What is Hyatt Hash

[kgburcham]

FireBird 1.0 beta (for Windows) is UCI only, fast, and stable.

rating:
~ 3400 ELO

features:
bitboards
true SMP parallel search
configurable pawn hash
analysis mode
ponder
searchmoves
multiPV (in beta)
multiCentiPawnPV
Hyatt Hash
perft, benchmark, and drawboard utilites
validate functions (debug)
multi pos gain (for each CPU/core/thread)
multi history (for each CPU/core/thread)

most recent improvements:
native Windows SMP threads/multiprocessing
code cleanup and formatting
Max Threads UCI option
get_sys_info function:
queries Windows for # CPUs,
sets appropriate # threads
searchmoves
pondering improved
several bugfixes
SMP code vastly improved
auto CPU (Max Thread) detect/config
UCI Max Thread configuration option
improved prefetch routine
improved bitscan routines
sparse one POPCNT
note: because of licensing issues, mersenne twist is no longer utilized and has been removed

FireBird is based on the IppoLit family of engines, more info: http://ippolit.wikispaces.com

this chess ship is a truly glorious achievement of the October Revolution and Decembrists movement!

[benstoker]

FireBird 1.0 beta (for Windows) is UCI only, fast, and stable.

rating:
~ 3400 ELO

features:
bitboards
true SMP parallel search
configurable pawn hash
analysis mode
ponder
searchmoves
multiPV (in beta)
multiCentiPawnPV
Hyatt Hash
perft, benchmark, and drawboard utilites
validate functions (debug)
multi pos gain (for each CPU/core/thread)
multi history (for each CPU/core/thread)

most recent improvements:
native Windows SMP threads/multiprocessing
code cleanup and formatting
Max Threads UCI option
get_sys_info function:
queries Windows for # CPUs,
sets appropriate # threads
searchmoves
pondering improved
several bugfixes
SMP code vastly improved
auto CPU (Max Thread) detect/config
UCI Max Thread configuration option
improved prefetch routine
improved bitscan routines
sparse one POPCNT
note: because of licensing issues, mersenne twist is no longer utilized and has been removed

FireBird is based on the IppoLit family of engines, more info: http://ippolit.wikispaces.com

this chess ship is a truly glorious achievement of the October Revolution and Decembrists movement!

I think the ippoliterati declared they were implementing some code into the robbo from Dr. Hyatt's crafty.

I am sure Dr. Hyatt is tickled to death about that.

The virus is spreading ...

Seems to me somone out there needs to step up to the plate and do something about this. This code is out there. If someone owns it, step up to the plate, say so, make your case, and STOP THE CLOWNS NOW!

[bob]

FireBird 1.0 beta (for Windows) is UCI only, fast, and stable.

rating:
~ 3400 ELO

features:
bitboards
true SMP parallel search
configurable pawn hash
analysis mode
ponder
searchmoves
multiPV (in beta)
multiCentiPawnPV
Hyatt Hash
perft, benchmark, and drawboard utilites
validate functions (debug)
multi pos gain (for each CPU/core/thread)
multi history (for each CPU/core/thread)

most recent improvements:
native Windows SMP threads/multiprocessing
code cleanup and formatting
Max Threads UCI option
get_sys_info function:
queries Windows for # CPUs,
sets appropriate # threads
searchmoves
pondering improved
several bugfixes
SMP code vastly improved
auto CPU (Max Thread) detect/config
UCI Max Thread configuration option
improved prefetch routine
improved bitscan routines
sparse one POPCNT
note: because of licensing issues, mersenne twist is no longer utilized and has been removed

FireBird is based on the IppoLit family of engines, more info: http://ippolit.wikispaces.com

this chess ship is a truly glorious achievement of the October Revolution and Decembrists movement!

I think the ippoliterati declared they were implementing some code into the robbo from Dr. Hyatt's crafty.

I am sure Dr. Hyatt is tickled to death about that.

The virus is spreading ...

Seems to me somone out there needs to step up to the plate and do something about this. This code is out there. If someone owns it, step up to the plate, say so, make your case, and STOP THE CLOWNS NOW!

No idea what that means. I suppose they might like the recent hash probe changes in Crafty and use that, but no idea. And the genie is out of the bottle, so worrying about who is copying what is pretty much pointless...

[Graham Banks]

so worrying about who is copying what is pretty much pointless...

I think that as a community, we should surely be condemning it, even if you believe that nothing can be done about it?

Cheers,
Graham.

[Mike S.]

Mentioning "Hyatt Hash" does not mean that they did copy and paste code. I am confident that you wouldn't want to condemn legal use of ideas, or of public concepts for techniques. I think one motivation behind open source software is to encourage that.

For me, the critical point is if code was "taken directly" by copy & paste or equivalent methods. If so, then there could be something to condemn. But as long as that has not been done (or at least not been proven), I'd say a so called clone can be treated as being legal.

[Graham Banks]

Mentioning "Hyatt Hash" does not mean that they did copy and paste code. I am confident that you wouldn't want to condemn legal use of ideas, or of public concepts for techniques. I think one motivation behind open source software is to encourage that.

