64 bit chess software

[seriously disabled]

Hi all,

 

I have a question. It's regarding chess software. The strongest playing chess playing chess program currently is Rybka 3 developed by Vasik Rajlich. Now there are two compiles of Rybka: there is a 32 version for 32 bit Windows and a 64 bit version for 64 bit Windows. I realized that the 64 bit version playes quite stronger than the 32 bit version.

 

What is the reason for this? Why does the 64 bit version play much stronger?

[Pangloss]

It's probably hard to say for sure without knowing a lot about the program. It's possible that the program was optimized for 64-bit and that the 32-bit version is the one that's actually showing a difference (i.e. hobbled by the "smaller" architecture). Or perhaps the routines it uses are just wildly enhanced by the wide register access of 64-bit chips (I'm told some mathematically-intense programs are much faster in 64-bit environments).

 

Check back, there are folks here who might have a better answer for you.

[seriously disabled]

It's probably hard to say for sure without knowing a lot about the program. It's possible that the program was optimized for 64-bit and that the 32-bit version is the one that's actually showing a difference (i.e. hobbled by the "smaller" architecture). Or perhaps the routines it uses are just wildly enhanced by the wide register access of 64-bit chips (I'm told some mathematically-intense programs are much faster in 64-bit environments).

 

Check back, there are folks here who might have a better answer for you.

 

I read somewhere that 64-bit RAM address space equals 2^64 bytes of RAM is something on the order of a billion gigabytes of RAM. Why is that?

[SH3RL0CK]

You might gain some insights by reading up on the most powerful chess program ever developed:

 

http://en.wikipedia.org/wiki/Deep_Blue_(chess_computer)

[seriously disabled]

You might gain some insights by reading up on the most powerful chess program ever developed:

 

http://en.wikipedia.org/wiki/Deep_Blue_(chess_computer)

 

Deep Blue is no longer the most powerful chess program. Deep Blue was simply a very fast computer (called the RS 6000) but the chess program was crappy.

 

Modern chess programs like Deep Rybka 3, Deep Shredder XP, Crafty (developed by Robert Hyatt), Naum 4 are much stronger than the chess program Deep Blue was running.

[SH3RL0CK]

How can you call the program crappy when it beat the reigning world champion? Do you think Kasparov would struggle to beat Rybka 3?

 

Maybe the software wasn't as elegant as todays versions, but you cannot deny its victory over Kasparov.

[seriously disabled]

How can you call the program crappy when it beat the reigning world champion? Do you think Kasparov would struggle to beat Rybka 3?

 

Kasparov is no longer the world champion. Kasparov resigned from competitive chess on March 10, 2005 to devote his time to politics and writing.

 

Viswanathan Anand is currently the chess world champion but Veselin Topalov is currently the highest rated player in FIDE (World Chess Federation).

 

Maybe the software wasn't as elegant as todays versions, but you cannot deny its victory over Kasparov.

 

Deep Blue was the best chess entity at this time but chess software technology has advanced a lot since Deep Blue. Modern chess programs will beat Deep Blue rather easily. Deep Rybka 3 running on a 52-core computer cluster proved to be almost unbeatable.

 

Cluster computing (also called distributed computing) is the future.

Edited May 25, 2009 by Uri

[SH3RL0CK]

Be that as it may, the wiki article describes some of the techniques used; and links for additional information on these. See for example from the wiki article:

 

The system derived its playing strength mainly out of brute force computing power. It was a massively parallel, RS/6000 SP Thin P2SC-based system with 30-nodes, with each node containing a 120 MHz P2SC microprocessors for a total of 30, enhanced with 480 special purpose VLSI chess chips. Its chess playing program was written in C and ran under the AIX operating system. It was capable of evaluating 200 million positions per second, twice as fast as the 1996 version. In June 1997, Deep Blue was the 259th most powerful supercomputer according to the TOP500 list, achieving 11.38 GFLOPS on the High-Performance LINPACK benchmark.[12]

 

 

Being massively parallel gave Deep Blue its computational power. The differences pointed out by Pangloss allude to the "larger" architecture which may provide the computational difference you notice.

