7-men Syzygy attempt

[noobpwnftw]

Also this, following the line result in a repetition.

https://syzygy-tables.info/?fen=1N4n1/3 ... _w_-_-_0_1

I guess it supposed to be probed differently as this table was in 16-bit format before compression?

[noobpwnftw]

Looks like it takes one less step of symbol conversion in

Code: Select all

#define read_symbol(x) _Generic((x), \
  u8: symcode[x][(&(x))[1]], \
  u16: x \
);

The decompression part probably should do the same although it hasn't been updated.

[syzygy]

The decompressor is more or less OK, but I forgot to store the map[] array as 16-bit values...

The DTZ flags also need a flag indicating whether map[] stores 8- or 16-bit values.

[duncan]

That does look like a deal, however, mine had serious memory bus congestion, not sure how would it go with less than a half of memory clock, you might want a test run if that's possible.

any machine you could recommend ?

[syzygy]

The decompressor is more or less OK, but I forgot to store the map[] array as 16-bit values...

The DTZ flags also need a flag indicating whether map[] stores 8- or 16-bit values.

This should fix the problem:
https://github.com/syzygy1/tb/commit/21 ... 09e0cc516e
Not fully implemented for pawnful tables, since the pawnful generator anyway cannot yet compress DTZ tables with max DTZ >= 290 moves (and such 7-men tables may not exist).

This needs to be changed in the probing code:
https://github.com/syzygy1/Cfish/commit ... cd62f8afeb

This is still untested. I will generate some pawnless test tables while forcing the use of a 16-bit map[] array and see if it works.

[syzygy]

KRBNvKQ.rtbw should be fine.
Unfortunately, KRBNvKQ.rtbz will have to be regenerated (-z).

[Dann Corbit]

That does look like a deal, however, mine had serious memory bus congestion, not sure how would it go with less than a half of memory clock, you might want a test run if that's possible.

any machine you could recommend ?

I guess that the one he is using costs at least $40,000, and probably more.

[Dann Corbit]

KRBNvKQ.rtbw should be fine.
Unfortunately, KRBNvKQ.rtbz will have to be regenerated (-z).

Are the statistics the program generated for that file correct?

[syzygy]

KRBNvKQ.rtbw should be fine.
Unfortunately, KRBNvKQ.rtbz will have to be regenerated (-z).

I have now tested the change and it seems to work well.

[syzygy]

KRBNvKQ.rtbw should be fine.
Unfortunately, KRBNvKQ.rtbz will have to be regenerated (-z).

Are the statistics the program generated for that file correct?

Yes.

(Or at least, this change won't change the statistics )