For me, the critical point is if code was "taken directly" by copy & paste or equivalent methods. If so, then there could be something to condemn. But as long as that has not been done (or at least not been proven), I'd say a so called clone can be treated as being legal.

I agree in essence. I have nothing against open source development that follows the accepted rules of compliance.
In fact, despite what people may think of Norman Schmidt, I still test Cyclone because it is a GPL derivative.

Cheers,
Graham.

[bob]

so worrying about who is copying what is pretty much pointless...

I think that as a community, we should surely be condemning it, even if you believe that nothing can be done about it?

Cheers,
Graham.

I don't have any problem with someone copying the hash code from Crafty. Any more than I object to them copying "printf()" from the standard C I/O library. Hashing is not what makes Crafty "crafty". It's the search and evaluation. Not the bitboards, nor input/output, etc.

[bob]

Mentioning "Hyatt Hash" does not mean that they did copy and paste code. I am confident that you wouldn't want to condemn legal use of ideas, or of public concepts for techniques. I think one motivation behind open source software is to encourage that.

For me, the critical point is if code was "taken directly" by copy & paste or equivalent methods. If so, then there could be something to condemn. But as long as that has not been done (or at least not been proven), I'd say a so called clone can be treated as being legal.

Several things I do have been somewhat immortalized for no good reason. "delta pruning" is one idea, when it is nothing new or exciting and has been around for many years. I try to explain ideas that influence performance (my newest hashing code was directly addressed at performance (speed)) so that anyone can use the idea to produce a more efficient program.

[benstoker]

so worrying about who is copying what is pretty much pointless...

I think that as a community, we should surely be condemning it, even if you believe that nothing can be done about it?

Cheers,
Graham.

I don't have any problem with someone copying the hash code from Crafty. Any more than I object to them copying "printf()" from the standard C I/O library. Hashing is not what makes Crafty "crafty". It's the search and evaluation. Not the bitboards, nor input/output, etc.

In any case, here's HYATT_HASH

Code: Select all

00132 #if 0 /* compiler */
00133 #define HYATT_HASH()   asm volatile ("movq (%0),%%r15\n" "movq %%r15,(%1)\n"            "movq 0x8(%0),%%r15\n" "movq %%r15,0x8(%1)\n"           /* "xorq %%r15,(%1)\n" "xorq %%r15,(%1)\n" */           "movq %1,%0\n"          : : "r" (tr), "r" (H) : "%r15");
00134 #define UPDATE_ANNI() (HashTable + (k + i))->age = AGE;
00135 #if 0
00136 #define HYATT_HASH()   H->h1 = ((uint64 *) tr)[0];   H->h2 = ((uint64 *) tr)[1];   H->h1 ^= H->h2;   H->h1 ^= H->h2;   tr = (typeHash *) (H);
00137 #endif
00138 #else /* non funzione ! */
00139 #define HYATT_HASH() /* */
00140 #define UPDATE_ANNI() tr->age = AGE;
00141 #endif

and it's referenced by:

Code: Select all

Information(), MyAll(), MyAllCheck(), MyCut(), MyCutCheck(), MyExclude(), MyExcludeCheck(), MyLowDepth(), MyLowDepthCheck(), MyPV(), MyPVQsearch(), MyPVQsearchCheck(), MyQsearch(), MyQsearchCheck(), MyTop(), MyTopAnalysis(), and OutputBestMove().

[benstoker]

so worrying about who is copying what is pretty much pointless...

I think that as a community, we should surely be condemning it, even if you believe that nothing can be done about it?

Cheers,
Graham.

I don't have any problem with someone copying the hash code from Crafty. Any more than I object to them copying "printf()" from the standard C I/O library. Hashing is not what makes Crafty "crafty". It's the search and evaluation. Not the bitboards, nor input/output, etc.

In any case, here's HYATT_HASH

Code: Select all

00132 #if 0 /* compiler */
00133 #define HYATT_HASH()   asm volatile ("movq (%0),%%r15\n" "movq %%r15,(%1)\n"            "movq 0x8(%0),%%r15\n" "movq %%r15,0x8(%1)\n"           /* "xorq %%r15,(%1)\n" "xorq %%r15,(%1)\n" */           "movq %1,%0\n"          : : "r" (tr), "r" (H) : "%r15");
00134 #define UPDATE_ANNI() (HashTable + (k + i))->age = AGE;
00135 #if 0
00136 #define HYATT_HASH()   H->h1 = ((uint64 *) tr)[0];   H->h2 = ((uint64 *) tr)[1];   H->h1 ^= H->h2;   H->h1 ^= H->h2;   tr = (typeHash *) (H);
00137 #endif
00138 #else /* non funzione ! */
00139 #define HYATT_HASH() /* */
00140 #define UPDATE_ANNI() tr->age = AGE;
00141 #endif

and it's referenced by:

Code: Select all

Information(), MyAll(), MyAllCheck(), MyCut(), MyCutCheck(), MyExclude(), MyExcludeCheck(), MyLowDepth(), MyLowDepthCheck(), MyPV(), MyPVQsearch(), MyPVQsearchCheck(), MyQsearch(), MyQsearchCheck(), MyTop(), MyTopAnalysis(), and OutputBestMove().