http://cgi.ebay.com.au/ws/eBayISAPI.dll ... K:MEWAX:IT
UK993... close to $2k including shipping in my money. My collection numbers
a little over 30 now, but not a single Mephisto wooden board.
It pays to have a 100% feedback rating. This machine came with a couple
of extra eproms, although they seem to be cheap on ebay at the moment. I'm
currently waiting for two of them, one for Montreux and one for Berlin Pro London. The shipping is peanuts.
take care all...
Larry
This one brought a pretty penny...
Moderators: Harvey Williamson, Steve B, Watchman
Forum rules
This textbox is used to restore diagrams posted with the fen tag before the upgrade.
This textbox is used to restore diagrams posted with the fen tag before the upgrade.
-
AMD64inside
- Member
- Posts: 20
- Joined: Tue Jul 31, 2007 7:31 pm
It appears the sale fell through as the seller has a new listing with different pictures for another / same? Lyon with upgrade Vancover and London epproms.
http://cgi.ebay.co.uk/Chess-computer-Me ... 1e5ea437d8
The seller also recently had a Exclusive Polger + the 10 MHz upgrade chip up for auction, but with about 2 days to go and bids up to £220, the seller suddenly cancelled all bids and cancelled the auction.
Puzzled Regards
Darrell
http://cgi.ebay.co.uk/Chess-computer-Me ... 1e5ea437d8
The seller also recently had a Exclusive Polger + the 10 MHz upgrade chip up for auction, but with about 2 days to go and bids up to £220, the seller suddenly cancelled all bids and cancelled the auction.
Puzzled Regards
Darrell
"The game itself is nothing, the playing of it is everything"
-
Steve B
- Site Admin
- Posts: 10144
- Joined: Sun Jul 29, 2007 10:02 am
- Location: New York City USofA
- Contact:
That is oddAMD64inside wrote:It appears the sale fell through as the seller has a new listing with different pictures for another / same? Lyon with upgrade Vancover and London epproms.
http://cgi.ebay.co.uk/Chess-computer-Me ... 1e5ea437d8
The seller also recently had a Exclusive Polger + the 10 MHz upgrade chip up for auction, but with about 2 days to go and bids up to £220, the seller suddenly cancelled all bids and cancelled the auction.
Puzzled Regards
Darrell
two sales fall through for the same seller
seems there might be something wrong with the Seller's description of the computers?
Unusual Regards
Steve
-
AMD64inside
- Member
- Posts: 20
- Joined: Tue Jul 31, 2007 7:31 pm
-
AMD64inside
- Member
- Posts: 20
- Joined: Tue Jul 31, 2007 7:31 pm
Well this time the auction fetched £1,019 so a little more than the first attempt.
Also the seller has relisted the previously mention Exclusive Polgar complete with 10Mhz upgrade chip that mysteriously ended abruptly.
http://cgi.ebay.co.uk/Chess-computer-Me ... 1e5ede4ec2
Second time lucky Regards .....
Darrell
Also the seller has relisted the previously mention Exclusive Polgar complete with 10Mhz upgrade chip that mysteriously ended abruptly.
http://cgi.ebay.co.uk/Chess-computer-Me ... 1e5ede4ec2
Second time lucky Regards .....
Darrell
"The game itself is nothing, the playing of it is everything"
Lucky guy indeed.
About a month ago, the Mephisto Lyon seller asked for 'only' GBP 500 in this Craigslist London listing:
http://london.craigslist.co.uk/tag/1961944896.html
He was a lucky mofo that no one contacted him to buy the computer at that price.
He who laughs last laughs best.
Kostea
PS Craigslist listings expire or are deleted by their author, so if you read my post 3 days after I added it, the Craigslist link may not lead to anything.
About a month ago, the Mephisto Lyon seller asked for 'only' GBP 500 in this Craigslist London listing:
http://london.craigslist.co.uk/tag/1961944896.html
He was a lucky mofo that no one contacted him to buy the computer at that price.
