Komodo eaten by Firebird

[beram]

In a Nunn2 match at 4m2sec on my PC, T4300 (Fritzmark 6,14),
Firebird 32bit (2CPU) crushes Komodo 1.0 32bit (1CPU) with 42,5 – 7,5 .
37 wins and only 2 losses for Firebird (one on time in winning position)

1 FireBird 1.0 w32 +37/=11/-2 85.00% 42.5/50
2 Doch32 1.0 JA +2/=11/-37 15.00% 7.5/50

against Stockfish 1.6.2 JA, it was a 29,5 - 20,5 win

Grts Bram

[kingliveson]

In a Nunn2 match at 4m2sec on my PC, T4300 (Fritzmark 6,14),
Firebird 32bit (2CPU) crushes Komodo 1.0 32bit (1CPU) with 42,5 – 7,5 .
37 wins and only 2 losses for Firebird (one on time in winning position)

1 FireBird 1.0 w32 +37/=11/-2 85.00% 42.5/50
2 Doch32 1.0 JA +2/=11/-37 15.00% 7.5/50

against Stockfish 1.6.2 JA, it was a 29,5 - 20,5 win

Grts Bram

Since Komodo is a single CPU...one-on-one test would have been better because the bird is not a chick learning to fly.

[beram]

In a Nunn2 match at 4m2sec on my PC, T4300 (Fritzmark 6,14),
Firebird 32bit (2CPU) crushes Komodo 1.0 32bit (1CPU) with 42,5 – 7,5 .
37 wins and only 2 losses for Firebird (one on time in winning position)

1 FireBird 1.0 w32 +37/=11/-2 85.00% 42.5/50
2 Doch32 1.0 JA +2/=11/-37 15.00% 7.5/50

against Stockfish 1.6.2 JA, it was a 29,5 - 20,5 win

Grts Bram

Since Komodo is a single CPU...one-on-one test would have been better because the bird is not a chick learning to fly.

And here are the results for 1 CPU Firebird 1.0 against Komodo 1.0

28 wins one loss and 21 draws
T8100, 2100 Mhz, Blitz 4m+2s

1 FireBird 1.0 w32 +28/=21/-1 77.00% 38.5/50
2 Doch32 1.0 JA +1/=21/-28 23.00% 11.5/50

[Dr.Wael Deeb]

How did this little birdy managed to swallow such a huge Lizard

[BubbaTough]

Not terribly shocking. I have seen many similar posts...in fact I don't think I have ever seen so many 50 game matches reported as I have in the last few months. I guess its a new trend by new posters. Usually its at least 1cpu versus 1 cpu though. Anyway, a couple nitpicks:

First, why do you list Doch but say Komodo?
Second, how does Firebird lose on time in a 4 minute game with 2 second increment? Was it a crash or was it a time control bug?

-Sam

[yanquis1972]

im somewhat stunned by this result. i am convinced the 64 bit komodo would fare a bit better against 64 bit FB. admittedly the elo gap is enormous but 28-1 is somewhat preposterous.

sam, initially due to a mistake the first komodo compile was still named doch, & i don't think chessbase lets you change engine names. since this was very quickly remedied it has now become a collector's item and should be worth a considerable sum of money in a few years time.

[BubbaTough]

sam, initially due to a mistake the first komodo compile was still named doch, & i don't think chessbase lets you change engine names. since this was very quickly remedied it has now become a collector's item and should be worth a considerable sum of money in a few years time.

Ahh, gotcha thanks. It looks like I missed out on this collector's edition. I guess you could open the pgn in a word processor and do a global replace on the name, but I must admit I am too lazy to do that kind of thing myself.

-Sam

[beram]

Not terribly shocking. I have seen many similar posts...in fact I don't think I have ever seen so many 50 game matches reported as I have in the last few months. I guess its a new trend by new posters. Usually its at least 1cpu versus 1 cpu though. Anyway, a couple nitpicks:

First, why do you list Doch but say Komodo?
Second, how does Firebird lose on time in a 4 minute game with 2 second increment? Was it a crash or was it a time control bug?

-Sam

It was my fault I had opened a Fritz window while watching that game

[yanquis1972]

i'm running an alternative match, FB 1.1 default with triplebases; same TC, only 64 bit and i guess i will use ponder on (never sure what do about ponder). i'll probably get bored & abandon it but if not, i'll post results. for whatever reason i think K will hold up much better than in bram's test. using silversuite for the OPBK. also have GUI TBs turned on, which in theory could benefit komodo a slight amount (it can't access TBs while ippolit has its own set that serve the same function of GUI TBs).

[yanquis1972]

stopping the test after an even 50 games, or halfway thru albert's excellent opening suite. may finish tomorrow.

result is +27 =17 -6 +156 elo for firebird.

according to ingo bauer's ipon site, komodo is ~120 elo weaker than rybka 3 at 5+3 time controls with ponder on, so this result is about in line with my expectations (even before seeing this), maybe a litte low for firebird (offhand i guessed ~175 elo at 4+2). at any rate i have confirmed that komodo is not completely impotent against the monster, scoring 6 hits.

i haven't really been watching the match, but one thing i've noticed nevertheless is that komodo's time management is quite a bit more intelligent, or certainly more 'human like', spending a good chunk of time on the opening, then leaving a comfortable margin for the rest of the moves. firebird in comparison seems to play with a casual capa-esque arrogance, for example in the game i am now watching it is 11 moves out of book yet still at 3:58 minutes, while komodo has 2:16 remaining on the clock (FB is nevertheless ahead by a bit over a quarter pawn according to both engines). just to pre-empt anyone wondering, the time management for FB isn't exactly broken (it doesn't hover at 4 minutes for a 4+2 game & gradually winds down at least somewhat, tho i dont think in my glances i've seen it below even 2 minutes), it's just not nearly as productive imo as komodo's.