Question about Program called Belle

[bigo]

I read somewhere that Belle was the first Master level Chess program, i find this hard to believe so far back. Where could I find some of it's games? I assume it had master opposition from humans?

[Tobias Lagemann]

Belle Games -> http://www.chessgames.com/perl/chessplayer?pid=23230

Tobias

[bob]

I read somewhere that Belle was the first Master level Chess program, i find this hard to believe so far back. Where could I find some of it's games? I assume it had master opposition from humans?

Belle was awarded a USCF life master certificate (which you earn when you produce a USCF rating of 2200 or higher, based on 24 games or more (so the rating is not provisional) at the 1983 WCCC in New York. At the time we had a 2250 rating with Cray Blitz, but lacked something like 5 games in rated USCF events. it was the first master-level chess program (with an official USCF rating). But things were moving quickly and Belle did not win the WCCC that year, losing to us in the final round. And then along came deep thought and the records were rewritten monthly...

The requirements back then were that you played 24 rated games in USCF sponsored tournaments. No matches for computers. At that time belle was searching 5K nodes per second and searching 8 plies deep, 6 of those in hardware.

[bob]

Belle Games -> http://www.chessgames.com/perl/chessplayer?pid=23230

Tobias

Huge error there.

Cray Blitz - Belle was most definitely 1-0, that was the year we won our first WCCC.

[Murat]

Cray Blitz - Belle was most definitely 1-0, that was the year we won our first WCCC.

Hi Bob,

I know the hardware for Cray Blitz was a Cray-1 supercomputer in 1982 geting 20.000 positions per second.

Was the hardware any different in 1983?

best regards

Murat :.

[bob]

Cray Blitz - Belle was most definitely 1-0, that was the year we won our first WCCC.

Hi Bob,

I know the hardware for Cray Blitz was a Cray-1 supercomputer in 1982 geting 20.000 positions per second.

Was the hardware any different in 1983?

best regards

Murat :.

Yes. In mid-Summer, 1983, Cray built the prototype for the first dual-CPU machine, called the Cray-XMP2. We used that prototype in the WCCC that year and developed the first parallel version of Cray Blitz for that machine. We were searching something like 20K nodes per second by then, which was about 4x faster than Belle. Note that the 20K is an estimate. When I get to the office, I have an old file folder with various test results for most versions, so I can flip through that to see if I can find a more accurate NPS. Unfortunately most of it is on the old Texas Instruments thermal paper, which is not known for remaining readable for long periods of time.

[Terry McCracken]

Cray Blitz - Belle was most definitely 1-0, that was the year we won our first WCCC.

Hi Bob,

I know the hardware for Cray Blitz was a Cray-1 supercomputer in 1982 geting 20.000 positions per second.

Was the hardware any different in 1983?

best regards

Murat :.

Yes. In mid-Summer, 1983, Cray built the prototype for the first dual-CPU machine, called the Cray-XMP2. We used that prototype in the WCCC that year and developed the first parallel version of Cray Blitz for that machine. We were searching something like 20K nodes per second by then, which was about 4x faster than Belle. Note that the 20K is an estimate. When I get to the office, I have an old file folder with various test results for most versions, so I can flip through that to see if I can find a more accurate NPS. Unfortunately most of it is on the old Texas Instruments thermal paper, which is not known for remaining readable for long periods of time.

Bob, I have an old computer textbook and it claimed that Belle could do 30K per sec.

Was that when you were down to very few pieces in an endgame, or basic ending, probably no Queens or Rooks?

Curious Regards,
Terry

[bob]

Cray Blitz - Belle was most definitely 1-0, that was the year we won our first WCCC.

Hi Bob,

I know the hardware for Cray Blitz was a Cray-1 supercomputer in 1982 geting 20.000 positions per second.

Was the hardware any different in 1983?

best regards

Murat :.

Yes. In mid-Summer, 1983, Cray built the prototype for the first dual-CPU machine, called the Cray-XMP2. We used that prototype in the WCCC that year and developed the first parallel version of Cray Blitz for that machine. We were searching something like 20K nodes per second by then, which was about 4x faster than Belle. Note that the 20K is an estimate. When I get to the office, I have an old file folder with various test results for most versions, so I can flip through that to see if I can find a more accurate NPS. Unfortunately most of it is on the old Texas Instruments thermal paper, which is not known for remaining readable for long periods of time.

Bob, I have an old computer textbook and it claimed that Belle could do 30K per sec.

Was that when you were down to very few pieces in an endgame, or basic ending, probably no Queens or Rooks?