He who laughs last laughs best.
Kostea
PS Craigslist listings expire or are deleted by their author, so if you read my post 3 days after I added it, the Craigslist link may not lead to anything.
Hi Ricardo,
There is no need for faster RAM.
Actually, believe it or not, the 6502 CPU is the bottleneck in the POLGAR module.
Here is a simple method to calculate timings.
The upgrade CPU has a 10 MHz frecquency ( 10 million Hertz).
That means that it can run 10 000 000 cycles per second.
A cycle time duration is 100 ns. (T = 1/F = 1/10 000 000 = 100 nanoseconds)
A 6502 CPU needs at least 4 cycles to read or write from/to a location of RAM memory.
This results in a:
4 x 100 ns = 400 ns operation
(this is the amount of time it takes the CPU to execute a software instruction for reading/writing to RAM memory).
Immediately after the read/write software instruction finished execution, the 6502 CPU needs some extra nanoseconds to have 'stable' signals on the address and data buses, and READ or WRITE pins.
So the 6502 CPU needs at least 400 ns before it is ready to do anything with the RAM memory.
The ROM and RAM in the POLGAR module have a 150 ns access time, meaning that they are almost 3 times faster than what the system can throw at them.
The more correct/complex way to calculate timing takes into account signal propagation delays thru some other separate chips that are interconnected between the CPU and ROM/RAM. For example, most old chess computers have chips for memory decoders, the 'newer' ones from around 1990s have extra chips for bus drivers.
For the POLGAR module it may take about 500 ns before a CPU 'fart' is smelled by the RAM/ROM chips.
Speedy Gonzalez regards,
Kostea
There is no need for faster RAM.
Actually, believe it or not, the 6502 CPU is the bottleneck in the POLGAR module.
Here is a simple method to calculate timings.
The upgrade CPU has a 10 MHz frecquency ( 10 million Hertz).
That means that it can run 10 000 000 cycles per second.
A cycle time duration is 100 ns. (T = 1/F = 1/10 000 000 = 100 nanoseconds)
A 6502 CPU needs at least 4 cycles to read or write from/to a location of RAM memory.
This results in a:
4 x 100 ns = 400 ns operation
(this is the amount of time it takes the CPU to execute a software instruction for reading/writing to RAM memory).
Immediately after the read/write software instruction finished execution, the 6502 CPU needs some extra nanoseconds to have 'stable' signals on the address and data buses, and READ or WRITE pins.
So the 6502 CPU needs at least 400 ns before it is ready to do anything with the RAM memory.
The ROM and RAM in the POLGAR module have a 150 ns access time, meaning that they are almost 3 times faster than what the system can throw at them.
The more correct/complex way to calculate timing takes into account signal propagation delays thru some other separate chips that are interconnected between the CPU and ROM/RAM. For example, most old chess computers have chips for memory decoders, the 'newer' ones from around 1990s have extra chips for bus drivers.
For the POLGAR module it may take about 500 ns before a CPU 'fart' is smelled by the RAM/ROM chips.
Speedy Gonzalez regards,
Kostea
Thanks Kostea,
Three more questions:
1) What about the heating. Can the old 65c02 operate for hours at this new frecuency of 10Mhz?.
2) If the 65c02 needs 4 cycles to read or write in the memory that means that i can still go up to 15 Mhz. The ram that it is installed in my polgar is a SRM2264LC10 , the datasheet for that RAM says is a 100 ns.
3)Can i climb up to 20 Mhz?
Frankenstein regards
Ricardo
Three more questions:
1) What about the heating. Can the old 65c02 operate for hours at this new frecuency of 10Mhz?.
2) If the 65c02 needs 4 cycles to read or write in the memory that means that i can still go up to 15 Mhz. The ram that it is installed in my polgar is a SRM2264LC10 , the datasheet for that RAM says is a 100 ns.