 

Maybe someone else might have more insight.

[seriously disabled]

Maybe someone else might have more insight.

 

If you want more insight about computer chess then visit:

 

http://www.talkchess.com/forum/viewforum.php?f=2&sid=f4720db1e54ac301c1e1044a0133c499

 

Some of the people there are much more knowledgeable on the subject than me.

Edited May 26, 2009 by Uri

[Reaper]

How can you call the program crappy when it beat the reigning world champion? Do you think Kasparov would struggle to beat Rybka 3?

 

 

He was only beat in the first game though. The final result of the 1996 match was 4-2, with Kasparov having 3 wins and 2 draws; he was clearly a superior player. The 1997 match is not quite as clear, as there is the possibility that the IBM team may have cheated.

 

Today's computers are way stronger than Deep Blue though. While Kasparov has only played Blitz games with Deep Fritz, the Hydra-Adams match of 2005 gives insight into what would probably happen if the best human players were to go up against the best machines of today:

 

http://www.chessbase.com/newsdetail.asp?newsid=2476

---

Merged post follows:

Consecutive posts merged

Anyways, I would have to agree with Pangloss here. 64-bit machines, so I'm told, tend to have better performance than 32 bit machines in general.

 

As for chess programs in general, the focus nowadays tends to rely much more on software rather than hardware. Deep Blue had dedicated chess hardware, which allowed it to calculate upwards of 200 million positions per second. As of right now, brute force calculations still remains the best way of creating programs capable of defeating the best human players.

 

However, modern software utilizes a large registry of opening, middlegame, and endgame databases, which allows it to retain a high level of play even on a PC.

 

For technical details of Deep Blue and modern computer programs, you can refer to this article:

 

http://www.extremetech.com/article2/0,2845,1915528,00.asp

 

This one is a good article on how, exactly, chess engines work:

http://www.cs.ualberta.ca/~tony/ICCA/anatomy.html

Edited July 1, 2009 by Reaper
Consecutive posts merged.

[seriously disabled]

For those of you interested in computer chess you need to check this stuff out...

 

Hardware expert Lukas Cimiotti (nick "Kullberg" in the forum) built the ultimate chess computer system, a cluster consisting of four Nehalem dual socket Intel Xeon W5580 (=8 cores each) computers 2.93 - 3.33 GHz and 5 Nehalem quads 4.2 GHz.

 

 

The cluster uses a private version of Rybka (Rybka 3 is currently the strongest chess software in the market), which, besides of supporting clusters, is stronger than Rybka 3.

 

So far the cluster lost only one game (of 24) against a human + computer team!

 

Link:

 

http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?pid=170376

[seriously disabled]

Lukas Cimiotti has already upgraded his cluster from 52 cores to 72 cores. He is waiting another 4 months to upgrade to 76 cores which is the next small change (for him).

 

The cluster now has 9 Nehalem EP (2x4 cores) systems. When there is a dual socket six core system by Intel, we will see a cluster with 76-, 80-, ..., 108 (9x2x6) cores!

 

AMD systems don´t count!

 

Lukas's cluster is already alot stronger than Deep Blue and even much stronger than Garry Kasparov who beat Deep Blue.

[bascule]

The simple answer:

 

There are 64 squares on a checkerboard. With a 64-bit CPU this allows you to build "masks" of the board as a single integer. For example, you might represent all of the possible spaces a given piece can move to as a single integer.

 

This is very handy for performance reasons.

[Chriton]

A chess game is all about looking ahead and predicting what your opponent will do.

 

A Computer can be programmed to look more ahead, this is better than a Human and Chess programmes are based on this, I know because I wrote a chess program, so many "If....then".

 

The only thing a Computer has against it is the unpredictable moves of Humans.

 

Chess is a mental and logical battle between people, a Computer cannot do that.