (once again)
Moderators: hgm, Rebel, chrisw
-
- Posts: 6997
- Joined: Thu Aug 18, 2011 12:04 pm
- Full name: Ed Schröder
Re: Measuring Hash Collisions (once again)
Yes, good point. On the other hand I was deliberately avoiding zobrist as checksum as the weakness of the algorithm is in the key set and wanted something totally independent. And, if the checksums don't match, that is a collision. The only problem is that it's possible in theory I don't get all the collisions, and for that purpose I will finish the current phase, then move to zobrist as checksum and compare if it reports more collisions.
90% of coding is debugging, the other 10% is writing bugs.
-
- Posts: 27817
- Joined: Fri Mar 10, 2006 10:06 am
- Location: Amsterdam
- Full name: H G Muller
Re: Measuring Hash Collisions (once again)
I would be very surprised if the checksum would not catch more that 99% of all collisions, so you are probably safe enough. Having 1 collision in 64M nodes would completely cripple your program if it caused a crash. But if it just causes that 1 node to have a wrong score, the chances that this wrong score would propagate all the way to the root are minute. The search would just pick anoher move in the same node to avoid the move if it was bad, or another move in the parent to avoid the entire node if it was good.
-
- Posts: 550
- Joined: Tue Nov 19, 2019 8:48 pm
- Full name: Alayan Feh
Re: Measuring Hash Collisions (once again)
A few dozens of TB hits were enough to change the root move on a start position search for Stockfish in a recent experiment by Aloril (both single-core of the same dev version, one without TB use, one with 7-men TB). But for such things to happen, the moves must be ranked extremely close in the first place.
-
- Posts: 6997
- Joined: Thu Aug 18, 2011 12:04 pm
- Full name: Ed Schröder
Re: Measuring Hash Collisions (once again)
Final results PART-TWORebel wrote: ↑Fri Feb 28, 2020 8:44 am PART-TWO
Moving from a 48 bit hash key to a 56 bit hash key and repeat the 4 x 100 runs.
So far:
Code: Select all
KEY = 48 bit KEY = 56 bit POS Time Coll TP Coll TP 100 0:30 20 13 0 0 100 1:00 35 19 100 2:30 87 57 100 5:00 192 144
Code: Select all
KEY = 48 bit KEY = 56 bit
POS Time Coll TP Coll TP
100 0:30 20 13 0 0
100 1:00 35 19 0 0
100 2:30 87 57 0 0
100 5:00 192 144 0 0
100 10:00 --- --- 0 0
90% of coding is debugging, the other 10% is writing bugs.
-
- Posts: 4367
- Joined: Fri Mar 10, 2006 5:23 am
- Location: http://www.arasanchess.org
Re: Measuring Hash Collisions (once again)
Why not 64-bit, since this is more or less standard?Moving from a 48 bit hash key to a 56 bit hash key and repeat the 4 x 100 runs.
-
- Posts: 6997
- Joined: Thu Aug 18, 2011 12:04 pm
- Full name: Ed Schröder
Re: Measuring Hash Collisions (once again)
Perhaps as last step since I have to rewrite my ASM HT code to win a byte.
First I want to know (pure curiosity) how well (or bad) a 48-bit key does, then a 56-bit key.
So far not a single collision seen with a 56-bit key, not even at 5 minutes per move.
Will post the results after the 100 positions at 10 minutes is finished.
90% of coding is debugging, the other 10% is writing bugs.
-
- Posts: 27817
- Joined: Fri Mar 10, 2006 10:06 am
- Location: Amsterdam
- Full name: H G Muller
Re: Measuring Hash Collisions (once again)
With 8 bits extra key you should have 256 times fewer collisions.
But the question is: "what is good enough"? I would say that 1 collision in 64M nodes is already far better than you would need to have totally undetectable impact on Elo. According to Bob 1 collison in 10K nodes should already have no measurable impact, so it seems you are several thousand times better than you need to be, with the 48-bit key. (Which, due to the way you calculate the index in a 128MB table, has 26 significant bits, as the other 22 are redundant.)
Note that I am not convinced by the effect of 7-men EGT in Stockfish. There should not be any lines of chessic relevance from the opening position that reach the 7-men stage, meaning that the scores obtained in these probes should never be able to propagate to the root. It sounds more like an accidental effect of replacing different TT entries because you don't search exactly the same tree, similar to what you would have if you run with a TT of a different size, or using a different set of Zobrist keys. Or that EGT probing slightly slows down the NPS. If root moves are so close in score that the choice between them is basically random, such insignificant things can affect the choice. (And Elo-wise you would not care at all.)
Using longer key will reduce the number of collisions, but it will never make it zero, if you run long enough. If you get one every minute with a 48-bit key, you will get one every 4 hours with a 56-bit key, etc. This would only be of any concern if even a single collision could give you an observable effect, like a program crash (e.g. due to playing an invalid hash move). Crashing once every 4 hours still seems quite unacceptable.
So if your only defense against crashing is the hash key, you need to maintain a much higher standard of collision avoidance. And even then it depends on whether you want to make virtually sure that you would never experience such a crash in a lifetime of your personal use of the program, or whether you have 10,000 users and want to make sure that not a single one of those will ever experience a crash in his lifetime (which he would then post on the forum, to point out, "hey, look how crappy this program is: it crashes reproducibly on this position!").
But the question is: "what is good enough"? I would say that 1 collision in 64M nodes is already far better than you would need to have totally undetectable impact on Elo. According to Bob 1 collison in 10K nodes should already have no measurable impact, so it seems you are several thousand times better than you need to be, with the 48-bit key. (Which, due to the way you calculate the index in a 128MB table, has 26 significant bits, as the other 22 are redundant.)
Note that I am not convinced by the effect of 7-men EGT in Stockfish. There should not be any lines of chessic relevance from the opening position that reach the 7-men stage, meaning that the scores obtained in these probes should never be able to propagate to the root. It sounds more like an accidental effect of replacing different TT entries because you don't search exactly the same tree, similar to what you would have if you run with a TT of a different size, or using a different set of Zobrist keys. Or that EGT probing slightly slows down the NPS. If root moves are so close in score that the choice between them is basically random, such insignificant things can affect the choice. (And Elo-wise you would not care at all.)
Using longer key will reduce the number of collisions, but it will never make it zero, if you run long enough. If you get one every minute with a 48-bit key, you will get one every 4 hours with a 56-bit key, etc. This would only be of any concern if even a single collision could give you an observable effect, like a program crash (e.g. due to playing an invalid hash move). Crashing once every 4 hours still seems quite unacceptable.
So if your only defense against crashing is the hash key, you need to maintain a much higher standard of collision avoidance. And even then it depends on whether you want to make virtually sure that you would never experience such a crash in a lifetime of your personal use of the program, or whether you have 10,000 users and want to make sure that not a single one of those will ever experience a crash in his lifetime (which he would then post on the forum, to point out, "hey, look how crappy this program is: it crashes reproducibly on this position!").
-
- Posts: 690
- Joined: Mon Apr 19, 2010 7:07 pm
- Location: Sweden
- Full name: Peter Osterlund
Re: Measuring Hash Collisions (once again)
I tested this in Texel about two weeks ago. Texel uses 64 bit Zobrist hash keys. A bucket contains 4 entries and a store can write to any one of the 4 entries. If no entry is empty, the least valuable entry is overwritten.hgm wrote: ↑Sat Feb 29, 2020 9:20 am But the question is: "what is good enough"? I would say that 1 collision in 64M nodes is already far better than you would need to have totally undetectable impact on Elo. According to Bob 1 collison in 10K nodes should already have no measurable impact, so it seems you are several thousand times better than you need to be, with the 48-bit key. (Which, due to the way you calculate the index in a 128MB table, has 26 significant bits, as the other 22 are redundant.)
Some of the bits in the Zobrist key are used to determine which bucket a position belongs to. Each entry also stores the whole 64 bit key, even though some of the bits are redundant since they were used to determine the bucket and thus are the same for all positions belonging to that bucket.
