|
< Earlier Kibitzing · PAGE 2 OF 4 ·
Later Kibitzing> |
| Jan-10-15 | | zanzibar: I just submitted all 64 games from TCEC Season 7 Superfinal (Komodo--Stockfish). Hopefully they get added in soon. |
|
| Jul-02-15 | | truefriends: Komodo lost to Jonny in round 5 of the WCCC 2015.
After 7/9 rounds Jonny leads .5 points ahead of Komodo. The last 2 rounds will be played today, July 3th. |
|
Jul-02-15
 | | AylerKupp: The WCCC is a joke and has been that way for several years. Imagine having a Candidates Tournament without Anand, Topalov, Nakamura, Caruana, Giri, Kramnik, and So participating. That's the equivalent of having a computer "world championship" without Stockfish, Houdini, Fire, Gull, Equinox, Critter, and Rybka participating. I have no idea why Larry Kaufman bothered to enter Komodo in the tournament. Allowing Jonny to run on a system with 2,400 cores when Komodo is running on a system with the next highest number of cores (24) is another indication of this silly state of affairs. What would a victory by Jonny prove? When running on similar computer systems, e.g. the latest CCRL tournament, http://computerchess.org.uk/ccrl/40..., Jonny is ranked #56-57. |
|
| Jul-03-15 | | truefriends: < That's the equivalent of having a computer "world championship" without Stockfish, Houdini, Fire, Gull, Equinox, Critter, and Rybka participating.> SF is this only engine in this group with active development and maybe Equinox. So comparing them to the top (active) super GM's is kind of strange TMHO. And btw I don't think Wesley So will make it to the candidates out of your list ;-) |
|
| Aug-18-15 | | Zhbugnoimt: When is TCEC season 8? |
|
Aug-18-15
 | | tamar: Starts this Friday. http://tcec.chessdom.com/live.php |
|
| Aug-23-15 | | NeverAgain: I just posted some research on using Stockfish 6 for game analysis on SF's page:
Stockfish (Computer) (kibitz #50) It's a bit long (over six pages split into four posts) but I think anyone who takes engine analysis seriously should find the posts interesting. There are some Komodo results in there as well. As I'm currently limited to the last free version (v6) it would be great if anyone with the latest version could independently test any or all cases listed in my article. I also noticed that LMR pruning is a toggable option in Komodo 6's settings. From some limited tests I did here, it seems that turning it off results in a speed penalty of about 25-33%, with a much higher node count for a given ply depth. Is that setting still present in the latest version? Has anyone tested it and observed if it makes any practical impact on analysis? I mean as in significantly changing a position's eval, rather than the obvious changes in the speed and node count numbers. |
|
Oct-23-15
 | | alexmagnus: Crazy king walk today against Protector. |
|
| Dec-01-15 | | ColdSong: Congrats to the best player of the world for this TCEC 8 victory.Impressive solidity and deeper positionnal understanding. |
|
Jan-16-16
 | | Penguincw: Komodo defeats Nakamura in a 4-game match: http://www.chess.com/news/komodo-be.... |
|
| Mar-19-16 | | scholes: Komodo 9.4 has been released |
|
Mar-19-16
| | AylerKupp: <scholes> Thanks for the news. I am hardly surprised since I would think that it galled the Komodo developers to have Stockfish be ranked #1 in both the CCRL 40/40 and CEGT 40/20 and 40/120 tournament rating lists by 1 point (CCRL), 3 points (CEGT 40/40), and 6 points (CEGT 40/120). I'm sure that they were working around the clock to get this release out, and they only needed an improvement of a few rating points to get back its #1 ranking. Of course, all that means is there will likely be a Stockfish 8 released soon which will regain the #1 ranking. That will be followed by a Komodo 9.5 or Komodo 10, etc. |
|
| Mar-19-16 | | scholes: <AylerKupp> Stockfish development versions are released every week or two. Latest regression test showed latest development version of sf is 13 elo stronger than sf7 on single core. As per their official release, Komodo 9.4 is 15 elo stronger than Komodo 9.3 on a single core. |
|
Mar-19-16
