Komodo 12.2 Announcement

[mjlef]

That does look odd. I will take a look. I have not seen that before. Was this from the start position?

Mark, I also just tried the Windows version.
Same issue.

Code: Select all

uciok
setoption name Use MCTS value true
ucinewgame
isready
readyok
go infinite
info depth 10 seldepth 5 time 2002 score cp 33 nodes 157 nps 78 pv e2e4 e7e5 g1f3 b8c6 f1c4 f8e7
info depth 12 seldepth 9 time 4003 score cp 32 nodes 340 nps 84 pv e2e4 e7e5 g1f3 b8c6 f1b5 a7a6 b5c6 d7c6 e1g1 c8e6
info depth 13 seldepth 10 time 6004 score cp 29 nodes 592 nps 98 pv e2e4 e7e5 g1f3 b8c6 f1b5 a7a6 b5c6 d7c6 e1g1 c8e6 d2d4
info depth 14 seldepth 10 time 8004 score cp 28 nodes 740 nps 92 pv e2e4 e7e5 g1f3 b8c6 f1b5 a7a6 b5c6 d7c6 e1g1 c8e6 d2d4
stop
info nodes 830
info depth 14 seldepth 10 time 9721 nodes 830 nps 85 score cp 8
bestmove e2e4 ponder e7e5

Thanks, I would expect it to behave the same whatever the version is. And I promise to look into it, but am quite tired today after sending out all the copies to testers and answering questions. One thing to keep in mind is the lines with times like this: time 8004 , are simply updates every two seconds of the best move (or best moves if MultiPV). There is always going to be some time between those updates and the final score and estmove output. I should have forced a full PV line display at the end just before the bestmove line. There were another 90 scores backed up between the 8 second update and final bestmove which will generally change the score, but going from 28 to 8 does seem like too much for 90 score backups. I will report what I find and make any improvement/changes available, of course.

[MonteCarlo]

Just another example along these lines, but more extreme. Maybe this will help shed some light:

Code: Select all

setoption name use mcts value true
position startpos moves g2g4 e7e5 f2f4
go infinite
info depth 8 seldepth 1 time 2001 score cp 2858469 nodes 54 nps 26 pv d8h4
info depth 10 seldepth 1 time 4003 score cp 3871595 nodes 118 nps 29 pv d8h4
info depth 11 seldepth 2 time 6006 score cp 5105332 nodes 184 nps 30 pv d8h4
info depth 12 seldepth 3 time 8009 score cp 6951263 nodes 357 nps 44 pv d8h4
info depth 12 seldepth 3 time 10010 score cp 7611787 nodes 425 nps 42 pv d8h4
info depth 13 seldepth 3 time 12013 score cp 8411022 nodes 491 nps 40 pv d8h4
info depth 13 seldepth 4 time 14014 score cp 9217555 nodes 612 nps 43 pv d8h4
stop
info nodes 670
info depth 13 seldepth 4 time 15319 nodes 670 nps 43 score cp 49
bestmove d8h4

[simon]

I think the software licence is too restrictive. It should allow to you run the software on machines you rent as well as those you own.

[lkaufman]

Komodo 12.2 has been released at komodochess.com. As usual it is free for subscribers and discounted 20% for purchasers of Komodo 11 or higher. It includes both the normal and MCTS versions. This is a rather major release, for several reasons.

The strength of the MCTS version is vastly improved, up around 200 elo points at blitz levels, perhaps 150 at much longer time controls. Our predicted ratings for it on the CCRL 40/4 scale are 3320 on 1 thread, 3350 on 2, and 3380 on 4 threads. We now believe that the MCTS version is stronger than the normal one when both are using MultiPV = 3 or more, and would rank in the top five independent engines if substituted for normal Komodo.

The low thread limit for MCTS is gone, we believe it can benefit from up to 64 threads, although this has not yet been proven beyond 30 threads.
The many-years old slow startup of Komodo on big machines with big hash tables is solved! Also Komodo is now able to make better use of big hardware with more than one “NUMA Node”. These are important for major tournaments like CCC and TCEC, and most people with 12 or more cores should see some benefits.

The playing strength of normal Komodo has been improved about 10 elo on four threads at blitz time controls, somewhat less on single thread. In general MP play has been improved noticeably since last (12.1.1) release.
Several minor bugs have been fixed.

We are considering splitting normal and MCTS Komodo into two separate engines in the future, but if so subscribers will be eligible to receive both. As to whether the MCTS version will surpass the normal one and become our main (or even only) version, only time will tell. Now that the gap is down to about 100 elo, it is at least a realistic possibility.