I modified the hash probe/insert code so that some of the bits in the Zobrist key are ignored when searching for an entry. The code to determine the bucket was not modified.
I ran some tests with different hash table size, time control and number of verification bits in the hash entries. All tests were using a single core. The result was:
Code: Select all
TT bits TC elo delta games win draw loss
1 4 6+0.06 -942.6 43.9 2054 1 16 2037
1 7 6+0.06 -177.2 6.2 2060 165 762 1133
1 8 6+0.06 -84.9 5.5 1998 274 971 753
1 9 6+0.06 -47.4 5.1 2028 353 1047 628
1 10 6+0.06 -28.0 5.2 2040 414 1048 578
1 12 6+0.06 -13.0 5.2 2030 441 1072 517
1 13 6+0.06 -4.3 2.2 11040 2469 5965 2606
1 16 6+0.06 5.6 5.1 2060 495 1103 462
1 10 24+0.24 -38.7 3.1 5024 748 2971 1305
1 14 24+0.24 -2.4 1.7 13644 2488 8572 2584
1 16 24+0.24 -0.8 2.0 10060 1870 6297 1893
1 10 96+0.96 -52.4 2.9 5052 570 3156 1326
1 14 96+0.96 -6.4 1.9 10018 1539 6756 1723
1 15 96+0.96 -1.2 1.6 13918 2203 9462 2253
1 16 96+0.96 0.1 1.9 10122 1587 6952 1583
4 8 24+0.24 -105.7 5.3 2002 186 1039 777
4 9 24+0.24 -57.7 4.9 2024 267 1157 600
4 10 24+0.24 -35.7 2.8 6002 903 3581 1518
4 13 24+0.24 -8.6 2.0 10014 1759 6249 2006
4 16 24+0.24 0.1 2.0 10362 1906 6552 1904
16 10 96+0.96 -34.1 4.3 2084 262 1356 466
128 8 6+0.06 -18.2 5.3 2048 439 1063 546
"bits" is the number of bits used when comparing hash entries.
"TC" is the time control in seconds, base time + increment.
"elo" is the elo difference compared to unmodified Texel.
"delta" is the estimated standard deviation of the elo.
When a hash probe is made, if an entry exists in the hash table for the probed position, the code is guaranteed to find the correct entry. If an entry does not exist for the probed position, the probability to get a probe result anyway (the collision probability) is 1/2^(bits-2), assuming the hash keys are "random enough".
I ran most tests with a 1 MB hash table to make sure the table is full almost all the time and to get lots of probe misses, to magnify the effect of hash collisions. Texel writes about 16MB/s to the hash table on my test computers when using a single core.
It can be seen that even in the extreme case of 1MB hash table and time control "96+0.96" the effect of hash collisions when using 15 verification bits was too small to measure using about 14000 games.
-
- Posts: 27817
- Joined: Fri Mar 10, 2006 10:06 am
- Location: Amsterdam
- Full name: H G Muller
Re: Measuring Hash Collisions (once again)
That pretty much confirms that 16 (non-redundant) bits in the signature is already good enough from an Elo perspective. And with full verification of the hash move, Elo would be the only concern.
So XOR the signature (or index) with the incremental eval, to make those bits non-redundant!petero2 wrote: ↑Sat Feb 29, 2020 1:22 pmSome of the bits in the Zobrist key are used to determine which bucket a position belongs to. Each entry also stores the whole 64 bit key, even though some of the bits are redundant since they were used to determine the bucket and thus are the same for all positions belonging to that bucket.
-
- Posts: 6997
- Joined: Thu Aug 18, 2011 12:04 pm
- Full name: Ed Schröder
Re: Measuring Hash Collisions (once again)
With Zobrist 8-bit checksum, 48 vs 56 bit hash key.
No single collision with a 56 byte key.
The 10 minutes per move:
No single collision with a 56 byte key.
With games it's different, first the HT is never cleared and secondly the nature of the start position only slightly changes move after move, this contradictory using EPD positions. In some of the 40/2:00 games I am playing I see a collision occur. The game, pos 6 and 7
Code: Select all
KEY = 48 bit KEY = 56 bit
POS Time Coll TP Coll TP
100 0:30 26 17 0 0
100 1:00 41 23 0 0
100 2:30 136 98 0 0
100 5:00 263 211 0 0
100 10:00 438 355 0 0
The 10 minutes per move:
Code: Select all
Pos 1 335.753.226 146.513.477 (43%) 0 0 00:04:44 21.00 0.95 1..Bxe5 2.dxe5 Nd3 3.Qxc2 Nxe1 4.Qe2 g6 5.Qxe1
Pos 2 379.761.209 165.236.968 (43%) 0 0 00:06:52 20.00 0.32 1..Nxf3 2.Rxd8 Rxd8 3.Qxf3 h5 4.Rd1 Rd4 5.Qe3
Pos 3 354.355.462 179.218.307 (50%) 0 0 00:05:32 19.00 0.44 1..Bxa6 2.Qxa6 Rde8 3.a5 Rxe4 4.Qd3 Ree8 5.Bd2