| | AylerKupp: <scholes> I take those regression test figures with a grain of salt. For example, Komodo 9.3 was quoted as being several 10s of Elo points better than Komodo 9.2, but that's not what engine vs. engine competition in the CCRL and CEGT tournaments showed. So I'll just wait and see if Komodo 9.4 is stronger in engine vs. engine competitions than Komodo 9.3 or, for that matter, Komodo 9.2. |
|
| May-21-16 | | scholes: Komodo wins stage 1a with 14/15 score |
|
May-22-16
| | Penguincw: Yep, congrats to Komodo 9.42 for winning Season 9, Stage 1a with a score of 14/15 (+13,-0,=2), which is also the only undefeated score.. Komodo entered the stage as the highest rated engine overall at 3225 (60.1% score), and it proved why it's the best. http://tcec.chessdom.com/archive.php
http://tcec.chessdom.com/live.php |
|
Jul-07-16
 | | HeMateMe: </5/2016 – It's stiffest competition came from the German program Jonny, running on a giant 2400-core machine. With only 48 cores the US program Komodo finished the computer tournament in Leiden, Netherlands, equal first with Jonny, then went on to win a hard-fought tiebreak, making it the 22nd World Computer Chess Champion. > Don't cores = computing power? If Johnny had 2400 cores, how does Komodo keep up? Better written program? Different type of hardware/interface? |
|
| Jul-07-16 | | scholes: <HeHateMe>Komodo is almost 200 elo stronger than Johnny on a single core. Further both engines were using their own opening books. At their level they would play almost perfect chess after opening book. |
|
| Jul-07-16 | | Appaz: There is little help in massive computing power and high ply depth if your eval algorithm is slightly off, which most are compared to Komodo. Good moves might be pruned quite early. There seem to be a critical area, maybe at 25-35 ply (depending on the position of course), where further depth seems to be of less importance relative to the eval. |
|
| Jul-07-16 | | Appaz: <scholes> I'm pretty sure that Komodo is very far from playing perfect chess. On the contrary, it probably plays quite lousy chess compared to "perfect". Komodo of the year 2020 will probably crush it on the same hardware and even at present time, Stockfish can beat it at regular intervals. Its strength is way beyond most humans chess understanding, that's why its play looks pretty decent to us (okay, <awesome> then). |
|
Jul-07-16
| | HeMateMe: I find it interesting that the elite chess programs don't draw as much against each other as human GMs do. It seems if they have an advantage, they squeeze blood out of it. |
|
Jul-07-16
 | | Sally Simpson: Hi Appaz,
Perhaps a slight misuse of the word 'understanding.'
"Its strength is way beyond most humans chess understanding." It understands nothing about chess. It can calculate way beyond any human and that is it. It does not know the human concept that is more important to want to win than play good moves. The thing does not even know it has won a game. All it's doing is calculating, not understanding. Have they (computers) understood this one yet.
Black to play.
click for larger viewBlack played 45...Qb6 (No computer would play this and every computer would give taking the Queen - with a check. It's top choice. How long after 46.Nxb6+ cxb6 does it take for it to realise it cannot win - or has it found a way after 46.Nxb6+ to win it.) It was a puzzle on here and humans who got it understood the position right away. To get it the thing will have to calculate, it does not understand what is going on. A Petrosian vs L Hazai, 1970 |
|
| Jul-07-16 | | scholes: <HeHateMe> TCEC stage 2 started with 12 straight draws. If even slightly worse, engines rush to exchange pieces to drawn endgame. But even slightly better they try to squeeze blood out of stone. <Appaz> I agree there is lot of improvement possible. But most of improvements will be in play in opening phase of game and detection of fortress. |
|
Jul-07-16
| | tamar: The quality of play Komodo 10 has attained is astounding, and it is about time chessgames started uploading TCEC games here. |
|
Jul-07-16
| | HeMateMe: I find the middle games fascinating, so creative. |
|
 |
|
|
< Earlier Kibitzing · PAGE 2 OF 4 ·
Later Kibitzing> |
Kibitzing
