|
< Earlier Kibitzing · PAGE 3 OF 3 ·
Later Kibitzing> |
| Sep-27-07 | | Brown: 9 games done, and there are only 5 here... ;-( |
 |
Sep-27-07
 | | MrSpock: Rybka vs. Zappa in Mexico: 4,5-5,5 (Zappa wins) |
|
| Sep-27-07 | | Brown: More proof that the best programs are not flawless. There is still hope for all those GM's that relish playing super-accurately for over 100 moves ;-) |
|
Sep-27-07
 | | Hesam7: <Brown: More proof that the best programs are not flawless. There is still hope for all those GM's that relish playing super-accurately for over 100 moves ;-)> I think everyone should take into consideration that the time control for this event was: <60 minutes per game plus 20 seconds per move> that is why the games are rather low in quality. On another note this defeat will hurt Rybka. Now the Junior team (chessbase) can avoid a match with Rybka and play Zappa Mexico instead. |
|
| Sep-29-07 | | sanyas: Rybka scales badly to 8-cores. Zappa was better able to take advantage of the hardware. This, despite Cozzie saying that Zappa's strength is in its evaluation, while Rubka relies on intelligent search. |
|
Sep-29-07
| | shr0pshire: <sanyas> Can you explain to me why Rybka scales badly to 8-cores? I don't understand why an engine would perform better or worse on different hardware. I would imagine that it would only perform faster or slower.... |
|
Sep-29-07
| | rover: <On another note this defeat will hurt Rybka. Now the Junior team (chessbase) can avoid a match with Rybka and play Zappa Mexico instead.> If it does indeed happen it won't help Junior much. Not much hope for them either way. |
|
Sep-30-07
| | heuristic: <shr0pshire> here's an abbreviated synopsys. Today's computers are symmetric multiprocessors. The SMPs that are multi-core (like the Xeon 5300 quad-core used in this tournament) are suited for multi-tasking. This means multiple tasks can be run in parallel. This is perfect for servers.
The other type of parallel programming is multi-programming. This means that ONE program (like a chess program) is written to the paradigm appropriate for parallelism. This is much different than serial programming and usually revolves around three concepts : - partitioning : the program may need to be broken up differently - communication : the partitions need to share data and pass control information. - coordination : at specific points in the program, the parallel tasks need to be synchronized. when people talk about "scaling", invariably it boils down to partitioning and communication. the methodology for writing parallel program is nascent. recall that google developed "map-reduce" as their usage model for parallel programming. note that cozzie specifically mentions "shared memory" usage. this means he has an IPC mechanism to do the aformentioned communication. also note that his academic & professional work is in parallel programming (which is a HOT field right now) for me at least, this tournament was a duel between a strong ox with very good chess-specific knowledge and a bunch of chickens with less chess knowledge. (this is a paraphrase of seymour cray's comment about vector machines) anyway, the short answer is that your multicore laptop is much faster at doing multiple things but still about the same speed at doing one thing.
a chess program ("engine") is just one thing to your computer. for it to get faster, it needs to undergo major modifications.
Hope this helps.
|
|
| Oct-01-07 | | bala81: Can anyone tell me where I can get the pgn file for this games, thx. |
|
| Oct-01-07 | | ChessJock: <Can anyone tell me where I can get the pgn file for this games, thx.> Um... right here ;) |
|
Oct-01-07
| | shr0pshire: <hueristic> Thanks for that explanation. I appreciate the explanation because I don't know much more beyond a beginner's level of how most chess engines operate. Knowing what you told me, it will be more interesting to compare engines. Thank you again. |
|
| Oct-03-07 | | timhortons: whats the commercial form of software(zappa) that beats rybka? and why most grandmaster still prefer to use rybka in game analysis despite it was beaten by zappa? is zappa mexico the one that beats rybka? i mean is that the commercial form of it? |
|
Oct-03-07
| | zanshin: <ChessJock: <bala81: Can anyone tell me where I can get the pgn file for this games, thx.>Um... right here ;)> Right. If you are a Premium member, you can download all the games at once using the PGN download button located in the upper righthand corner. Regular members can still download the pgn files one at a time. |
|
Oct-03-07
| | weisyschwarz: <bala81> do you like the playing of Anand? Where in India is he from? |
|
Oct-03-07
 | | WannaBe: <weisyschwartz> Chennai, India, according to: http://en.wikipedia.org/wiki/Viswan... (Right hand side, underneath his picture) |
|
Oct-03-07
| | shr0pshire: <timhortons <whats the commercial form of software(zappa) that beats rybka?>>
The program is called Zappa Mexico.
http://www.shredderchess.com/chess-... |
|
| Oct-04-07 | | BAG: I wonder if they have to both use the same hardware? |
|
| Oct-08-07 | | tal lover: In my opinion Rybka played a better chess in this match, rybka got a winning position in game 9 and blow away the victory, and 2 zappa wins was draw positions that rybka tried to force a win and lose |
|
| Oct-08-07 | | s4life: are there any head-to-head results between these two engines under classical time controls? |
|
| Oct-16-07 | | pawnofdoom: <BAG>Both engines had to play on 8-core machines, so yes. Zappa is supposedly stronger than Rybka on those platforms |
|
Oct-26-07
| | Dionyseus: <pawnofdoom> <Both engines had to play on 8-core machines, so yes. Zappa is supposedly stronger than Rybka on those platforms > Zappa scales very well but Zappa isn't necessarily stronger than Rybka on 8-core machines, statically speaking close results in such short matches doesn't prove one player is stronger than the other. |
|
Apr-17-08
| | Whitehat1963: Latest computer rankings:
http://www.computerchess.org.uk/ccr... |
|
| Apr-17-08 | | D.Observer: On this event, are the only participants Zappa (Computer) and Rybka (Computer)? |
|
Apr-18-08
| | Karpova: <D.Observer>
This was simply a match between Rybka and Zappa so the answer is yes. |
|
Oct-06-08
| | Whitehat1963: Interesting commentary on this match from Zappa's creator: http://74.125.95.104/search?q=cache... |
|
|
|
|
< Earlier Kibitzing · PAGE 3 OF 3 ·
Later Kibitzing> |
Kibitzing