Pos 4 396.025.267 177.082.718 (44%) 0 0 00:07:31 22.00 -0.55 1..Qb3 2.Qc2 Qxc2 3.Nxc2 Rc4 4.Nb4
Pos 5 475.310.830 198.611.473 (41%) 0 0 00:03:53 23.00 0.00 1..Bxe3 2.Rexe3 Rcc8 3.Bg4 a4 4.Rg3
Pos 6 435.763.543 234.728.172 (53%) 0 0 00:09:54 21.00 0.88 1..Rxd1 2.Qxd1 Ka7 3.Qe2 h5
Pos 7 394.939.990 150.234.654 (38%) 0 0 00:03:38 21.00 0.46 1.Qxf5 Bxf5 2.Nc3 Bf8 3.g4 Be6 4.Bb5 Kb7
Pos 8 388.180.455 211.088.380 (54%) 0 0 00:07:39 20.00 0.53 1.Qg6 Rxd1 2.Rxd1 c3 3.bxc3 Rxc3
Pos 9 369.208.766 170.708.055 (46%) 0 0 00:05:11 20.00 0.21 1.Bxb7 Rxb7 2.Rdd2 Rc4 3.Nd5 Nc6 4.Rxc4 bxc4
Pos 10 410.069.015 136.289.021 (33%) 0 0 00:04:08 19.00 1.00 1.Rxc8 Qxc8 2.Nc3 Ng6 3.b4 Ne8
Pos 11 448.228.317 185.623.736 (41%) 0 0 00:04:56 22.00 0.28 1..Rxb2 2.Qxb2 Qa8 3.Nh4
Pos 12 469.430.870 177.120.591 (37%) 0 0 00:03:57 19.00 0.72 1..Nb4 2.Qxd8 Bxd8 3.Nxb4 Rxb4 4.Bc6 Re6 5.Bg2
Pos 13 371.166.156 165.018.896 (44%) 0 0 00:08:09 20.00 1.03 1.Bxd5 exd5 2.Nd2 Nxd2 3.Bxd2
Pos 14 334.790.575 157.207.142 (46%) 0 0 00:08:02 19.00 0.36 1.Bxg7 Kxg7 2.Qh4 f6
Pos 15 415.507.048 172.373.175 (41%) 0 0 00:04:27 20.00 -0.20 1.Rxb8 Rxb8 2.Nd2 h6 3.Nc4 f5 4.Rc1 Rb1
Pos 16 351.660.800 147.201.036 (41%) 0 0 00:05:23 20.00 0.71 1.Bxg5 hxg5 2.Nf3
Pos 17 509.296.055 226.553.336 (44%) 0 0 00:04:38 21.00 0.86 1.h5 Rxe1 2.Rxe1 Qg7 3.Qxg7 Kxg7 4.Ne6
Pos 18 527.238.966 192.376.521 (36%) 0 0 00:05:29 20.00 0.54 1.Rxd4 Nxd4 2.Bd1 Kf8
Pos 19 509.421.438 176.256.785 (34%) 0 0 00:03:07 23.00 0.80 1.Rxb8 Nxb8 2.Re1 Nd7 3.Re2 Bd3 4.Nxc6 Rxc6
Pos 20 378.370.342 188.915.282 (49%) 0 0 00:04:30 20.00 0.65 1..Bxd2 2.Qxd2 Ra8 3.g3 Ra7 4.h4
Pos 21 385.671.591 165.688.547 (42%) 0 0 00:07:25 19.00 0.59 1..Rbd8 2.Nc4 Rg6 3.Kh1 Rdg8
Pos 22 576.474.868 240.742.667 (41%) 0 0 00:08:16 23.00 0.71 1.Rxb4 Bxb4 2.d5 exd5 3.a7 Ra8 4.f4 Bc3 5.Bb6
Pos 23 493.732.234 216.520.585 (43%) 0 0 00:08:39 23.00 0.84 1.Bxc5 Nxc5 2.Rxb8 Bxb8 3.Rb1 Be5 4.Ne7 Ke6
Pos 24 530.123.018 230.696.760 (43%) 0 0 00:08:39 22.00 0.71 1.Rxb8 Bxb8 2.Rb1 Be5 3.Nxe4 Kxe4 4.f3
Pos 25 346.055.697 183.905.412 (53%) 0 0 00:09:36 20.00 0.54 1.Nxf8 Kxf8 2.Rc1 Nxe2 3.Qxe2 Ba6
Pos 1 365.746.424 173.016.816 (47%) 0 0 00:04:54 21.00 0.77 1.Nxc5 bxc5 2.Nxe6 Kh8 3.b3 Qe7 4.f4 Bc7 5.Nxf8
Pos 2 517.539.915 201.783.322 (38%) 0 0 00:05:05 22.00 1.00 1..Qxa1 2.Rxa1 Rd3 3.Re3 Rxb3
Pos 3 344.541.838 116.738.050 (33%) 0 0 00:05:30 21.00 0.96 1.Nxc8 Qxc8 2.b5 axb5 3.Qxb5
Pos 4 418.014.809 152.761.408 (36%) 0 0 00:08:37 23.00 0.00 1..Qxe3 2.fxe3
Pos 5 335.216.254 125.575.156 (37%) 0 0 00:05:22 20.00 0.41 1.Rxe8 Rxe8 2.Rxe8 Bxe8 3.h5
Pos 6 340.117.039 144.231.791 (42%) 0 0 00:09:46 20.00 0.61 1.Bxg7 Kxg7 2.h5 e5 3.dxe6 Bxe6 4.Qb2 Kg8 5.h6
Pos 7 383.016.504 176.050.827 (45%) 0 0 00:06:04 21.00 0.50 1.Rxc2 Rxc2 2.Rd5 Bxg5 3.Bxg5 Qf3 4.Qe3 Qxe3
Pos 8 337.886.645 145.249.777 (42%) 0 0 00:09:00 22.00 0.58 1.Bxc6 Rxc6 2.Rxc6 Qxc6 3.Qxc6
Pos 9 505.576.915 250.804.389 (49%) 0 0 00:04:00 21.00 0.95 1..Rxc1 2.Qxc1 Qh4 3.Qc8 Kh7 4.Qc1 Qh3 5.Qf1
Pos 10 417.889.254 172.170.902 (41%) 0 0 00:04:07 20.00 0.38 1.Rxc8 Qxc8 2.Qc3 Qb8 3.Bc5
Pos 11 401.546.544 173.084.900 (43%) 0 0 00:08:38 22.00 0.36 1..Rxd1 2.Rxd1 Qc7 3.Bb5 Qc2 4.Qxc2 Rxc2 5.Ba1
Pos 12 329.309.216 150.959.805 (45%) 0 0 00:04:15 20.00 0.97 1.Nxe7 Bxe7 2.axb5 Rg6 3.Rg1 fxe4 4.Bxe4 Rh6
Pos 13 420.088.459 195.873.209 (46%) 0 0 00:07:53 21.00 0.63 1.Rxc8 Bxc8 2.Qb8 Qd7
Pos 14 349.996.906 150.055.663 (42%) 0 0 00:06:09 22.00 0.50 1.Bxd5 Rxc1 2.Bxc1 exd5 3.Ne5 Nhf6 4.f4 Bh6
Pos 15 374.545.273 157.827.546 (42%) 0 0 00:05:57 22.00 0.86 1.Rxc8 Qxc8 2.Rc3 Qd8 3.Qc2 f6 4.Bxg5 fxg5
Pos 16 414.561.343 151.735.847 (36%) 0 0 00:06:57 22.00 0.36 1..Qxg4 2.hxg4 Ng6 3.g3 dxc4 4.Rxc4
Pos 17 423.320.802 171.641.777 (40%) 0 0 00:05:38 22.00 0.15 1..Qxd2 2.Nxd2 Rxc1 3.Rxc1 Bb7 4.Bb1 Rxd4 5.Nf1
Pos 18 407.566.815 149.359.262 (36%) 0 0 00:09:36 21.00 0.00 1..Bxc4 2.bxc4 Nd7 3.Be3 Nc5 4.Bg5
Pos 19 365.090.611 194.631.694 (53%) 0 0 00:05:19 19.00 -0.27 1.Rxc7 Bxc7 2.Bd3 Bxg3 3.Bxg5 Kxg5 4.Qe3 Kg4
Pos 20 472.106.867 255.510.432 (54%) 0 0 00:08:12 20.00 0.77 1.Rxc6 bxc6 2.Qd2 Qd5 3.Qd3 Qe6 4.e3 c5
Pos 21 558.050.379 219.242.807 (39%) 0 0 00:06:56 22.00 0.57 1.Rxb6 axb6 2.Rb1 Rb8 3.g4 c4
