h48

A prototype for an optimal Rubik's cube solver, work in progress.
git clone https://git.tronto.net/h48
Download | Log | Files | Refs | README | LICENSE

commit 08efbd62ef956327924aaea1fcfa97fde4dc1678
parent d65ce8e86b517ef3f7d48cf5ded1101444e05b2b
Author: Sebastiano Tronto <sebastiano@tronto.net>
Date:   Wed,  1 Nov 2023 19:26:09 +0100

Reworked moves and transformations, moved some stuff

Diffstat:
MREADME.md | 44+++++++++++---------------------------------
Rsrc/_trans_move_arr.c -> old/_trans_move_arr.c | 0
Aold/inverses.c | 71+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Rexperiments/cube.c -> old/manualsimd/cube.c | 0
Rexperiments/cube.h -> old/manualsimd/cube.h | 0
Dsrc/_constants.c | 286-------------------------------------------------------------------------------
Dsrc/_move_logic_arr.c | 149-------------------------------------------------------------------------------
Dsrc/_move_logic_avx2.c | 6------
Asrc/_moves_arr.c | 264+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Asrc/_moves_avx2.c | 2++
Asrc/_trans_arr.c | 952+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Asrc/_trans_avx2.c | 0
Dsrc/_trans_move_avx2.c | 2--
Msrc/cube.c | 438+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++------------
Msrc/cube.h | 3---
Mutils/gentranscode.sh | 47+++++++++++++++++++++++++++++++++++++----------
Autils/gentransswitch.sh | 8++++++++
17 files changed, 1720 insertions(+), 552 deletions(-)

diff --git a/README.md b/README.md @@ -13,51 +13,29 @@ $ make test ## TODO: -### Tests +### Make AVX2 work -* inverse_move -* inverse_trans - -### Simple additions - -* Write to AVX2-src format -* move() that takes a string (alg) as input - -### Changes - -* write one function (static inline or public?) for each move (and trans) -* only call the specific functions in performance-critical steps (i.e. solve - - if the functions are static inline, all performance-critical steps must - be internal to cube.c) -* this also enables skipping some unnecessary work (e.g. flip edges, twist - corners) for many moves (and mirror transformations) -* add benchmarks with moves / trans called directly instead of via the - generic function -* keep generic move and trans functions with big switch case -* bring back constants into cube.c, and maybe also moves (TBD: what to - do with architecture-specific code? leave in separate files like now, - use just one file for each architecture...) +* writecube to AVX2-src format (+ tests) +* generate moves and transformations with scripts in utils/ +* fix base get_ and set_ macros (constant arguments?) +* optimize things that use get_ and set_ ### Documentation and interface * inline some documentation as comments in cube.h or cube.c * README.md (maybe convert to txt?) becomes the reference documentation -### AVX2 - -* fix base get_ and set_ macros (constant arguments?) -* implement missing stuff (moves, transform) -* optimize things that use get_ and set_ - ### More features +* move() that takes a string (alg) as input * coordinates: co, eo, epsep, cpsep_sym, cocpsep_sym, cphtr_sym, cocphtr_sym -* pruning tables (1 bit per entry + fallback) -* solve.c -### Optimizations: +### Solving -* multi-move (up to 4/5 moves at once) +* Fixed depth +* pruning tables (1 bit per entry + fallback) +* Takes as parameters the amount of memory to use and a FILE for the tables +* Use multi-move (up to 4/5 moves at once) ### Things I need to learn: diff --git a/src/_trans_move_arr.c b/old/_trans_move_arr.c diff --git a/old/inverses.c b/old/inverses.c @@ -0,0 +1,71 @@ +static move_t inverse_move_arr[] = { + [U] = U3, + [U2] = U2, + [U3] = U, + [D] = D3, + [D2] = D2, + [D3] = D, + [R] = R3, + [R2] = R2, + [R3] = R, + [L] = L3, + [L2] = L2, + [L3] = L, + [F] = F3, + [F2] = F2, + [F3] = F, + [B] = B3, + [B2] = B2, + [B3] = B, +}; + +static trans_t inverse_trans_arr[] = { + [UFr] = UFr, + [ULr] = URr, + [UBr] = UBr, + [URr] = URr, + [DFr] = DFr, + [DLr] = DLr, + [DRr] = DRr, + [DBr] = DBr, + [RUr] = FRr, + [RFr] = LFr, + [RDr] = BLr, + [RBr] = RBr, + [LUr] = FLr, + [LFr] = RFr, + [LDr] = BRr, + [LBr] = LBr, + [FUr] = FUr, + [FRr] = RUr, + [FDr] = BUr, + [FLr] = LUr, + [BUr] = FDr, + [BLr] = RDr, + [BDr] = BDr, + [BRr] = LDr, + [UFm] = UFm, + [ULm] = ULm, + [UBm] = UBm, + [URm] = ULm, + [DFm] = DFm, + [DLm] = DRm, + [DRm] = DLm, + [DBm] = DBm, + [RUm] = FLm, + [RFm] = RFm, + [RDm] = BRm, + [RBm] = LBm, + [LUm] = FRm, + [LFm] = LFm, + [LDm] = BLm, + [LBm] = RBm, + [FUm] = FUm, + [FRm] = LUm, + [FDm] = BUm, + [FLm] = RUm, + [BUm] = FDm, + [BLm] = LDm, + [BDm] = BDm, + [BRm] = RDm, +}; diff --git a/experiments/cube.c b/old/manualsimd/cube.c diff --git a/experiments/cube.h b/old/manualsimd/cube.h diff --git a/src/_constants.c b/src/_constants.c @@ -1,286 +0,0 @@ -#define U 0U -#define U2 1U -#define U3 2U -#define D 3U -#define D2 4U -#define D3 5U -#define R 6U -#define R2 7U -#define R3 8U -#define L 9U -#define L2 10U -#define L3 11U -#define F 12U -#define F2 13U -#define F3 14U -#define B 15U -#define B2 16U -#define B3 17U - -#define UFr 0 -#define ULr 1 -#define UBr 2 -#define URr 3 -#define DFr 4 -#define DLr 5 -#define DBr 6 -#define DRr 7 -#define RUr 8 -#define RFr 9 -#define RDr 10 -#define RBr 11 -#define LUr 12 -#define LFr 13 -#define LDr 14 -#define LBr 15 -#define FUr 16 -#define FRr 17 -#define FDr 18 -#define FLr 19 -#define BUr 20 -#define BRr 21 -#define BDr 22 -#define BLr 23 - -#define UFm 24 -#define ULm 25 -#define UBm 26 -#define URm 27 -#define DFm 28 -#define DLm 29 -#define DBm 30 -#define DRm 31 -#define RUm 32 -#define RFm 33 -#define RDm 34 -#define RBm 35 -#define LUm 36 -#define LFm 37 -#define LDm 38 -#define LBm 39 -#define FUm 40 -#define FRm 41 -#define FDm 42 -#define FLm 43 -#define BUm 44 -#define BRm 45 -#define BDm 46 -#define BLm 47 - -#define errormove 99U -#define errortrans 99U - -#define _c_ufr 0U -#define _c_ubl 1U -#define _c_dfl 2U -#define _c_dbr 3U -#define _c_ufl 4U -#define _c_ubr 5U -#define _c_dfr 6U -#define _c_dbl 7U - -#define _e_uf 0U -#define _e_ub 1U -#define _e_db 2U -#define _e_df 3U -#define _e_ur 4U -#define _e_ul 5U -#define _e_dl 6U -#define _e_dr 7U -#define _e_fr 8U -#define _e_fl 9U -#define _e_bl 10U -#define _e_br 11U - -#define _eoshift 4U -#define _coshift 5U - -#define _pbits 0xFU -#define _eobit 0x10U -#define _cobits 0xF0U -#define _cobits2 0x60U -#define _ctwist_cw 0x20U -#define _ctwist_ccw 0x40U -#define _eflip 0x10U -#define _error 0xFFU - -static move_t inverse_move_arr[] = { - [U] = U3, - [U2] = U2, - [U3] = U, - [D] = D3, - [D2] = D2, - [D3] = D, - [R] = R3, - [R2] = R2, - [R3] = R, - [L] = L3, - [L2] = L2, - [L3] = L, - [F] = F3, - [F2] = F2, - [F3] = F, - [B] = B3, - [B2] = B2, - [B3] = B, -}; - -static trans_t inverse_trans_arr[] = { - [UFr] = UFr, - [ULr] = URr, - [UBr] = UBr, - [URr] = URr, - [DFr] = DFr, - [DLr] = DLr, - [DRr] = DRr, - [DBr] = DBr, - [RUr] = FRr, - [RFr] = LFr, - [RDr] = BLr, - [RBr] = RBr, - [LUr] = FLr, - [LFr] = RFr, - [LDr] = BRr, - [LBr] = LBr, - [FUr] = FUr, - [FRr] = RUr, - [FDr] = BUr, - [FLr] = LUr, - [BUr] = FDr, - [BLr] = RDr, - [BDr] = BDr, - [BRr] = LDr, - [UFm] = UFm, - [ULm] = ULm, - [UBm] = UBm, - [URm] = ULm, - [DFm] = DFm, - [DLm] = DRm, - [DRm] = DLm, - [DBm] = DBm, - [RUm] = FLm, - [RFm] = RFm, - [RDm] = BRm, - [RBm] = LBm, - [LUm] = FRm, - [LFm] = LFm, - [LDm] = BLm, - [LBm] = RBm, - [FUm] = FUm, - [FRm] = LUm, - [FDm] = BUm, - [FLm] = RUm, - [BUm] = FDm, - [BLm] = LDm, - [BDm] = BDm, - [BRm] = RDm, -}; - -static char *cornerstr[] = { - [_c_ufr] = "UFR", - [_c_ubl] = "UBL", - [_c_dfl] = "DFL", - [_c_dbr] = "DBR", - [_c_ufl] = "UFL", - [_c_ubr] = "UBR", - [_c_dfr] = "DFR", - [_c_dbl] = "DBL" -}; - -static char *cornerstralt[] = { - [_c_ufr] = "URF", - [_c_ubl] = "ULB", - [_c_dfl] = "DLF", - [_c_dbr] = "DRB", - [_c_ufl] = "ULF", - [_c_ubr] = "URB", - [_c_dfr] = "DRF", - [_c_dbl] = "DLB" -}; - -static char *edgestr[] = { - [_e_uf] = "UF", - [_e_ub] = "UB", - [_e_db] = "DB", - [_e_df] = "DF", - [_e_ur] = "UR", - [_e_ul] = "UL", - [_e_dl] = "DL", - [_e_dr] = "DR", - [_e_fr] = "FR", - [_e_fl] = "FL", - [_e_bl] = "BL", - [_e_br] = "BR" -}; - -static char *movestr[] = { - [U] = "U", - [U2] = "U2", - [U3] = "U'", - [D] = "D", - [D2] = "D2", - [D3] = "D'", - [R] = "R", - [R2] = "R2", - [R3] = "R'", - [L] = "L", - [L2] = "L2", - [L3] = "L'", - [F] = "F", - [F2] = "F2", - [F3] = "F'", - [B] = "B", - [B2] = "B2", - [B3] = "B'", -}; - -static char *transstr[] = { - [UFr] = "rotation UF", - [UFm] = "mirrored