Thank you for this new version.

However, there seems to be a problem/bug with the displayed score in the MCTS version when outputting the final info line after the search stops.
Didn't check the A/B version.

Example 1

Code: Select all

info depth 22 seldepth 29 time 58059 score cp -112 nodes 34122 nps 587 pv f1f4 c6a4 f4g4 e7f6 f5h5 f8f7 g4g6 e4g5 h5h7 g5h7 g6f7 b7f7 b3f7 c3b2 b1b2 a4c6 d4f6 h7f6 g2g3 g7g5 b2c1
info nodes 35333
info depth 22 seldepth 29 time 60001 nodes 35333 nps 588 score cp -28
bestmove f1f4 ponder c6a4

Example 2

Code: Select all

info depth 22 seldepth 16 time 58044 score cp 992 nodes 34904 nps 601 pv g2g7 f7e8 d6e7
info nodes 36421
info depth 22 seldepth 16 time 60001 nodes 36421 nps 607 score cp 45
bestmove g2g7 ponder f7e8

Please note the difference in the final score compared to the former info line.

Komodo MCTS was running in a terminal under Linux, 1024 MB MCTS Hash, MCTS Explore set to 12, 1 min per position.
It also happens after a 'go infinite' and 'stop'.

Yes, it seems that the final displayed score is typically a quarter or less of the real score. It's probably only a display issue, but we'll investigate further.

[duncan]

Komodo 12.2 has been released at komodochess.com. As usual it is free for subscribers and discounted 20% for purchasers of Komodo 11 or higher. It includes both the normal and MCTS versions. This is a rather major release, for several reasons.

The strength of the MCTS version is vastly improved, up around 200 elo points at blitz levels, perhaps 150 at much longer time controls. Our predicted ratings for it on the CCRL 40/4 scale are 3320 on 1 thread, 3350 on 2, and 3380 on 4 threads. We now believe that the MCTS version is stronger than the normal one when both are using MultiPV = 3 or more, and would rank in the top five independent engines if substituted for normal Komodo.

The low thread limit for MCTS is gone, we believe it can benefit from up to 64 threads, although this has not yet been proven beyond 30 threads.
The many-years old slow startup of Komodo on big machines with big hash tables is solved! Also Komodo is now able to make better use of big hardware with more than one “NUMA Node”. These are important for major tournaments like CCC and TCEC, and most people with 12 or more cores should see some benefits.

The playing strength of normal Komodo has been improved about 10 elo on four threads at blitz time controls, somewhat less on single thread. In general MP play has been improved noticeably since last (12.1.1) release.
Several minor bugs have been fixed.

We are considering splitting normal and MCTS Komodo into two separate engines in the future, but if so subscribers will be eligible to receive both. As to whether the MCTS version will surpass the normal one and become our main (or even only) version, only time will tell. Now that the gap is down to about 100 elo, it is at least a realistic possibility.

Good achievement. Thanks.

[Gian-Carlo Pascutto]

Do you support 7 men EGTB in this update?

Does the "smart TB access mode" work for them?

[mjlef]

Do you support 7 men EGTB in this update?

Does the "smart TB access mode" work for them?

No, we do not yet have support for 7 piece Syzygy, but it is on the list.

[mjlef]

Yes, it seems that the final displayed score is typically a quarter or less of the real score. It's probably only a display issue, but we'll investigate further.

I traced the issue down to a missing pi (3.141592653....) in the final conversion of the win percentage to centipawns. This scales it down a lot. I will include the fix in a free update. Note this does not change the move the MCTS search selects, it just pulls the score towards zero. Another user reported in the Fritz GUI that the eval will appear to "flip" when Komodo MCTS is winning. I have traced that to Fritz being unable to handles scores above 32768, where it apparently wraps around to a negative score (looks like they use 16 bit signed integers for that). Note MCTS has a highest win percentage, but as you approach that it can to to very high numbers. I will include a change to accommodate that popular GUI as well by limiting the scores to -32000...32000. We will need to test these changes and see if anyone else finds something, but expect the free update in a few days. Thanks again again for the report.

[Jouni]

When running test suite (Arasan) in Fritz GUI (4 cores bmi version) I get constantly error message as: "Engine Komodo 12.2 64b caused an exception". With 1 core I didn't see that error. Previously with any Komodo no crashes.

[Jouni]

Additional testing: crashes also with one core in Fritz GUI. And log file don't say anything helpful ..