Pos 22 422.050.223 161.086.835 (38%) 0 0 00:04:26 20.00 0.84 1.Rxd8 Qxd8 2.Qa1 Re6 3.Rc1 Bf8
Pos 23 474.875.791 209.506.298 (44%) 0 0 00:03:54 22.00 0.37 1.Rxd8 Rxd8 2.Ne4 Rd1 3.Kg2 Rb1 4.Bxe5
Pos 24 339.826.638 138.766.293 (40%) 0 0 00:02:25 19.00 0.16 1.Bxb7 Rxb7 2.Ng5 Qb8 3.h5 Rbd7 4.Qg4
Pos 25 463.947.062 176.093.878 (37%) 0 0 00:06:33 23.00 0.38 1..Qxd4 2.Rxd4 Be7 3.Bf3 Rb6 4.Rxd5 Rxb3
Pos 1 517.100.466 175.785.631 (33%) 0 0 00:04:07 23.00 0.75 1..Rxd3 2.cxd3 b5 3.Nb6 fxg3 4.hxg3
Pos 2 462.291.231 190.795.770 (41%) 0 0 00:05:17 21.00 1.01 1.Qxg5 hxg5 2.Nxe5 dxe5 3.Kc2 Bc5 4.Rxc5 Rd1
Pos 3 434.037.624 148.290.561 (34%) 0 0 00:06:37 22.00 -0.29 1.Qxe8 Bxe8 2.Bd1 Nd5 3.Bd2 Bd6
Pos 4 403.351.907 167.403.451 (41%) 0 0 00:04:48 19.00 1.02 1.Bxe6 Qxe6 2.Qe3 Rc8 3.Qb6
Pos 5 416.910.898 198.393.330 (47%) 0 0 00:01:41 21.00 1.11 1..Rxe1 2.Qxe1 Re6 3.Qd2 Bf6 4.a5 g5
Pos 6 517.631.466 215.924.219 (41%) 0 0 00:05:42 23.00 0.59 1..Rxc1 2.Bxc1 Rc8 3.Ra2 Rxc1 4.Rxa4 Re1 5.Kf4
Pos 7 351.214.180 149.953.472 (42%) 0 0 00:04:18 19.00 -0.06 1.Bxc5 Qxc5 2.c4 dxc4 3.Bxc4 Nd4
Pos 8 368.230.830 132.379.801 (35%) 0 0 00:05:37 20.00 0.51 1.Bxf6 Bxf6 2.Bd5 Qe7 3.Rc3 Bg5 4.Rf3 e4 5.Rf7
Pos 9 385.101.389 176.067.598 (45%) 0 0 00:01:27 20.00 0.37 1..Rxd4 2.cxd4 Nf6 3.Be5 Nd5 4.Rf3 f6 5.Bf4
Pos 10 549.144.434 246.251.166 (44%) 0 0 00:04:12 22.00 0.88 1.Rxh5 Bxh5 2.Ne5 Bxf3 3.Nxf3 Nb5 4.Kd3 Kf6
Pos 11 363.787.231 183.595.698 (50%) 0 0 00:06:06 20.00 0.55 1.Rxf8 Rxf8 2.Qg2 b4 3.Nd1 Qd4
Pos 12 430.535.907 207.195.828 (48%) 0 0 00:07:58 21.00 0.93 1.g3 g4 2.Qxe6 Nxe6 3.Rf5 Nxf8 4.Rxf8 Kg7 5.Rf5
Pos 13 528.932.724 213.019.915 (40%) 0 0 00:07:42 22.00 0.53 1..Rxd3 2.Rxd3 Ra6 3.Nd5 Rxa4
Pos 14 351.314.669 134.528.976 (38%) 0 0 00:05:42 21.01 0.58 1.Bxd5 Rxd5 2.Rc4 Bd6 3.Rxc5
Pos 15 406.152.371 193.580.205 (47%) 0 0 00:04:46 21.00 0.69 1.Rxb4 axb4 2.Ne3 Bb5 3.Kh1
Pos 16 411.308.280 200.907.694 (48%) 0 0 00:05:38 22.00 0.69 1..Rxd4 2.Qxd4 Qc7 3.Rd3 Rb4 4.Nxd5 Rxd4 5.Nxc7
Pos 17 367.672.233 163.984.051 (44%) 0 0 00:06:56 20.00 -0.52 1..Rxd3 2.Bxd3 Rd6 3.Rd2 Qa7 4.Bf1 Rxd2 5.Bxd2
Pos 18 504.690.101 170.622.592 (33%) 0 0 00:06:42 21.00 0.05 1.Qxd7 Nxd7 2.a6 Nb6 3.Bd2 Nc8 4.b3 Bd6
Pos 19 498.964.890 225.404.127 (45%) 0 0 00:07:39 20.00 0.75 1.Bxd6 Qxd6
Pos 20 437.448.759 213.261.148 (48%) 0 0 00:05:35 20.00 0.97 1..Bxh4 2.Kxh4 Qd4 3.Re1 Qe5 4.Kh3 Raa8
Pos 21 424.465.175 192.765.505 (45%) 0 0 00:06:12 21.00 1.28 1.Qxe5 dxe5 2.Bxa5 Ra7 3.Bb4 Kf6 4.Rd1 h5
Pos 22 380.463.439 172.902.746 (45%) 0 0 00:07:09 19.00 0.34 1..Bf2 2.Ng2 Bd4
Pos 23 468.508.963 241.384.030 (51%) 0 0 00:08:51 22.00 0.00 1..Bxh2 2.Kxh2 Nf6 3.Bg3
Pos 24 430.253.259 218.118.205 (50%) 0 0 00:05:30 19.00 0.02 1.Rxc7 Nxc7 2.Qc4 Qd7 3.f4 exf4 4.gxf4 Ke7
Pos 25 476.816.121 230.348.655 (48%) 0 0 00:07:59 25.00 0.21 1..Qxf3 2.Kxf3 Bb2 3.Rxb4 cxb4 4.Ra4 Bc3 5.Bxe3
Pos 1 378.662.293 175.876.763 (46%) 0 0 00:08:49 21.00 0.09 1..Nb3 2.Kb1 Nxc1 3.Kxc1 Qxg5 4.Qe3 Rb5 5.Qxg5
Pos 2 389.408.542 129.243.062 (33%) 0 0 00:05:58 21.00 -0.10 1..Rxb2 2.Qxb2 Qc7 3.Nxd4 cxd4 4.Bd2 Qd7 5.h3
Pos 3 441.848.010 202.416.344 (45%) 0 0 00:07:04 21.00 0.94 1.Rxa4 Rxa4 2.Rxa4 Qxa4 3.Qb8
Pos 4 408.008.622 168.733.818 (41%) 0 0 00:04:39 24.00 0.54 1.Rxb8 Qxb8 2.Rb3 Qc7 3.Kg1 Bc2 4.Rb6 Be4 5.Nxe4
Pos 5 429.330.565 170.372.582 (39%) 0 0 00:03:13 21.00 0.37 1..Rxb1 2.Nxb1 Qb7 3.Nfd2 Qb2 4.Qe7 Nxc3 5.Qxd7
Pos 6 471.643.922 202.513.091 (42%) 0 0 00:04:45 19.00 0.58 1.Bxc8 Nxc8 2.Kc2 Be7 3.f3 Rg6 4.Rha1 f5 5.Kd3
Pos 7 344.136.403 133.331.484 (38%) 0 0 00:05:21 20.00 1.27 1.Nxf6 Qxf6 2.Be4 Rb8 3.Ba7
Pos 8 339.029.227 149.015.998 (43%) 0 0 00:09:11 20.00 0.00 1.Rxe6 fxe6 2.c5 e5 3.Ra1 Nab6 4.Rd1 Na4 5.Ra1
Pos 9 354.258.822 160.654.573 (45%) 0 0 00:05:52 22.00 0.50 1..Qxd4 2.Rxd4 Rc6 3.Rd2 Rac8 4.Nd1 Rxc2 5.Rxd6
Pos 10 360.432.054 156.752.504 (43%) 0 0 00:06:19 20.00 1.29 1.Qc4 Rxe1 2.Rxe1 Qd7 3.Rc1 Ne3 4.Qxc7 Nexd5
Pos 11 381.920.064 149.411.603 (39%) 0 0 00:07:59 21.00 0.16 1..Bxd5 2.Rxd5 Nf8 3.Rxd8
Pos 12 441.175.233 237.319.446 (53%) 0 0 00:03:37 18.00 0.35 1..Rxd3 2.Rxd3 Re8 3.h3 Qe4 4.Qd2 Kg7 5.Qc3