UF", - [ULr] = "rotation UL", - [ULm] = "mirrored UL", - [UBr] = "rotation UB", - [UBm] = "mirrored UB", - [URr] = "rotation UR", - [URm] = "mirrored UR", - [DFr] = "rotation DF", - [DFm] = "mirrored DF", - [DLr] = "rotation DL", - [DLm] = "mirrored DL", - [DBr] = "rotation DB", - [DBm] = "mirrored DB", - [DRr] = "rotation DR", - [DRm] = "mirrored DR", - [RUr] = "rotation RU", - [RUm] = "mirrored RU", - [RFr] = "rotation RF", - [RFm] = "mirrored RF", - [RDr] = "rotation RD", - [RDm] = "mirrored RD", - [RBr] = "rotation RB", - [RBm] = "mirrored RB", - [LUr] = "rotation LU", - [LUm] = "mirrored LU", - [LFr] = "rotation LF", - [LFm] = "mirrored LF", - [LDr] = "rotation LD", - [LDm] = "mirrored LD", - [LBr] = "rotation LB", - [LBm] = "mirrored LB", - [FUr] = "rotation FU", - [FUm] = "mirrored FU", - [FRr] = "rotation FR", - [FRm] = "mirrored FR", - [FDr] = "rotation FD", - [FDm] = "mirrored FD", - [FLr] = "rotation FL", - [FLm] = "mirrored FL", - [BUr] = "rotation BU", - [BUm] = "mirrored BU", - [BRr] = "rotation BR", - [BRm] = "mirrored BR", - [BDr] = "rotation BD", - [BDm] = "mirrored BD", - [BLr] = "rotation BL", - [BLm] = "mirrored BL", -}; diff --git a/src/_move_logic_arr.c b/src/_move_logic_arr.c @@ -1,149 +0,0 @@ -#define PERM4(r, i, j, k, l) \ - aux = r[i]; \ - r[i] = r[l]; \ - r[l] = r[k]; \ - r[k] = r[j]; \ - r[j] = aux; -#define PERM22(r, i, j, k, l) \ - aux = r[i]; \ - r[i] = r[j]; \ - r[j] = aux; \ - aux = r[k]; \ - r[k] = r[l]; \ - r[l] = aux; -#define CO(a, b) \ - aux = (a & _cobits) + (b & _cobits); \ - auy = (aux + _ctwist_cw) >> 2U; \ - auz = (aux + auy) & _cobits2; \ - a = (a & _pbits) | auz; -#define CO4(r, i, j, k, l) \ - CO(r[i], _ctwist_cw) \ - CO(r[j], _ctwist_cw) \ - CO(r[k], _ctwist_ccw) \ - CO(r[l], _ctwist_ccw) -#define EO4(r, i, j, k, l) \ - r[i] ^= _eobit; \ - r[j] ^= _eobit; \ - r[k] ^= _eobit; \ - r[l] ^= _eobit; - -uint8_t aux, auy, auz; -cube_t ret = c; - -switch (m) { -case U: - PERM4(ret.e, _e_uf, _e_ul, _e_ub, _e_ur) - PERM4(ret.c, _c_ufr, _c_ufl, _c_ubl, _c_ubr) - - return ret; -case U2: - PERM22(ret.e, _e_uf, _e_ub, _e_ul, _e_ur) - PERM22(ret.c, _c_ufr, _c_ubl, _c_ufl, _c_ubr) - - return ret; -case U3: - PERM4(ret.e, _e_uf, _e_ur, _e_ub, _e_ul) - PERM4(ret.c, _c_ufr, _c_ubr, _c_ubl, _c_ufl) - - return ret; -case D: - PERM4(ret.e, _e_df, _e_dr, _e_db, _e_dl) - PERM4(ret.c, _c_dfr, _c_dbr, _c_dbl, _c_dfl) - - return ret; -case D2: - PERM22(ret.e, _e_df, _e_db, _e_dr, _e_dl) - PERM22(ret.c, _c_dfr, _c_dbl, _c_dbr, _c_dfl) - - return ret; -case D3: - PERM4(ret.e, _e_df, _e_dl, _e_db, _e_dr) - PERM4(ret.c, _c_dfr, _c_dfl, _c_dbl, _c_dbr) - - return ret; -case R: - PERM4(ret.e, _e_ur, _e_br, _e_dr, _e_fr) - PERM4(ret.c, _c_ufr, _c_ubr, _c_dbr, _c_dfr) - - CO4(ret.c, _c_ubr, _c_dfr, _c_ufr, _c_dbr) - - return ret; -case R2: - PERM22(ret.e, _e_ur, _e_dr, _e_fr, _e_br) - PERM22(ret.c, _c_ufr, _c_dbr, _c_ubr, _c_dfr) - - return ret; -case R3: - PERM4(ret.e, _e_ur, _e_fr, _e_dr, _e_br) - PERM4(ret.c, _c_ufr, _c_dfr, _c_dbr, _c_ubr) - - CO4(ret.c, _c_ubr, _c_dfr, _c_ufr, _c_dbr) - - return ret; -case L: - PERM4(ret.e, _e_ul, _e_fl, _e_dl, _e_bl) - PERM4(ret.c, _c_ufl, _c_dfl, _c_dbl, _c_ubl) - - CO4(ret.c, _c_ufl, _c_dbl, _c_dfl, _c_ubl) - - return ret; -case L2: - PERM22(ret.e, _e_ul, _e_dl, _e_fl, _e_bl) - PERM22(ret.c, _c_ufl, _c_dbl, _c_ubl, _c_dfl) - - return ret; -case L3: - PERM4(ret.e, _e_ul, _e_bl, _e_dl, _e_fl) - PERM4(ret.c, _c_ufl, _c_ubl, _c_dbl, _c_dfl) - - CO4(ret.c, _c_ufl, _c_dbl, _c_dfl, _c_ubl) - - return ret; -case F: - PERM4(ret.e, _e_uf, _e_fr, _e_df, _e_fl) - PERM4(ret.c, _c_ufr, _c_dfr, _c_dfl, _c_ufl) - - EO4(ret.e, _e_uf, _e_fr, _e_df, _e_fl) - CO4(ret.c, _c_ufr, _c_dfl, _c_dfr, _c_ufl) - - return ret; -case F2: - PERM22(ret.e, _e_uf, _e_df, _e_fr, _e_fl) - PERM22(ret.c, _c_ufr, _c_dfl, _c_ufl, _c_dfr) - - return ret; -case F3: - PERM4(ret.e, _e_uf, _e_fl, _e_df, _e_fr) - PERM4(ret.c, _c_ufr, _c_ufl, _c_dfl, _c_dfr) - - EO4(ret.e, _e_uf, _e_fr, _e_df, _e_fl) - CO4(ret.c, _c_ufr, _c_dfl, _c_dfr, _c_ufl) - - return ret; -case B: - PERM4(ret.e, _e_ub, _e_bl, _e_db, _e_br) - PERM4(ret.c, _c_ubr, _c_ubl, _c_dbl, _c_dbr) - - EO4(ret.e, _e_ub, _e_br, _e_db, _e_bl) - CO4(ret.c, _c_ubl, _c_dbr, _c_dbl, _c_ubr) - - return ret; -case B2: - PERM22(ret.e, _e_ub, _e_db, _e_br, _e_bl) - PERM22(ret.c, _c_ubr, _c_dbl, _c_ubl, _c_dbr) - - return ret; -case B3: - PERM4(ret.e, _e_ub, _e_br, _e_db, _e_bl) - PERM4(ret.c, _c_ubr, _c_dbr, _c_dbl, _c_ubl) - - EO4(ret.e, _e_ub, _e_br, _e_db, _e_bl) - CO4(ret.c, _c_ubl, _c_dbr, _c_dbl, _c_ubr) - - return ret; -default: -#ifdef DEBUG - fprintf(stderr, "mover error, unknown move\n"); -#endif - goto move_error; -} diff --git a/src/_move_logic_avx2.c b/src/_move_logic_avx2.c @@ -1,6 +0,0 @@ -/* TODO: -static cube_t move_cube[18]; -return compose(move_cube[m], c); -*/ - -goto move_error; diff --git a/src/_moves_arr.c b/src/_moves_arr.c @@ -0,0 +1,264 @@ +#define PERM4(r, i, j, k, l) \ + aux = r[i]; \ + r[i] = r[l]; \ + r[l] = r[k]; \ + r[k] = r[j]; \ + r[j] = aux; +#define PERM22(r, i, j, k, l) \ + aux = r[i]; \ + r[i] = r[j]; \ + r[j] = aux; \ + aux = r[k]; \ + r[k] = r[l]; \ + r[l] = aux; +#define CO(a, b) \ + aux = (a & _cobits) + (b & _cobits); \ + auy = (aux + _ctwist_cw) >> 2U; \ + auz = (aux + auy) & _cobits2; \ + a = (a & _pbits) | auz; +#define CO4(r, i, j, k, l) \ + CO(r[i], _ctwist_cw) \ + CO(r[j], _ctwist_cw) \ + CO(r[k], _ctwist_ccw) \ + CO(r[l], _ctwist_ccw) +#define EO4(r, i, j, k, l) \ + r[i] ^= _eobit; \ + r[j] ^= _eobit; \ + r[k] ^= _eobit; \ + r[l] ^= _eobit; + +static inline cube_t +inline_move_U(cube_t c) +{ + uint8_t aux; + cube_t ret = c; + + PERM4(ret.e, _e_uf, _e_ul, _e_ub, _e_ur) + PERM4(ret.c, _c_ufr, _c_ufl, _c_ubl, _c_ubr) + + return ret; +} + +static inline cube_t +inline_move_U2(cube_t c) +{ + uint8_t aux; + cube_t ret = c; + + PERM22(ret.e, _e_uf, _e_ub, _e_ul, _e_ur) + PERM22(ret.c, _c_ufr, _c_ubl, _c_ufl, _c_ubr) + + return ret; +} + +static inline cube_t +inline_move_U3(cube_t c) +{ + uint8_t aux; + cube_t ret = c; + + PERM4(ret.e, _e_uf, _e_ur, _e_ub, _e_ul) + PERM4(ret.c, _c_ufr, _c_ubr, _c_ubl, _c_ufl) + + return ret; +} + +static inline cube_t +inline_move_D(cube_t c) +{ + uint8_t aux; + cube_t ret = c; + + PERM4(ret.e, _e_df, _e_dr, _e_db, _e_dl) + PERM4(ret.c, _c_dfr, _c_dbr, _c_dbl, _c_dfl) + + return ret; +} + +static inline cube_t +inline_move_D2(cube_t c) +{ + uint8_t aux; + cube_t ret = c; + + PERM22(ret.e, _e_df, _e_db, _e_dr, _e_dl) + PERM22(ret.c, _c_dfr, _c_dbl, _c_dbr, _c_dfl) + + return ret; +} + +static inline cube_t +inline_move_D3(cube_t c) +{ + uint8_t aux; + cube_t ret = c; + + PERM4(ret.e, _e_df, _e_dl, _e_db, _e_dr) + PERM4(ret.c, _c_dfr, _c_dfl, _c_dbl, _c_dbr) + + return ret; +} + +static inline cube_t +inline_move_R(cube_t c) +{ + uint8_t aux, auy, auz; + cube_t ret = c; + + PERM4(ret.e, _e_ur, _e_br, _e_dr, _e_fr) + PERM4(ret.c, _c_ufr, _c_ubr, _c_dbr, _c_dfr) + + CO4(ret.c, _c_ubr, _c_dfr, _c_ufr, _c_dbr) + + return ret; +} + +static inline cube_t +inline_move_R2(cube_t c) +{ + uint8_t aux; + cube_t ret = c; + + PERM22(ret.e, _e_ur, _e_dr, _e_fr, _e_br) + PERM22(ret.c, _c_ufr, _c_dbr, _c_ubr, _c_dfr) + + return ret; +} + +static inline cube_t +inline_move_R3(cube_t c) +{ + uint8_t aux, auy, auz; + cube_t ret = c; + + PERM4(ret.e, _e_ur, _e_fr, _e_dr, _e_br) + PERM4(ret.c, _c_ufr, _c_dfr, _c_dbr, _c_ubr) + + CO4(ret.c, _c_ubr, _c_dfr, _c_ufr, _c_dbr) + + return ret; +} + +static inline cube_t +inline_move_L(cube_t c) +{ + uint8_t aux, auy, auz; + cube_t ret = c; + + PERM4(ret.e, _e_ul, _e_fl, _e_dl, _e_bl) + PERM4(ret.c, _c_ufl, _c_dfl, _c_dbl, _c_ubl) + + CO4(ret.c, _c_ufl, _c_dbl, _c_dfl, _c_ubl) + + return ret; +} + +static inline cube_t +inline_move_L2(cube_t c) +{ + uint8_t aux; + cube_t ret = c; + + PERM22(ret.e, _e_ul, _e_dl, _e_fl, _e_bl) + PERM22(ret.c, _c_ufl, _c_dbl, _c_ubl, _c_dfl) + + return ret; +} + +static inline cube_t +inline_move_L3(cube_t c) +{ + uint8_t aux, auy, auz; + cube_t ret = c; + + PERM4(ret.e, _e_ul, _e_bl, _e_dl, _e_fl) + PERM4(ret.c, _c_ufl, _c_ubl, _c_dbl, _c_dfl) + + CO4(ret.c, _c_ufl, _c_dbl, _c_dfl, _c_ubl) + + return ret; +} + +static inline cube_t +inline_move_F(cube_t c) +{ + uint8_t aux, auy, auz; + cube_t ret = c; + + PERM4(ret.e, _e_uf, _e_fr, _e_df, _e_fl) + PERM4(ret.c, _c_ufr, _c_dfr, _c_dfl, _c_ufl) + + EO4(ret.e, _e_uf, _e_fr, _e_df, _e_fl) + CO4(ret.c, _c_ufr, _c_dfl, _c_dfr, _c_ufl) + + return ret; +} + +static inline cube_t +inline_move_F2(cube_t c) +{ + uint8_t