3)Can i climb up to 20 Mhz?
Frankenstein regards
Ricardo
-
Steve B
- Site Admin
- Posts: 10144
- Joined: Sun Jul 29, 2007 10:02 am
- Location: New York City USofA
- Contact:
Its too much work Ricardoricard60 wrote:Thanks Kostea,
Three more questions:
1) What about the heating. Can the old 65c02 operate for hours at this new frecuency of 10Mhz?.
2) If the 65c02 needs 4 cycles to read or write in the memory that means that i can still go up to 15 Mhz. The ram that it is installed in my polgar is a SRM2264LC10 , the datasheet for that RAM says is a 100 ns.
3)Can i climb up to 20 Mhz?
Frankenstein regards
Ricardo
just stop fooling around and get a REAL Polgar 10 Mhz already
http://www.flickr.com/photos/10261668@N ... 922171154/
Non-Modified Regards
Steve
This guy sells it:
http://cgi.ebay.co.uk/Chess-computer-Me ... 1e5ede4ec2
but i already have the exclusive board.
no room for more boards regards
Ricardo
http://cgi.ebay.co.uk/Chess-computer-Me ... 1e5ede4ec2
but i already have the exclusive board.
no room for more boards regards
Ricardo
Hi Ricardo,
What I said below is not an attempt to discourage anyone from overclocking.
Just some things to think about.
Before I attempt to answer your questions, let me clarify one thing. The CPU in the POLGAR reads a lot from ROM too (where the chess engine is stored). The ROM is generally considered slower than the RAM, but for a correct answer you have to compare the ROM and RAM chip IDs in your particular case and find out their speeds.
Besides the RAM, you have to have a fast ROM too. The sticker on the upgrade ROM that says '10 MHz', and that only means that the chess software has been modified to run at 10 MHz.
The ROM speed is stamped on its capsule by the manufacturer.
One thing that others have mentioned before. The Saitek RISC 2500 (and Montreux) load their chess engine from ROM to RAM and then execute the software from RAM. In that particular chess computer, the RAM is 2 times faster than the ROM.
1. That's a good question. A CPU built and stamped as a 10 MHz must reliably work at that speed. The problem here is that the seller only provides a ROM and a quartz crystal. No new CPU, so the old CPU is still a 5 MHz.
Some manufacturers made CPUs at the highest frequency the technology permitted, then after testing, the ones running fine were stamped with that frequency, the ones which were not stable were marked with a downgraded frequency. This reduced waste at a time CPUs were expensive to make.
In most cases, overclocking will overheat the CPU. More thoughts about overclocking at the bottom of this post.
2. I made a mistake in my previous post.
This is correct:
- the 65C02 needs 4 cycles to read/write from RAM/ROM, and the process of read/write somewhat overlaps and is done during the last 2 cycles.
This is wrong:
- the 65C02 CPU needs 4 cycles before it even begins to read/write.
However what I said that the RAM memory was fast enough for that particular case, and that the CPU was a bottleneck still stands.
The 65C02 spends cycle 1 and 2 to initiate and stabilize the signals on its pins, then during approx cycles 3 and 4 performs the read/write from memory.
65C02 at 15 MHz
one cycle = 66 ns
cycle 1 + cycle 2 = 132 ns ; stabilize signals
cycle 3 + cycle 4 = 132 ns ; still slower than 100 ns access time of RAM
RAM is faster than CPU, and is appropiate for this set up
65C02 at 20 MHz
one cycle = 50 ns
cycle 1 + cycle 2 = 100 ns ; stabilize signals
cycle 3 + cycle 4 = 100 ns ; equal with 100 ns access time of RAM
RAM and CPU have the same speed, it's still ok to use the 100 ns RAM chips, but I would not go above 20 MHz until I install faster RAM.
3. Theoretically, yes you can. In practice, I would not overclock ANY type of CPU more than twice the frequency stamped on it. I'd rather buy a faster CPU.