Pos 13 345.295.502 135.809.237 (39%) 0 0 00:09:53 19.02 0.75 1..Bxf3 2.Qxf3 Qd6 3.Rd1 Ra5 4.c5 bxc5 5.dxc5
Pos 14 406.161.267 171.640.000 (42%) 0 0 00:06:40 21.00 0.39 1..Qxg3 2.Rxg3 b4 3.Nb5 Bh6 4.Kb1 Bf4
Pos 15 392.564.375 183.438.505 (46%) 0 0 00:04:55 21.00 -0.34 1..Bxh3 2.Qxh3 Rac8 3.Rhf1 Qe5 4.Qd3 Rc4
Pos 16 358.695.361 144.637.223 (40%) 0 0 00:09:02 21.00 0.66 1..Rxe1 2.Qxe1 Bxf3 3.gxf3 Qxd4 4.Qe4 Rd8
Pos 17 331.665.817 117.723.775 (35%) 0 0 00:01:31 20.00 0.41 1..Nxg5 2.Qxg5 Be7 3.Qe3 Rad8 4.Nf3 Nc5 5.bxc5
Pos 18 349.120.661 133.571.886 (38%) 0 0 00:06:44 19.00 0.45 1..Bxd5 2.exd5 g6 3.Rd1 Bg5 4.Rff1 Rae8 5.Kh1
Pos 19 343.732.407 134.159.151 (39%) 0 0 00:09:07 19.00 0.73 1..Bxe4 2.Nxe4
Pos 20 360.792.212 161.588.053 (44%) 0 0 00:03:47 20.00 0.50 1.Bxc5 dxc5 2.fxe5 Bxc1 3.Qxc1
Pos 21 372.617.915 199.583.836 (53%) 0 0 00:09:49 20.00 0.71 1.Bxg4 fxg4 2.Nd1 Rxf1 3.Rxf1 Qxe4 4.Qxb7 Rc8
Pos 22 509.285.870 285.440.351 (56%) 0 0 00:09:19 19.00 -0.20 1..Rxf2 2.Qxf2 hxg5
Pos 23 357.595.842 142.283.377 (39%) 0 0 00:07:40 20.00 -0.07 1..Nxe4 2.Qxe4 Rb3 3.Rd2 Qb5 4.Rh3 Re8 5.Bb4
Pos 24 375.835.832 186.640.088 (49%) 0 0 00:07:44 20.00 1.49 1..Kc8 2.Qxf4 Rxe7 3.Rxe7 Bxe7 4.h5 Qa1
Pos 25 476.996.117 220.554.597 (46%) 0 0 00:09:56 21.02 0.76 1.Kg3 Ra4 2.Rd2 Rxa6
With games it's different, first the HT is never cleared and secondly the nature of the start position only slightly changes move after move, this contradictory using EPD positions. In some of the 40/2:00 games I am playing I see a collision occur. The game, pos 6 and 7
Code: Select all
Pos 1 105.142.242 37.713.107 (35%) 0 0 0 0 00:02:21 18.00 -0.00 1..g5 2.Bg3 c5 3.f3 cxd4
Pos 2 84.466.202 34.687.002 (41%) 0 0 0 0 00:02:27 18.00 0.27 2.Bg3 c5 3.dxc5 Ne4 4.Qc2
Pos 3 101.107.778 39.429.095 (38%) 0 0 0 0 00:02:11 18.00 -0.13 2..c5 3.dxc5 Ne4 4.Qc2 bxc5 5.f3 Nxg3 6.hxg3
Pos 4 112.798.936 47.228.775 (41%) 0 0 0 0 00:01:16 17.08 0.09 3.Rd1 Ne4 4.Qc2 cxd4 5.Rxd4 d5 6.e3 f5 7.cxd5
Pos 5 58.945.119 25.252.328 (42%) 0 0 0 0 00:01:34 17.01 0.00 3..Ne4 4.Qd3 b5 5.dxc5 Na6 6.cxb5 Qa5 7.b4
Pos 6 55.527.648 24.343.630 (43%) 1 0 1 0 00:01:41 17.00 0.15 4.Qc2 cxd4 5.Rxd4 f5
Pos 7 119.408.690 54.179.797 (45%) 1 0 2 0 00:02:40 18.00 0.11 4..cxd4 5.Rxd4 f5 6.f3 Nc6 7.Rd1 Nxg3
Pos 8 20.092.497 9.261.524 (46%) 0 0 2 0 00:00:34 16.00 0.09 5.Rxd4 d5 6.f3 Qf6 7.Rd1 Nxg3 8.hxg3 Nc6 9.cxd5
Pos 9 118.194.459 53.802.132 (45%) 0 0 2 0 00:03:03 18.00 0.09 5..f5 6.f3 Nc6 7.Rd1 Nxg3 8.hxg3 Rc8 9.e4 Ne5
Pos 10 76.762.479 35.628.401 (46%) 0 0 2 0 00:02:11 18.00 -0.06 6.f3 Nc6 7.Rd1 Nxg3 8.hxg3 d5 9.cxd5 exd5
Pos 11 70.668.462 32.968.157 (46%) 0 0 2 0 00:02:09 17.03 0.31 6..Nc6 7.Rd1 Nxg3 8.hxg3
Pos 12 171.000.899 84.374.603 (49%) 0 0 2 0 00:04:17 19.00 0.00 7.Rd1 Nxg3 8.hxg3 d5 9.f4 g4 10.cxd5 exd5
Pos 13 81.351.228 40.623.516 (49%) 0 0 2 0 00:02:22 18.00 0.09 7..Nxg3 8.hxg3 d5 9.cxd5 exd5 10.Qd2
Pos 14 12.700.738 6.484.122 (51%) 0 0 2 0 00:00:23 15.00 0.00 8.hxg3 d5 9.f4 g4 10.cxd5 exd5 11.Nf2 Qf6 12.e4
Pos 15 66.542.687 34.132.262 (51%) 0 0 2 0 00:02:04 16.00 0.23 8..d5 9.f4 g4 10.cxd5 exd5 11.Nf2 Qf6 12.e4
Pos 16 90.695.619 46.043.000 (50%) 0 0 2 0 00:02:26 18.00 0.00 9.f4 g4 10.cxd5 exd5 11.Nf2
Pos 17 104.596.327 52.453.875 (50%) 0 0 2 0 00:02:59 18.02 0.09 9..Na5 10.fxg5 Rc8 11.gxh6 Nxc4 12.Nf4 Qf6
Pos 18 67.523.989 35.896.176 (53%) 0 0 2 0 00:01:22 18.00 0.00 10.fxg5 Nxc4 11.Qc1 Rc8 12.b3 Ne5 13.Qa1 Ng4
Pos 19 81.406.068 40.982.679 (50%) 0 0 2 0 00:02:06 18.00 0.00 10..Nxc4 11.Qc1 Rc8 12.b3 Ne5 13.Qf4 Qc7 14.gxh6
Pos 20 120.792.676 64.558.896 (53%) 0 0 2 0 00:02:12 19.00 -0.22 11.Qc3 hxg5 12.Nxg5 Qxg5 13.Rh8
Pos 21 129.755.527 69.287.671 (53%) 0 0 2 0 00:02:31 19.00 0.05 11..hxg5 12.Nxg5
Pos 22 99.448.743 54.838.154 (55%) 0 0 2 0 00:02:08 19.00 0.00 12.Nxg5 Qxg5 13.Rh8 Kf7 14.Rh7 Ke8 15.Rxb7
Pos 23 10.445.250 5.940.089 (56%) 0 0 2 0 00:00:14 17.00 0.04 12..Qxg5 13.Rh8 Kf7 14.Rh7 Ke8 15.Rxb7 Qf6
Pos 24 188.544.657 106.656.922 (56%) 0 0 2 0 00:03:30 21.00 -0.79 13.Rh8 Kf7 14.Rh7 Ke8 15.Rxb7 Qf6 16.Rd4 Rf7
Pos 25 10.337 6.281 (60%) 0 0 2 0 00:00:00 7.00 -0.31 13..Kf7 14.Rh7 Ke8 15.Rxb7 Rf7 16.Qh8 Rf8
Pos 26 97.376.011 54.814.258 (56%) 0 0 2 0 00:02:12 21.00 -0.78 14.Rh7 Ke8 15.Rxb7 Qf6 16.Rd4 Rf7 17.Rxf7
