Sesse wrote:
These things tend to be exponential. In particular, you'd need a machine with a lot more RAM than 1TB.
I thought roughly speaking you need 60 times more hard disk and ram for each extra piece.
I’m sure the factor should be much smaller than 60. Long time ago someone has already predicted that building Syzygy 7 men would take 1TB Ram and 3-5 years. Actually it took not that RAM size and only 5 months!
I don't know where you got your numbers from, but it did take two machines with >= 1 TB RAM to generate them. Generating 8-piece tables with the same algorithm would require a machine with 64 TB RAM.
7 men is about 17 times larger than 6 men. Thus I guess the factor for 8 men is reasonably about 10-25. Magic may happens again
Your numbers are off. 150Gb vs 17TB is not a factor of 17.
syzygy wrote: ↑Tue Aug 21, 2018 4:14 pm
Generating 8-piece tables with the same algorithm would require a machine with 64 TB RAM.
There exist algorithms that use 1-bit per position in RAM for random access and that do sequential disk I/O to complement it. That would lower the requirement to 1/8 * 462 * 62! / 56! = 2.55 Tb of RAM. Will be much slower of course.
whereagles wrote: ↑Tue Aug 21, 2018 4:41 pm
One such machine would be difficult to get. Could that be divided into a cluster?
Yes, it can. One of the very first tablebase calculations was, as far as I know, done on a cluster, through some clever restructuring of the algorithm through group theory. I remember trying to follow the mathematics at some point, and even though I do know some abstract algebra, it was over my head
Sesse wrote:
These things tend to be exponential. In particular, you'd need a machine with a lot more RAM than 1TB.
I thought roughly speaking you need 60 times more hard disk and ram for each extra piece.
I’m sure the factor should be much smaller than 60. Long time ago someone has already predicted that building Syzygy 7 men would take 1TB Ram and 3-5 years. Actually it took not that RAM size and only 5 months!
I don't know where you got your numbers from, but it did take two machines with >= 1 TB RAM to generate them. Generating 8-piece tables with the same algorithm would require a machine with 64 TB RAM.
7 men is about 17 times larger than 6 men. Thus I guess the factor for 8 men is reasonably about 10-25. Magic may happens again
Your numbers are off. 150Gb vs 17TB is not a factor of 17.
"EGTB
6 man tablebase syzygy {EGTB} 90 GB
7 man tablebase The high speed of generating the tablebases was because of using a supercomputer named Lomonosov (top500). The size of seven-man tablebases is about 140 TB.
Many show interest in what is to expect from 8-man endings. First, take note that the longest 6-man mate took 262 moves (KRN-KNN). Moving to 7-man endings doubled this value. Second, 8-man tablebases include much more endings with both sides having relatively equal strength. All this gives us a strong hope to discover a mate in more than 1000 moves in one of 8-man endgames. Unfortunately the size of 8-man tablebases will be 100 times larger than the size of 7-man tablebases. To fully compute them, one will need about 10 PB (10,000 TB) of disk space and 50 TB of RAM. Only the top 10 supercomputers can solve the 8-man problem in 2014. The first 1000-move mate is unlikely to be found until 2020 when a part of a TOP100 supercomputer may be allowed to be used for solving this task."
The fathom code that read up to 6 man syzygy tables will continue to work and simply ignore 7 man tables? Or it will miserably crash if used in a 7 man configuration?
elcabesa wrote: ↑Fri Aug 24, 2018 6:59 pm
The fathom code that read up to 6 man syzygy tables will continue to work and simply ignore 7 man tables? Or it will miserably crash if used in a 7 man configuration?
Even Jon Dart's updated fathom code is two years old, so I guess the chance of being able to probe the 7 man files is close to zero with fathom.
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.