Overclocking
1. Overheating
- depends on the CPU ; some CPUs can take more heat than others because of construction (capsule size, capsule material, capsule shape, etc)
- depends on cooling ; having a fan and cooling vents helps, also a heatsink if there is room in the enclosure; the location of CPU inside the enclosure.
- depends where you live (geographically) ; cooler or hotter local weather
- depends on the room temperature where you play ; having a fan or air conditioning helps
- how long you had it powered on
2. Gain here, lose there
- risk of CPU failure (some CPUs are expensive, others are hard to find, some harder to replace on the circuit board)
- messed up time controls (if the chess software is not modified to run at that overclocking frequency)
- messed up display and sometimes button response (if the whole chess computer takes frequency from one single master crystal)
- ELO gain barely noticeable compared to the risk of damage
I want to elaborate on the 'ELO gain' because this is by far the reason why chess computers are overclocked.
How a chess engine works, I mean the software/algorithm.
For a long time (1970s and part of 80s), the chess moves were computed using the alpha-beta algorithm. It was estimated that for going an extra ply deeper (and about 200 extra ELO points) one would have to run the same chess engine on a 5-6 times faster computer.
Then came along improved algorithms such as: Negascout and PVS. They had about 20 percent speed gain over alpha-beta, so you would still need an approx 4-5 times faster CPU for gaining an extra ply.
Then the Null Move algorithm was invented in the beginning of 1990s, and that saved another 20-50 percent of the time used for computing the next move.
There are several algorithms for move search and tree pruning, and combinations of them, which I have not mentioned, in order to shorten this example.
Anyway, I think that today we need a computer 3 times faster for an extra ply.
Mephisto Polgar was made when none of the faster algorithms were around, so I would assume you'd need to overclock the CPU about 5 times faster to gain some significant ELO increase.
You would likely toast the CPU by doing that.
Most of the time used in calculating the next move is spent during the last ply, due to the search tree expanding exponentially.
For example, if POLGAR reaches 5 ply in 20 seconds, from that moment it will need an extra 100-120 seconds to advance to the 6th ply. Do you notice that it would take way more time for the last ply than it took for all last 5 plies taken together ?
Another example, this time on the good side, let's assume that a chess computer is set to 30 sec/move. Around the time it reaches 30 seconds, it was calculating the moves between the 5th and 6th ply, and when time ran out, since it did not reach a full 6 ply, it had to make the best move found at 5 ply. Let's say that the computer calculated 5.87 plies (there are no such things as fractional plies displayed by any chess engine, but there exists a 'fractional ply' theory in chess programming).
It might be possible that by overclocking only 2 times would provide the extra speed to make it to the next full 6 ply, so the computer makes a deeper search move. That's where I see the potential usefulness in overclocking. This has to be experimented with because it depends a lot on the specific chess engines and time controls.
Whoever knows more about chess engine programming, please feel free to add to this, because I'm far from being an expert in anything. I could have asked these questions in the TALK CHESS forum, but I did not want to be laughed at.
In conclusion, before overclocking chess computers, a lot of factors have to be taken into account.
I myself would rather buy a faster CPU.
Either way, overclocking or a new CPU, the limitations of the chess engines killed my interest in doing it this way.
The only way I'd see it worth doing, is redesigning the electronic schematic of module and using a CPU 100 times faster. Probably gaining 400 ELO points. Now that's what I call overclocking.
Blisters on my fingers (from typing, not from overclocking 3rd degree burns) regards,
Kostea
What I said below is not an attempt to discourage anyone from overclocking.
Just some things to think about.
Before I attempt to answer your questions, let me clarify one thing. The CPU in the POLGAR reads a lot from ROM too (where the chess engine is stored). The ROM is generally considered slower than the RAM, but for a correct answer you have to compare the ROM and RAM chip IDs in your particular case and find out their speeds.
Besides the RAM, you have to have a fast ROM too. The sticker on the upgrade ROM that says '10 MHz', and that only means that the chess software has been modified to run at 10 MHz.