Pos 27 274.107.271 157.971.529 (57%) 0 0 2 0 00:06:38 21.00 0.86 14..Ke8 15.Rxb7 Qf6
Pos 28 26.016.982 15.510.024 (59%) 0 0 2 0 00:00:33 19.00 0.00 15.Rxb7 Rf7 16.Qh8 Rf8 17.Qc3
Pos 29 83.972.652 45.276.701 (53%) 0 0 2 0 00:01:12 20.00 0.00 15..Rf7 16.Qh8 Rf8 17.Qc3 Rd3
Pos 30 197.557.833 109.993.818 (55%) 0 0 2 0 00:02:51 22.00 0.00 16.Qh8 Rf8 17.Qc3
Pos 31 2.004 1.288 (64%) 0 0 2 0 00:00:00 7.00 0.00 16..Rf8 17.Qc3
Pos 32 90.857.066 50.503.853 (55%) 0 0 2 0 00:00:52 22.00 0.00 17.Qc3
Pos 33 94.890.964 51.039.796 (53%) 0 0 2 0 00:01:43 19.00 0.02 17..Qf6 18.Rd4 Rf7 19.Rxf7 Kxf7 20.e4 fxe4
Pos 34 140.741.792 76.617.463 (54%) 0 0 2 0 00:02:02 20.00 -0.00 18.Rd4 Rf7 19.Rxf7
Pos 35 128.900.442 71.006.037 (55%) 0 0 2 0 00:02:36 20.00 0.09 18..Rf7 19.Rxf7 Kxf7 20.e3 Nd6 21.Qc7
Pos 36 92.965.930 54.056.445 (58%) 0 0 2 0 00:01:55 20.00 -0.17 19.Rxf7 Kxf7 20.e4 fxe4 21.Bxc4 dxc4 22.Qxc4
Pos 37 34.153.874 19.267.325 (56%) 0 0 2 0 00:00:47 18.00 0.14 19..Kxf7 20.e3 Nd6 21.Qc7 Qe7 22.Qxe7 Kxe7
Pos 38 166.023.508 96.773.393 (58%) 0 0 2 0 00:02:13 20.00 -0.23 20.e4 fxe4 21.Bxc4 Rc8 22.a4 Kg7 23.Ke2 Rxc4
Pos 39 208.367.056 125.067.082 (60%) 0 0 2 0 00:03:30 20.00 0.64 20..fxe4 21.Bxc4 Rc8 22.a4 Rc5 23.b3 dxc4 24.bxc4
Pos 40 162.052.827 101.135.237 (62%) 0 0 2 0 00:02:37 20.00 -0.11 21.Bxc4 Rc8 22.a4 Rxc4 23.Rxc4 dxc4 24.Qxc4
Pos 41 119.300.808 74.678.832 (62%) 0 0 2 0 00:01:27 19.00 0.05 21..Rc8 22.a4 Rc5 23.Qe3
Pos 42 96.503.250 61.343.343 (63%) 0 0 2 0 00:01:50 20.00 0.00 22.a4 Rc5 23.b3 dxc4 24.bxc4 Ke8 25.Qd2 Qf2
Pos 43 274.007.383 170.249.845 (62%) 0 0 2 0 00:05:41 20.00 0.00 22..Rc5 23.Qe3
Pos 44 99.987.042 64.069.812 (64%) 0 0 2 0 00:02:09 19.00 0.00 23.b3 dxc4 24.bxc4 Kg8
Pos 45 77.943.574 49.788.456 (63%) 0 0 2 0 00:01:35 18.00 0.44 23..dxc4 24.bxc4 Ra5 25.Rd7 Kg6 26.Qc2 Qa1
Pos 46 119.036.018 75.743.022 (63%) 0 0 2 0 00:02:04 19.00 -0.43 24.bxc4 Ra5 25.Rd7 Kg6 26.Qc2 Qa1 27.Rd1 Qe5
Pos 47 118.194.250 75.609.867 (63%) 0 0 2 0 00:01:57 19.00 0.26 24..Ra5 25.Ke2 Rxa4 26.Rd7 Kg6 27.Qc2 Ra3 28.Qxe4
Pos 48 165.448.898 108.340.438 (65%) 0 0 2 0 00:01:47 19.00 -0.48 25.Ke2 Rxa4 26.Rd7 Kg6 27.Qc2 Ra3 28.Qxe4
Pos 49 350.013.801 216.093.977 (61%) 0 0 2 0 00:07:27 19.04 0.38 25..Kg6 26.Qd2 Qf5 27.Qc2 Re5
Pos 50 154.311.786 95.752.815 (62%) 0 0 2 0 00:02:36 19.00 -0.58 26.Qe3 Rxa4 27.Qxe4 Qf5 28.Qxf5
Pos 51 152.203.434 93.445.372 (61%) 0 0 2 0 00:02:40 19.00 0.58 26..Rxa4 27.Qxe4 Qf5 28.Qg4 Qxg4 29.Rxg4 Kf5
Pos 52 126.501.668 76.099.411 (60%) 0 0 2 0 00:01:41 19.00 -0.68 27.Qxe4 Qf5 28.Qxf5 Kxf5 29.Rf4 Ke5 30.Rh4
Pos 53 241.206.277 142.350.978 (59%) 0 0 2 0 00:03:47 19.00 0.64 27..Qf5 28.g4 Ra2 29.Ke3 Qxe4 30.Rxe4
Pos 54 210.072.759 115.736.512 (55%) 0 0 2 0 00:02:46 19.00 -0.75 28.Qxf5 Kxf5 29.Kf3 Ra3 30.Kf2 a5 31.g4 Ke5
Pos 55 356.573.302 180.730.146 (50%) 0 0 2 0 00:05:28 21.00 1.01 28..Kxf5 29.g4 Ke5 30.Rd7 Rxc4
Pos 56 284.474.656 146.972.222 (51%) 0 0 2 0 00:04:02 21.00 -0.95 29.g4 Ke5 30.Rd7
Pos 57 217.494.416 113.103.679 (52%) 0 0 2 0 00:01:04 20.00 1.07 29..Ke5 30.Rd7 Rxc4 31.Rxa7 Rxg4 32.Kf3 Rf4
Pos 58 202.135.778 110.713.334 (54%) 0 0 2 0 00:02:54 21.00 -1.00 30.Rd7 Rxc4 31.Rxa7 Rxg4 32.Kf3
Pos 59 126.498.671 69.626.246 (55%) 0 0 2 0 00:01:11 20.00 1.08 30..Rxc4 31.Rxa7 Rxg4 32.Kf3 Rf4 33.Kg3 b5
Pos 60 287.992.404 189.362.157 (65%) 0 0 2 0 00:03:22 22.00 -1.01 31.Rxa7 Rxg4 32.Kf3 Rg5 33.Rb7
Pos 61 159.890.939 110.797.753 (69%) 0 0 2 0 00:01:52 21.00 1.10 31..Rxg4 32.Kf3 Rf4 33.Ke3 b5 34.Rb7 Rb4 35.Rb6
Pos 62 286.152.327 202.067.382 (70%) 0 0 2 0 00:03:12 21.00 -0.84 32.Kf3 Rf4 33.Ke3 b5 34.Rb7 Rd4
Pos 63 183.119.164 125.516.840 (68%) 0 0 2 0 00:02:09 20.00 1.21 32..Rf4 33.Kg3 Rb4 34.Kf3 Rd4 35.Ra3
Pos 64 340.699.380 230.189.702 (67%) 0 0 2 0 00:04:09 20.01 -1.03 33.Kg3 Rb4 34.Rf7 b5 35.Kf3 Rb3
Pos 65 249.694.462 161.946.085 (64%) 0 0 2 0 00:03:05 20.05 1.08 33..b5 34.Rb7 Rb4 35.Kf3 Rb3 36.Kg4 Kd4 37.Kh4
Pos 66 277.927.492 170.485.325 (61%) 0 0 2 0 00:03:04 21.00 -1.29 34.Rb7 Rb4 35.Kf3 Rf4
Pos 67 169.648.534 108.218.106 (63%) 0 0 2 0 00:02:05 20.00 1.21 34..Rb4 35.Kh3 Kf4 36.g3 Kg5 37.Ra7 Rd4 38.Rg7