aux; + cube_t ret = c; + + PERM22(ret.e, _e_uf, _e_df, _e_fr, _e_fl) + PERM22(ret.c, _c_ufr, _c_dfl, _c_ufl, _c_dfr) + + return ret; +} + +static inline cube_t +inline_move_F3(cube_t c) +{ + uint8_t aux, auy, auz; + cube_t ret = c; + + PERM4(ret.e, _e_uf, _e_fl, _e_df, _e_fr) + PERM4(ret.c, _c_ufr, _c_ufl, _c_dfl, _c_dfr) + + EO4(ret.e, _e_uf, _e_fr, _e_df, _e_fl) + CO4(ret.c, _c_ufr, _c_dfl, _c_dfr, _c_ufl) + + return ret; +} + +static inline cube_t +inline_move_B(cube_t c) +{ + uint8_t aux, auy, auz; + cube_t ret = c; + + PERM4(ret.e, _e_ub, _e_bl, _e_db, _e_br) + PERM4(ret.c, _c_ubr, _c_ubl, _c_dbl, _c_dbr) + + EO4(ret.e, _e_ub, _e_br, _e_db, _e_bl) + CO4(ret.c, _c_ubl, _c_dbr, _c_dbl, _c_ubr) + + return ret; +} + +static inline cube_t +inline_move_B2(cube_t c) +{ + uint8_t aux; + cube_t ret = c; + + PERM22(ret.e, _e_ub, _e_db, _e_br, _e_bl) + PERM22(ret.c, _c_ubr, _c_dbl, _c_ubl, _c_dbr) + + return ret; +} + +static inline cube_t +inline_move_B3(cube_t c) +{ + uint8_t aux, auy, auz; + cube_t ret = c; + + PERM4(ret.e, _e_ub, _e_br, _e_db, _e_bl) + PERM4(ret.c, _c_ubr, _c_dbr, _c_dbl, _c_ubl) + + EO4(ret.e, _e_ub, _e_br, _e_db, _e_bl) + CO4(ret.c, _c_ubl, _c_dbr, _c_dbl, _c_ubr) + + return ret; +} diff --git a/src/_moves_avx2.c b/src/_moves_avx2.c @@ -0,0 +1,2 @@ +/* TODO: return compose(c, (some avx garbage)); +*/ diff --git a/src/_trans_arr.c b/src/_trans_arr.c @@ -0,0 +1,952 @@ +static inline cube_t +flipallcorners(cube_t c) +{ + uint8_t i, piece, orien; + cube_t ret; + + ret = c; + for (i = 0; i < 8; i++) { + piece = get_corner(c, i); + orien = ((piece << 1) | (piece >> 1)) & _cobits2; + set_corner(ret, i, (piece & _pbits) | orien); + } + + return ret; +} + +static inline cube_t +inline_trans_UFr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {0, 1, 2, 3, 4, 5, 6, 7}, + .e = {0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11} + }; + cube_t ti = { + .c = {0, 1, 2, 3, 4, 5, 6, 7}, + .e = {0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_ULr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {4, 5, 7, 6, 1, 0, 2, 3}, + .e = {5, 4, 7, 6, 0, 1, 2, 3, 25, 26, 27, 24} + }; + cube_t ti = { + .c = {5, 4, 6, 7, 0, 1, 3, 2}, + .e = {4, 5, 6, 7, 1, 0, 3, 2, 27, 24, 25, 26} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_UBr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {1, 0, 3, 2, 5, 4, 7, 6}, + .e = {1, 0, 3, 2, 5, 4, 7, 6, 10, 11, 8, 9} + }; + cube_t ti = { + .c = {1, 0, 3, 2, 5, 4, 7, 6}, + .e = {1, 0, 3, 2, 5, 4, 7, 6, 10, 11, 8, 9} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_URr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {5, 4, 6, 7, 0, 1, 3, 2}, + .e = {4, 5, 6, 7, 1, 0, 3, 2, 27, 24, 25, 26} + }; + cube_t ti = { + .c = {4, 5, 7, 6, 1, 0, 2, 3}, + .e = {5, 4, 7, 6, 0, 1, 2, 3, 25, 26, 27, 24} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_DFr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {2, 3, 0, 1, 6, 7, 4, 5}, + .e = {3, 2, 1, 0, 6, 7, 4, 5, 9, 8, 11, 10} + }; + cube_t ti = { + .c = {2, 3, 0, 1, 6, 7, 4, 5}, + .e = {3, 2, 1, 0, 6, 7, 4, 5, 9, 8, 11, 10} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_DLr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {7, 6, 4, 5, 2, 3, 1, 0}, + .e = {6, 7, 4, 5, 2, 3, 0, 1, 26, 25, 24, 27} + }; + cube_t ti = { + .c = {7, 6, 4, 5, 2, 3, 1, 0}, + .e = {6, 7, 4, 5, 2, 3, 0, 1, 26, 25, 24, 27} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_DBr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {3, 2, 1, 0, 7, 6, 5, 4}, + .e = {2, 3, 0, 1, 7, 6, 5, 4, 11, 10, 9, 8} + }; + cube_t ti = { + .c = {3, 2, 1, 0, 7, 6, 5, 4}, + .e = {2, 3, 0, 1, 7, 6, 5, 4, 11, 10, 9, 8} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_DRr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {6, 7, 5, 4, 3, 2, 0, 1}, + .e = {7, 6, 5, 4, 3, 2, 1, 0, 24, 27, 26, 25} + }; + cube_t ti = { + .c = {6, 7, 5, 4, 3, 2, 0, 1}, + .e = {7, 6, 5, 4, 3, 2, 1, 0, 24, 27, 26, 25} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_RUr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {64, 67, 65, 66, 37, 38, 36, 39}, + .e = {20, 23, 22, 21, 24, 27, 26, 25, 0, 1, 2, 3} + }; + cube_t ti = { + .c = {32, 34, 35, 33, 70, 68, 69, 71}, + .e = {8, 9, 10, 11, 16, 19, 18, 17, 20, 23, 22, 21} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_RFr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {38, 37, 36, 39, 64, 67, 66, 65}, + .e = {24, 27, 26, 25, 23, 20, 21, 22, 19, 16, 17, 18} + }; + cube_t ti = { + .c = {36, 39, 38, 37, 66, 65, 64, 67}, + .e = {25, 26, 27, 24, 21, 22, 23, 20, 16, 19, 18, 17} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_RDr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {67, 64, 66, 65, 38, 37, 39, 36}, + .e = {23, 20, 21, 22, 27, 24, 25, 26, 2, 3, 0, 1} + }; + cube_t ti = { + .c = {33, 35, 34, 32, 71, 69, 68, 70}, + .e = {10, 11, 8, 9, 17, 18, 19, 16, 21, 22, 23, 20} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_RBr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {37, 38, 39, 36, 67, 64, 65, 66}, + .e = {27, 24, 25, 26, 20, 23, 22, 21, 17, 18, 19, 16} + }; + cube_t ti = { + .c = {37, 38, 39, 36, 67, 64, 65, 66}, + .e = {27, 24, 25, 26, 20, 23, 22, 21, 17, 18, 19, 16} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_LUr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {65, 66, 64, 67, 36, 39, 37, 38}, + .e = {21, 22, 23, 20, 26, 25, 24, 27, 1, 0, 3, 2} + }; + cube_t ti = { + .c = {34, 32, 33, 35, 68, 70, 71, 69}, + .e = {9, 8, 11, 10, 19, 16, 17, 18, 22, 21, 20, 23} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_LFr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {36, 39, 38, 37, 66, 65, 64, 67}, + .e = {25, 26, 27, 24, 21, 22, 23, 20, 16, 19, 18, 17} + }; + cube_t ti = { + .c = {38, 37, 36, 39, 64, 67, 66, 65}, + .e = {24, 27, 26, 25, 23, 20, 21, 22, 19, 16, 17, 18} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_LDr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {66, 65, 67, 64, 39, 36, 38, 37}, + .e = {22, 21, 20, 23, 25, 26, 27, 24, 3, 2, 1, 0} + }; + cube_t ti = { + .c = {35, 33, 32, 34, 69, 71, 70, 68}, + .e = {11, 10, 9, 8, 18, 17, 16, 19, 23, 20, 21, 22} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_LBr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {39, 36, 37, 38, 65, 66, 67, 64}, + .e = {26, 25, 24, 27, 22, 21, 20, 23, 18, 17, 16, 19} + }; + cube_t ti = { + .c = {39, 36, 37, 38, 65, 66, 67, 64}, + .e = {26, 25, 24, 27, 22, 21, 20, 23, 18, 17, 16, 19} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_FUr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {68, 70, 69, 71, 32, 34, 33, 35}, + .e = {16, 19, 18, 17, 9, 8, 11, 10, 5, 4, 7, 6} + }; + cube_t ti = { + .c = {68, 70, 69, 71, 32, 34, 33, 35}, + .e = {16, 19, 18, 17, 9, 8, 11, 10, 5, 4, 7, 6} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_FRr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {32, 34, 35, 33, 70, 68, 69, 71}, + .e = {8, 9, 10, 11, 16, 19, 18, 17, 20, 23, 22, 21} + }; + cube_t ti = { + .c = {64, 67, 65, 66, 37, 38, 36, 39}, + .e = {20, 23, 22, 21, 24, 27, 26, 25, 0, 1, 2, 3} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_FDr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {70, 68, 71, 69, 34, 32, 35, 33}, + .e = {19, 16, 17, 18, 8, 9, 10, 11, 7, 6, 5, 4} + }; + cube_t ti = { + .c = {69, 71, 68, 70, 33, 35, 32, 34}, + .e = {17, 18, 19, 16, 11, 10, 9, 8, 4, 5, 6, 7} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_FLr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {34, 32, 33, 35, 68, 70, 71, 69}, + .e = {9, 8, 11, 10, 19, 16, 17, 18, 22, 21, 20, 23} + }; + cube_t ti = { + .c = {65, 66, 64, 67, 36, 39, 37, 38}, + .e = {21, 22, 23, 20, 26, 25, 24, 27, 1, 0, 3, 2} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_BUr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {69, 71, 68, 70, 33, 35, 32, 34}, + .e = {17, 18, 19, 16, 11, 10, 9, 8, 4, 5, 6, 7} + }; + cube_t ti = { + .c = {70, 68, 71, 69, 34, 32, 35, 33}, + .e = {19, 16, 17, 18, 8, 9, 10, 11, 7, 6, 5, 4} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_BRr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {35, 33, 32, 34, 69, 71, 70, 68}, + .e = {11, 10, 9, 8, 18, 17, 16, 19, 23, 20, 21, 22} + }; + cube_t ti = { + .c = {66, 65, 67, 64, 39, 36, 38, 37}, + .e = {22, 21, 20, 23, 25, 26, 27, 24, 3, 2, 1, 0} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_BDr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {71, 69, 70, 68, 35, 33, 34, 32}, + .e = {18, 17, 16, 19, 10, 11, 8, 9, 6, 7, 4, 5} + }; + cube_t ti = { + .c = {71, 69, 70, 68, 35, 33, 34, 32}, + .e = {18, 17, 16, 19, 10, 11, 8, 9, 6, 7, 4, 5} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_BLr(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {33, 35, 34, 32, 71, 69, 68, 70}, + .e = {10, 11, 8, 9, 17, 18, 19, 16, 21, 22, 23, 20} + }; + cube_t ti = { + .c = {67, 64, 66, 65, 38, 37, 39, 36}, + .e = {23, 20, 21, 22, 27, 24, 25, 26, 2, 3, 0, 1} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + + return ret; +} + +static inline cube_t +inline_trans_UFm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {4, 5, 6, 7, 0, 1, 2, 3}, + .e = {0, 1, 2, 3, 5, 4, 7, 6, 9, 8, 11, 10} + }; + cube_t ti = { + .c = {4, 5, 6, 7, 0, 1, 2, 3}, + .e = {0, 1, 2, 3, 5, 4, 7, 6, 9, 8, 11, 10} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_ULm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {0, 1, 3, 2, 5, 4, 6, 7}, + .e = {4, 5, 6, 7, 0, 1, 2, 3, 24, 27, 26, 25} + }; + cube_t ti = { + .c = {0, 1, 3, 2, 5, 4, 6, 7}, + .e = {4, 5, 6, 7, 0, 1, 2, 3, 24, 27, 26, 25} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_UBm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {5, 4, 7, 6, 1, 0, 3, 2}, + .e = {1, 0, 3, 2, 4, 5, 6, 7, 11, 10, 9, 8} + }; + cube_t ti = { + .c = {5, 4, 7, 6, 1, 0, 3, 2}, + .e = {1, 0, 3, 2, 4, 5, 6, 7, 11, 10, 9, 8} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_URm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {1, 0, 2, 3, 4, 5, 7, 6}, + .e = {5, 4, 7, 6, 1, 0, 3, 2, 26, 25, 24, 27} + }; + cube_t ti = { + .c = {1, 0, 2, 3, 4, 5, 7, 6}, + .e = {5, 4, 7, 6, 1, 0, 3, 2, 26, 25, 24, 27} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_DFm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {6, 7, 4, 5, 2, 3, 0, 1}, + .e = {3, 2, 1, 0, 7, 6, 5, 4, 8, 9, 10, 11} + }; + cube_t ti = { + .c = {6, 7, 4, 5, 2, 3, 0, 1}, + .e = {3, 2, 1, 0, 7, 6, 5, 4, 8, 9, 10, 11} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_DLm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {3, 2, 0, 1, 6, 7, 5, 4}, + .e = {7, 6, 5, 4, 2, 3, 0, 1, 27, 24, 25, 26} + }; + cube_t ti = { + .c = {2, 3, 1, 0, 7, 6, 4, 5}, + .e = {6, 7, 4, 5, 3, 2, 1, 0, 25, 26, 27, 24} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_DBm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {7, 6, 5, 4, 3, 2, 1, 0}, + .e = {2, 3, 0, 1, 6, 7, 4, 5, 10, 11, 8, 9} + }; + cube_t ti = { + .c = {7, 6, 5, 4, 3, 2, 1, 0}, + .e = {2, 3, 0, 1, 6, 7, 4, 5, 10, 11, 8, 9} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_DRm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {2, 3, 1, 0, 7, 6, 4, 5}, + .e = {6, 7, 4, 5, 3, 2, 1, 0, 25, 26, 27, 24} + }; + cube_t ti = { + .c = {3, 2, 0, 1, 6, 7, 5, 4}, + .e = {7, 6, 5, 4, 2, 3, 0, 1, 27, 24, 25, 26} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_RUm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {68, 71, 69, 70, 33, 34, 32, 35}, + .e = {21, 22, 23, 20, 25, 26, 27, 24, 0, 1, 2, 3} + }; + cube_t ti = { + .c = {70, 68, 69, 71, 32, 34, 35, 33}, + .e = {8, 9, 10, 11, 19, 16, 17, 18, 23, 20, 21, 22} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_RFm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {34, 33, 32, 35, 68, 71, 70, 69}, + .e = {25, 26, 27, 24, 22, 21, 20, 23, 19, 16, 17, 18} + }; + cube_t ti = { + .c = {66, 65, 64, 67, 36, 39, 38, 37}, + .e = {25, 26, 27, 24, 22, 21, 20, 23, 19, 16, 17, 18} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_RDm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {71, 68, 70, 69, 34, 33, 35, 32}, + .e = {22, 21, 20, 23, 26, 25, 24, 27, 2, 3, 0, 1} + }; + cube_t ti = { + .c = {71, 69, 68, 70, 33, 35, 34, 32}, + .e = {10, 11, 8, 9, 18, 17, 16, 19, 22, 21, 20, 23} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_RBm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {33, 34, 35, 32, 71, 68, 69, 70}, + .e = {26, 25, 24, 27, 21, 22, 23, 20, 17, 18, 19, 16} + }; + cube_t ti = { + .c = {67, 64, 65, 66, 37, 38, 39, 36}, + .e = {27, 24, 25, 26, 23, 20, 21, 22, 18, 17, 16, 19} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_LUm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {69, 70, 68, 71, 32, 35, 33, 34}, + .e = {20, 23, 22, 21, 27, 24, 25, 26, 1, 0, 3, 2} + }; + cube_t ti = { + .c = {68, 70, 71, 69, 34, 32, 33, 35}, + .e = {9, 8, 