The ROM speed is stamped on its capsule by the manufacturer.
One thing that others have mentioned before. The Saitek RISC 2500 (and Montreux) load their chess engine from ROM to RAM and then execute the software from RAM. In that particular chess computer, the RAM is 2 times faster than the ROM.
1. That's a good question. A CPU built and stamped as a 10 MHz must reliably work at that speed. The problem here is that the seller only provides a ROM and a quartz crystal. No new CPU, so the old CPU is still a 5 MHz.
Some manufacturers made CPUs at the highest frequency the technology permitted, then after testing, the ones running fine were stamped with that frequency, the ones which were not stable were marked with a downgraded frequency. This reduced waste at a time CPUs were expensive to make.
In most cases, overclocking will overheat the CPU. More thoughts about overclocking at the bottom of this post.
2. I made a mistake in my previous post.
This is correct:
- the 65C02 needs 4 cycles to read/write from RAM/ROM, and the process of read/write somewhat overlaps and is done during the last 2 cycles.
This is wrong:
- the 65C02 CPU needs 4 cycles before it even begins to read/write.
However what I said that the RAM memory was fast enough for that particular case, and that the CPU was a bottleneck still stands.
The 65C02 spends cycle 1 and 2 to initiate and stabilize the signals on its pins, then during approx cycles 3 and 4 performs the read/write from memory.
65C02 at 15 MHz
one cycle = 66 ns
cycle 1 + cycle 2 = 132 ns ; stabilize signals
cycle 3 + cycle 4 = 132 ns ; still slower than 100 ns access time of RAM
RAM is faster than CPU, and is appropiate for this set up
65C02 at 20 MHz
one cycle = 50 ns
cycle 1 + cycle 2 = 100 ns ; stabilize signals
cycle 3 + cycle 4 = 100 ns ; equal with 100 ns access time of RAM
RAM and CPU have the same speed, it's still ok to use the 100 ns RAM chips, but I would not go above 20 MHz until I install faster RAM.
3. Theoretically, yes you can. In practice, I would not overclock ANY type of CPU more than twice the frequency stamped on it. I'd rather buy a faster CPU.
Overclocking
1. Overheating
- depends on the CPU ; some CPUs can take more heat than others because of construction (capsule size, capsule material, capsule shape, etc)
- depends on cooling ; having a fan and cooling vents helps, also a heatsink if there is room in the enclosure; the location of CPU inside the enclosure.
- depends where you live (geographically) ; cooler or hotter local weather
- depends on the room temperature where you play ; having a fan or air conditioning helps
- how long you had it powered on
2. Gain here, lose there
- risk of CPU failure (some CPUs are expensive, others are hard to find, some harder to replace on the circuit board)
- messed up time controls (if the chess software is not modified to run at that overclocking frequency)
- messed up display and sometimes button response (if the whole chess computer takes frequency from one single master crystal)
- ELO gain barely noticeable compared to the risk of damage
I want to elaborate on the 'ELO gain' because this is by far the reason why chess computers are overclocked.
How a chess engine works, I mean the software/algorithm.
For a long time (1970s and part of 80s), the chess moves were computed using the alpha-beta algorithm. It was estimated that for going an extra ply deeper (and about 200 extra ELO points) one would have to run the same chess engine on a 5-6 times faster computer.
Then came along improved algorithms such as: Negascout and PVS. They had about 20 percent speed gain over alpha-beta, so you would still need an approx 4-5 times faster CPU for gaining an extra ply.
Then the Null Move algorithm was invented in the beginning of 1990s, and that saved another 20-50 percent of the time used for computing the next move.
There are several algorithms for move search and tree pruning, and combinations of them, which I have not mentioned, in order to shorten this example.
Anyway, I think that today we need a computer 3 times faster for an extra ply.