Pos 68 305.384.854 189.020.616 (61%) 0 0 2 0 00:03:16 20.00 -1.15 35.Kh3 Rb3 36.Kg4 Rb4 37.Kf3 Rg4 38.Kxg4
Pos 69 304.427.812 188.891.051 (62%) 0 0 2 0 00:03:40 21.00 1.32 35..Rb1 36.g4 Kf4 37.Rf7 Kg5 38.Rg7 Kf6 39.Rb7
Pos 70 225.051.462 141.724.164 (62%) 0 0 2 0 00:02:17 21.00 -1.12 36.g4 Ke4 37.Kh4 Rh1 38.Kg3 Ra1 39.Rxb5 Rg1
Pos 71 250.291.912 160.475.783 (64%) 0 0 2 0 00:02:03 20.01 1.15 36..b4
Pos 72 179.901.333 115.398.880 (64%) 0 0 2 0 00:01:37 21.00 -1.20 37.Kg2 Kf4
Pos 73 277.253.801 183.014.433 (66%) 0 0 2 0 00:01:58 22.00 0.96 37..Kf4 38.Re7 Re1 39.Rb7 Re5 40.Kh3 Rc5 41.Kh4
Pos 74 366.624.090 246.042.189 (67%) 0 0 2 0 00:03:57 23.00 -1.39 38.Re7 e5 39.g5 Kxg5
Pos 75 263.872.608 187.293.664 (70%) 0 0 2 0 00:02:30 22.00 1.51 38..e5 39.g5 Rb2 40.Kf1 Kxg5 41.Rxe5 Kf4 42.Rh5
Pos 76 224.924.186 166.715.589 (74%) 0 0 2 0 00:01:55 22.00 -1.30 39.g5 Kxg5
Pos 77 277.371.573 233.811.881 (84%) 0 0 2 0 00:02:29 22.00 1.73 39..Kxg5 40.Rxe5 Kf4
Pos 78 216.646.037 195.362.966 (90%) 0 0 2 0 00:02:00 21.00 -1.64 40.Rxe5 Kf4 41.Re7 b4 42.Rh7 Rb2 43.Kh3 Rb3
Pos 79 239.370.305 220.683.919 (92%) 0 0 2 0 00:02:07 20.00 1.77 40..Kf4 41.Rd5 Rb2
Pos 80 248.638.021 229.006.406 (92%) 0 0 2 0 00:01:50 20.01 -1.72 41.Re7 b4
Pos 81 740.927.998 681.240.045 (91%) 0 0 2 0 00:06:29 22.00 2.21 41..b4 42.Rb7 Rb2 43.Kh3 b3 44.Rb4 Ke3 45.Kg3
Pos 82 379.400.107 356.408.306 (93%) 0 0 2 0 00:01:31 21.00 -2.41 42.Kf2 b3 43.Rb7 Rb2 44.Kg1
Pos 83 274.511.793 256.360.447 (93%) 0 0 2 0 00:01:48 23.00 1.50 42..b3 43.Rb7 Ke4 44.Rb4 Kd5 45.Kg3 Kc5 46.Ra4
Pos 84 541.406.249 515.558.573 (95%) 0 0 2 0 00:03:43 24.00 -2.50 43.Rb7 Rb2 44.Ke1 Ke4 45.Kd1 Kd4 46.Kc1 Rc2
Pos 85 351.792.498 338.435.781 (96%) 0 0 2 0 00:02:46 24.00 2.64 43..Rb2 44.Ke1 Ke4 45.Kd1 Rb1 46.Kd2 Rb2 47.Kc3
Pos 86 311.978.135 299.202.065 (95%) 0 0 2 0 00:02:17 23.00 -2.33 44.Ke1 Ke3
Pos 87 530.922.178 507.232.762 (95%) 0 0 2 0 00:04:06 25.00 1.53 44..Ke3 45.Re7 Kd4 46.Rd7 Kc3 47.Rc7
Pos 88 323.151.333 308.211.926 (95%) 0 0 2 0 00:02:25 25.00 -1.39 45.Re7 Kd4 46.Rd7 Ke5 47.Rd3 Kf6
Pos 89 359.822.827 341.685.862 (94%) 0 0 2 0 00:02:33 24.00 1.53 45..Kd3 46.Rd7 Kc4 47.Rc7 Kd4 48.Rd7 Kc5 49.Rc7
Pos 90 430.955.130 413.942.455 (96%) 0 0 2 0 00:03:14 25.00 -2.64 46.Rd7 Kc2 47.Rc7 Kb1 48.Rb7
Pos 91 212.550.848 205.045.707 (96%) 0 0 2 0 00:01:14 28.00 2.51 46..Kc2 47.Rc7 Kb1
Pos 92 727.580.246 703.804.746 (96%) 0 0 2 0 00:04:38 27.00 -3.15 47.Rc7 Kb1 48.Rb7
Pos 93 550.374.832 524.127.824 (95%) 0 0 2 0 00:00:05 25.00 2.27 47..Kb1 48.Rb7 Ka2 49.Ra7 Kb1 50.Rb7 Ka2 51.Ra7
Pos 94 232.977.327 222.712.962 (95%) 0 0 2 0 00:01:52 23.00 -2.58 48.Rb7 Kc1 49.Rb8 Kc2 50.Rc8 Kb1 51.Rb8 Ka2
Pos 95 329.704.629 322.775.143 (97%) 0 0 2 0 00:02:32 24.00 3.21 48..Kc1 49.Rb6 Rb1
Pos 96 334.116.712 326.561.262 (97%) 0 0 2 0 00:01:45 23.00 -3.09 49.Rb8 Rb1 50.Ke2 Kc2
Pos 97 242.383.740 237.583.015 (98%) 0 0 2 0 00:01:50 23.00 3.23 49..Rb1 50.Ke2
Pos 98 315.400.526 307.509.696 (97%) 0 0 2 0 00:01:41 24.00 -3.11 50.Ke2 Kc2
Pos 99 283.875.731 277.013.928 (97%) 0 0 2 0 00:02:15 23.00 3.21 50..Kb2 51.Kd2 Rh1
Pos 100 299.010.865 291.709.898 (97%) 0 0 2 0 00:01:30 22.00 -3.09 51.Kd2 Ra1 52.Rb5 Rh1 53.Rf5 Ka3
Pos 101 334.158.473 323.478.463 (96%) 0 0 2 0 00:02:42 23.05 3.23 51..Rg1
Pos 102 480.439.753 461.490.684 (96%) 0 0 2 0 00:03:04 23.01 -3.25 52.Kd3 Ka2 53.Kc3 Rc1 54.Kb4 Ra1 55.Ra8 Kb1
Pos 103 833.979.417 805.556.716 (96%) 0 0 2 0 00:01:41 23.01 3.81 52..Ra7 53.Rb6 Ka3 54.Kc3 Rc7 55.Kd2 Ka2 56.Rb8
Pos 104 538.224.717 516.801.167 (96%) 0 0 2 0 00:04:25 24.00 -5.45 53.Kd2 Ra2 54.Kd1 Ra7 55.Kd2 Rd7 56.Ke2 Re7
Pos 105 466.136.794 440.428.737 (94%) 0 0 2 0 00:01:23 23.00 5.81 53..Ra3 54.Rb4
Pos 106 523.316.054 497.157.767 (95%) 0 0 2 0 00:00:22 23.00 -5.63 54.Rb4 Kb1 55.Rb8 Ra2
Pos 107 550.085.739 527.539.724 (95%) 0 0 2 0 00:01:18 25.00 5.76 54..Kb1 55.Rb6 b2 56.Kd1 Rd3 57.Ke2 Kc2 58.Rxb2
Pos 108 694.112.743 680.803.041 (98%) 0 0 2 0 00:00:23 23.00 -5.62 55.Re4 b2
Pos 109 526.797.541 504.684.973 (95%) 0 0 2 0 00:02:36 23.02 5.85 55..Ra7 56.Rb4
Pos 110 359.722.284 334.955.713 (93%) 0 0 2 0 00:02:10 25.01 -5.63 56.Kc3
Pos 111 544.390.874 509.936.572 (93%) 0 0 2 0 00:00:39 21.00 5.81 56..b2 57.Kc3 Ra3 58.Kd4