Curious Regards,
Terry

Actually your old book is wrong. As was I. The original belle (1977) did about 1K, the next generation (1978) did 5K with hardware move generation and evaluation, software search. The 1980 version did 160K. And CB was doing 20K so we were 8x slower, but we were significantly more efficient. Belle could only use a very simple move ordering in the hardware, no killers, no hash move, just MVV/LVA ordered captures, then the rest of the moves. It was doing 6 plies in hardware, 2 in software. we were actually outsiearching belle by one ply in 1983 at the WCCC.

So sorry for that 5K error. That was the previous version. The belle in the picture at the top of this thread was the last version, and is at the Smithsonian. That was the 160K 8-ply version.

[sje]

It was along time ago, so I'm a bit hazy on details, so some of this might be inaccurate.

1) Belle existed in several versions, the first of which was software only. I believe that this was the same program that was distributed (source and object) as part of the standard Bell Labs Unix back in the mid 1970s. I can recall playing it on a DEC pdp11-70 (eight 16 bit registers with 22 bit physical address space) way back then. As I recall, it was the only program around that could handle move input in either algebraic or descriptive notation.

2) The software side of every Belle ran on an LSI-11 (or similar pdp-11 system) and was in charge of I/O, hardware control, and the upper region of a search tree. It was only the last version of Belle that had an A/B search in hardware; earlier hardware was similar in spirit to the Cheops machine from Greenblatt at MIT.

3) The 160 Knode/second figure was achieved in part because the Belle hardware would skip completion of position factor evaluation at a node if the material score was too far outside the current A/B window. The rate for quiet positions was rather less, but still good enough for a USCF Master rating.

4) In all of history, the State Department of the United States has gone chasing after only two strong chessplayers: Fischer and Belle. Belle was kidnapped by the US government and held for a couple of weeks prior to a tour of Russia because of the irrational fear that the Soviets would steal its secrets to win the Cold War.

5) Deep Blue and its predecessors were really not much more than replicated Belle-on-a-chip systems. I am somewhat saddened but not surprised that the media failed to mention this during IBM's marketing blitz.

[Carey]

It was along time ago, so I'm a bit hazy on details, so some of this might be inaccurate.

1) Belle existed in several versions, the first of which was software only. I believe that this was the same program that was distributed (source and object) as part of the standard Bell Labs Unix back in the mid 1970s. I can recall playing it on a DEC pdp11-70 (eight 16 bit registers with 22 bit physical address space) way back then. As I recall, it was the only program around that could handle move input in either algebraic or descriptive notation.

Pretty much. All I want to add is that there are actually several versions of T.Belle floating around.

Some with assembler and others with C, some with a little more advanced evaluator, etc.

As for the descriptive notation... I don't remember that part.

2) The software side of every Belle ran on an LSI-11 (or similar pdp-11 system) and was in charge of I/O, hardware control, and the upper region of a search tree. It was only the last version of Belle that had an A/B search in hardware; earlier hardware was similar in spirit to the Cheops machine from Greenblatt at MIT.

The last version is the only one that matters.

The first one was soley experimental proof of concept.

The second was more advanced, but was still a "let's see how well this works" kind of thing.


And speaking of Cheops, the source for the modified TECH-II that was used with Cheops is available. Unfortunately, the source for MacHackVI isn't. He doesn't seem to want to hunt for it, even if it still exists. I can't find an email address for him to ask him directly though.

3) The 160 Knode/second figure was achieved in part because the Belle hardware would skip completion of position factor evaluation at a node if the material score was too far outside the current A/B window. The rate for quiet positions was rather less, but still good enough for a USCF Master rating.

KT & JC said it had a node rate of 100k to 200k depending on the position.

4) In all of history, the State Department of the United States has gone chasing after only two strong chessplayers: Fischer and Belle. Belle was kidnapped by the US government and held for a couple of weeks prior to a tour of Russia because of the irrational fear that the Soviets would steal its secrets to win the Cold War.

More concerned about the hardware itself. Some of it was state of the art that wasn't supposed to be exported. Regardless whether it played chess or not.

The Soviets were already doing quite a bit of industrial espionage and reproducing US & European computer equipment.

They probably already had all the hardware info they could have gotten from Belle, but it did still contain equipment that wasn't supposed to be exported.

And I have to frown a bit on your use of 'kidnapped'. Impounded would be the right term for a computer.

5) Deep Blue and its predecessors were really not much more than replicated Belle-on-a-chip systems. I am somewhat saddened but not surprised that the media failed to mention this during IBM's marketing blitz.

Wouldn't have been much point. Nice, but not much point.

DB was so far beyond Belle that it would have made as much sense to mention CHEOPS.

Neither would have meant anything to most people.

Belle ran on, what, a base 10mhz clock? Cheops on 5.5mhz.

Hard to compare either of them to Deep Blue.