Mephisto Polgar was made when none of the faster algorithms were around, so I would assume you'd need to overclock the CPU about 5 times faster to gain some significant ELO increase.
You would likely toast the CPU by doing that.
Most of the time used in calculating the next move is spent during the last ply, due to the search tree expanding exponentially.
For example, if POLGAR reaches 5 ply in 20 seconds, from that moment it will need an extra 100-120 seconds to advance to the 6th ply. Do you notice that it would take way more time for the last ply than it took for all last 5 plies taken together ?
Another example, this time on the good side, let's assume that a chess computer is set to 30 sec/move. Around the time it reaches 30 seconds, it was calculating the moves between the 5th and 6th ply, and when time ran out, since it did not reach a full 6 ply, it had to make the best move found at 5 ply. Let's say that the computer calculated 5.87 plies (there are no such things as fractional plies displayed by any chess engine, but there exists a 'fractional ply' theory in chess programming).
It might be possible that by overclocking only 2 times would provide the extra speed to make it to the next full 6 ply, so the computer makes a deeper search move. That's where I see the potential usefulness in overclocking. This has to be experimented with because it depends a lot on the specific chess engines and time controls.
Whoever knows more about chess engine programming, please feel free to add to this, because I'm far from being an expert in anything. I could have asked these questions in the TALK CHESS forum, but I did not want to be laughed at.
In conclusion, before overclocking chess computers, a lot of factors have to be taken into account.
I myself would rather buy a faster CPU.
Either way, overclocking or a new CPU, the limitations of the chess engines killed my interest in doing it this way.
The only way I'd see it worth doing, is redesigning the electronic schematic of module and using a CPU 100 times faster. Probably gaining 400 ELO points. Now that's what I call overclocking.
Blisters on my fingers (from typing, not from overclocking 3rd degree burns) regards,
Kostea
Thanks Kostea for your explanation.
The fact is that a 10 Mhz polgar was really released and i have the software for it loaded in a new and faster eprom. I also buy a new microprocesor (WDC 65C02S) and a faster RAM. But i believe that i only have to change the EPROM and the 5 Mhz crystal for the 10 Mhz. With that i will add a new machine to my collection.
The 10 Mhz software not only fixes the mess up time control but also has some improve on endings. This machine has almost 100 elo points higher than the 5 Mhz version; in human chess playing this could be the difference between an expert and a master and that is a lot.
I saw the datasheet of the 65c02 for 10 Mhz and it is still ok with the temperature.
If everything goes ok i will post some games here with a photo.
testing regards
Ricardo
The fact is that a 10 Mhz polgar was really released and i have the software for it loaded in a new and faster eprom. I also buy a new microprocesor (WDC 65C02S) and a faster RAM. But i believe that i only have to change the EPROM and the 5 Mhz crystal for the 10 Mhz. With that i will add a new machine to my collection.
The 10 Mhz software not only fixes the mess up time control but also has some improve on endings. This machine has almost 100 elo points higher than the 5 Mhz version; in human chess playing this could be the difference between an expert and a master and that is a lot.
I saw the datasheet of the 65c02 for 10 Mhz and it is still ok with the temperature.
If everything goes ok i will post some games here with a photo.
testing regards
Ricardo
-
AMD64inside
- Member
- Posts: 20
- Joined: Tue Jul 31, 2007 7:31 pm
Unbelieveable
The Lyon 32Bit along with Vancouver and London upgrade roms is once again relisted after an apparently successful sale.
http://cgi.ebay.co.uk/Chess-computer-Me ... 1e5f38922e
Different pictures is seems but must be the same unit.
Third Time Unlucky Regards
Darrell
The Lyon 32Bit along with Vancouver and London upgrade roms is once again relisted after an apparently successful sale.
http://cgi.ebay.co.uk/Chess-computer-Me ... 1e5f38922e
Different pictures is seems but must be the same unit.
Third Time Unlucky Regards
Darrell
"The game itself is nothing, the playing of it is everything"