Pos 112 811.555.876 784.436.049 (96%) 0 0 2 0 00:00:46 23.00 -5.65 57.Kc3 Ka1 58.Rxb2 Rc7 59.Kb3 Rb7 60.Kc3 Rxb2
Pos 113 574.136.494 560.660.779 (97%) 0 0 2 0 00:02:25 23.00 6.38 57..Kc1 58.Rxb2 Rc7 59.Kb3 Rb7 60.Kc4 Kxb2
Pos 114 421.824.255 412.618.653 (97%) 0 0 2 0 00:01:44 21.00 -5.72 58.Rxb2 Rc7 59.Kb3 Rb7 60.Kc4 Rxb2 61.Kc5
Pos 115 578.629.581 566.415.289 (97%) 0 0 2 0 00:02:24 22.00 M 17 58..Rc7 59.Kb3 Rb7 60.Kc3 Rxb2 61.Kc4 Rc2 62.Kd5
Pos 116 585.047.316 573.023.550 (97%) 0 0 2 0 00:02:24 20.00 -5.67 59.Kb3 Rb7 60.Kc3 Rxb2 61.Kc4 Rd2 62.Kc5 Kb2
Pos 117 710.030.580 695.419.430 (97%) 0 0 2 0 00:02:45 20.00 5.88 59..Rb7 60.Kc3 Rxb2 61.Kd3
Pos 118 184.686.345 180.977.930 (97%) 0 0 2 0 00:00:47 20.00 -M15 60.Kc3 Rxb2 61.Kc4 Rc2 62.Kd5 Kd1 63.Ke5 Rc5
Pos 119 519.904.048 508.575.566 (97%) 0 0 2 0 00:02:16 21.00 M 13 60..Rxb2 61.Kc4 Kc2 62.Kd5 Rb5 63.Ke4 Kd2 64.Kd4
Pos 120 197.516.274 192.682.611 (97%) 0 0 2 0 00:00:59 21.00 -M15 61.Kd4 Kd2 62.Ke5 Ke3 63.Ke6 Rb6 64.Kf7 Rb7
Pos 121 94.516.016 92.303.761 (97%) 0 0 2 0 00:00:27 19.01 M 14 61..Kd2 62.Ke5 Rb5 63.Ke4 Rb4 64.Ke5 Ke3 65.Kd5
Pos 122 44.490.726 43.511.986 (97%) 0 0 2 0 00:00:13 19.00 -M13 62.Ke4 Kc3 63.Kf5 Rf2 64.Ke4 Rb2 65.Kf5 Rf2
Pos 123 88.710.878 86.529.144 (97%) 0 0 2 0 00:00:28 19.01 M 13 62..Rb4 63.Kd5 Kd3 64.Kc6 Ra4 65.Kd5
Pos 124 53.120.828 51.866.203 (97%) 0 0 2 0 00:00:16 19.00 -M13 63.Kd5 Kd3 64.Ke6 Re4 65.Kd5 Rb4 66.Ke6 Re4
Pos 125 43.995.240 42.970.081 (97%) 0 0 2 0 00:00:13 18.02 M 13 63..Kd3 64.Ke6 Kd4 65.Kf5 Kd5 66.Kg5 Ke5 67.Kh6
Pos 126 19.276.083 18.809.795 (97%) 0 0 2 0 00:00:06 18.00 -M11 64.Ke5 Rb5 65.Kf4 Kd4 66.Kf3
Pos 127 30.362.057 29.719.523 (97%) 0 0 2 0 00:00:09 18.00 M 13 64..Rb5 65.Kf4 Kd4 66.Kf3 Rf5 67.Kg4 Ke4 68.Kg3
Pos 128 4.794.286 4.702.866 (98%) 0 0 2 0 00:00:01 15.00 -M 9 65.Kd6 Ke4 66.Kc6 Rd5 67.Kc7 Kd4 68.Kb7 Rc5
Pos 129 4.747.062 4.662.636 (98%) 0 0 2 0 00:00:01 15.03 M 10 65..Ke4 66.Kc6 Rd5 67.Kc7 Ke5 68.Kc8
Pos 130 3.668.755 3.597.077 (98%) 0 0 2 0 00:00:01 16.00 -M 9 66.Kc6 Rd5 67.Kb7
Pos 131 3.209.236 3.140.611 (97%) 0 0 2 0 00:00:01 15.07 M 09 66..Rf5 67.Kd6 Re5
Pos 132 12.906.180 12.661.133 (98%) 0 0 2 0 00:00:03 17.00 -M11 67.Kd6 Rd5 68.Kc7 Re5 69.Kc8 Rc5 70.Kd8 Rd5
Pos 133 1.704.424 1.673.678 (98%) 0 0 2 0 00:00:00 14.00 M 09 67..Re5 68.Kc6 Rd5 69.Kc7 Kd4 70.Kc6 Kc4 71.Kb6
Pos 134 2.104.545 2.067.046 (98%) 0 0 2 0 00:00:00 15.00 -M 9 68.Kc6 Rd5 69.Kb6
Pos 135 623.976 613.617 (98%) 0 0 2 0 00:00:00 13.00 M 09 68..Rd5 69.Kb6 Ke5 70.Kc7 Ke6 71.Kc6 Rg5 72.Kb6
Pos 136 493.110 484.611 (98%) 0 0 2 0 00:00:00 13.00 -M 7 69.Kb6 Kd4
Pos 137 763.331 748.031 (97%) 0 0 2 0 00:00:00 13.00 M 09 69..Ke5 70.Kc6 Ke6 71.Kb6 Kd7 72.Ka7 Kc6 73.Ka6
Pos 138 328.586 321.172 (97%) 0 0 2 0 00:00:00 12.00 -M 8 70.Kc6 Ke6 71.Kc7 Rd6 72.Kc8 Kd5 73.Kc7 Kc5
Pos 139 191.228 186.669 (97%) 0 0 2 0 00:00:00 11.00 M 07 70..Ke6 71.Kc7 Rd6 72.Kc8 Rc6 73.Kb7 Kd6 74.Ka8
Pos 140 18.401 17.831 (96%) 0 0 2 0 00:00:00 8.00 -5.64 71.Kc7 Rd6 72.Kc8 Rd7 73.Kb8 Ke5 74.Kc8 Ra7
Pos 141 73.532 71.216 (96%) 0 0 2 0 00:00:00 10.00 M 06 71..Rd6 72.Kc8 Rc6 73.Kb7 Kd6 74.Kb8 Kd7 75.Kb7
Pos 142 21.327 20.353 (95%) 0 0 2 0 00:00:00 9.00 -M 6 72.Kb8 Rd7 73.Kc8 Kd6 74.Kb8 Kc6 75.Kc8 Rd1
Pos 143 83.349 80.799 (96%) 0 0 2 0 00:00:00 10.00 M 07 72..Rd7 73.Kc8 Kd6
Pos 144 38.429 37.317 (97%) 0 0 2 0 00:00:00 10.00 -M 5 73.Kc8 Kd6 74.Kb8 Rc7 75.Ka8 Kc5 76.Kb8 Kb6
Pos 145 28.223 27.412 (97%) 0 0 2 0 00:00:00 9.00 M 06 73..Kd6 74.Kb8 Rc7 75.Ka8 Rg7 76.Kb8 Kc6 77.Ka8
Pos 146 6.349 6.174 (97%) 0 0 2 0 00:00:00 7.00 -5.72 74.Kb8 Rc7 75.Ka8 Rh7 76.Kb8 Kc6 77.Ka8 Kb8
Pos 147 6.203 5.996 (96%) 0 0 2 0 00:00:00 7.00 M 05 74..Rc7 75.Ka8 Rh7 76.Kb8 Kc6 77.Ka8 Kb6 78.Kb8
Pos 148 1.852 1.807 (97%) 0 0 2 0 00:00:00 6.00 -M 4 75.Ka8 Rh7 76.Kb8 Kc6 77.Ka8 Kb6 78.Kb8 Rh8
Pos 149 2.792 2.712 (97%) 0 0 2 0 00:00:00 5.14 M 03 75..Kc5 76.Kb8 Kb6 77.Ka8 Rc8
Pos 150 108 100 (92%) 0 0 2 0 00:00:00 3.00 -M 2 76.Kb8 Kb6 77.Ka8 Rc8
Pos 151 74 66 (89%) 0 0 2 0 00:00:00 2.07 M 02 76..Kb6 77.Ka8 Rc8
Pos 152 35 30 (86%) 0 0 2 0 00:00:00 2.00 -M 1 77.Ka8 Rc8
Pos 153 9 9 (100%) 0 0 2 0 00:00:00 1.16 Mate Rc8
90% of coding is debugging, the other 10% is writing bugs.