11, 10, 16, 19, 18, 17, 21, 22, 23, 20} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_LFm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {32, 35, 34, 33, 70, 69, 68, 71}, + .e = {24, 27, 26, 25, 20, 23, 22, 21, 16, 19, 18, 17} + }; + cube_t ti = { + .c = {64, 67, 66, 65, 38, 37, 36, 39}, + .e = {24, 27, 26, 25, 20, 23, 22, 21, 16, 19, 18, 17} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_LDm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {70, 69, 71, 68, 35, 32, 34, 33}, + .e = {23, 20, 21, 22, 24, 27, 26, 25, 3, 2, 1, 0} + }; + cube_t ti = { + .c = {69, 71, 70, 68, 35, 33, 32, 34}, + .e = {11, 10, 9, 8, 17, 18, 19, 16, 20, 23, 22, 21} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_LBm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {35, 32, 33, 34, 69, 70, 71, 68}, + .e = {27, 24, 25, 26, 23, 20, 21, 22, 18, 17, 16, 19} + }; + cube_t ti = { + .c = {65, 66, 67, 64, 39, 36, 37, 38}, + .e = {26, 25, 24, 27, 21, 22, 23, 20, 17, 18, 19, 16} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_FUm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {64, 66, 65, 67, 36, 38, 37, 39}, + .e = {16, 19, 18, 17, 8, 9, 10, 11, 4, 5, 6, 7} + }; + cube_t ti = { + .c = {32, 34, 33, 35, 68, 70, 69, 71}, + .e = {16, 19, 18, 17, 8, 9, 10, 11, 4, 5, 6, 7} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_FRm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {36, 38, 39, 37, 66, 64, 65, 67}, + .e = {9, 8, 11, 10, 16, 19, 18, 17, 21, 22, 23, 20} + }; + cube_t ti = { + .c = {37, 38, 36, 39, 64, 67, 65, 66}, + .e = {20, 23, 22, 21, 27, 24, 25, 26, 1, 0, 3, 2} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_FDm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {66, 64, 67, 65, 38, 36, 39, 37}, + .e = {19, 16, 17, 18, 9, 8, 11, 10, 6, 7, 4, 5} + }; + cube_t ti = { + .c = {33, 35, 32, 34, 69, 71, 68, 70}, + .e = {17, 18, 19, 16, 10, 11, 8, 9, 5, 4, 7, 6} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_FLm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {38, 36, 37, 39, 64, 66, 67, 65}, + .e = {8, 9, 10, 11, 19, 16, 17, 18, 23, 20, 21, 22} + }; + cube_t ti = { + .c = {36, 39, 37, 38, 65, 66, 64, 67}, + .e = {21, 22, 23, 20, 25, 26, 27, 24, 0, 1, 2, 3} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_BUm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {65, 67, 64, 66, 37, 39, 36, 38}, + .e = {17, 18, 19, 16, 10, 11, 8, 9, 5, 4, 7, 6} + }; + cube_t ti = { + .c = {34, 32, 35, 33, 70, 68, 71, 69}, + .e = {19, 16, 17, 18, 9, 8, 11, 10, 6, 7, 4, 5} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_BRm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {39, 37, 36, 38, 65, 67, 66, 64}, + .e = {10, 11, 8, 9, 18, 17, 16, 19, 22, 21, 20, 23} + }; + cube_t ti = { + .c = {39, 36, 38, 37, 66, 65, 67, 64}, + .e = {22, 21, 20, 23, 26, 25, 24, 27, 2, 3, 0, 1} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_BDm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {67, 65, 66, 64, 39, 37, 38, 36}, + .e = {18, 17, 16, 19, 11, 10, 9, 8, 7, 6, 5, 4} + }; + cube_t ti = { + .c = {35, 33, 34, 32, 71, 69, 70, 68}, + .e = {18, 17, 16, 19, 11, 10, 9, 8, 7, 6, 5, 4} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + +static inline cube_t +inline_trans_BLm(cube_t c) +{ + cube_t ret; + cube_t tn = { + .c = {37, 39, 38, 36, 67, 65, 64, 66}, + .e = {11, 10, 9, 8, 17, 18, 19, 16, 20, 23, 22, 21} + }; + cube_t ti = { + .c = {38, 37, 39, 36, 67, 64, 66, 65}, + .e = {23, 20, 21, 22, 24, 27, 26, 25, 3, 2, 1, 0} + }; + + ret = compose(tn, c); + ret = compose(ret, ti); + ret = flipallcorners(ret); + + return ret; +} + diff --git a/src/_trans_avx2.c b/src/_trans_avx2.c diff --git a/src/_trans_move_avx2.c b/src/_trans_move_avx2.c @@ -1,2 +0,0 @@ -static cube_t trans_move_cube[48]; -static cube_t trans_move_cube_inverse[48]; diff --git a/src/cube.c b/src/cube.c @@ -2,17 +2,121 @@ #include <stdbool.h> #include <string.h> -#ifdef CUBE_AVX2 -#include <immintrin.h> -#endif - #ifdef DEBUG #include <stdio.h> #endif +#ifdef AVX2 +#include <immintrin.h> +#endif + #include "cube.h" -#include "_constants.c" +#define U 0U +#define U2 1U +#define U3 2U +#define D 3U +#define D2 4U +#define D3 5U +#define R 6U +#define R2 7U +#define R3 8U +#define L 9U +#define L2 10U +#define L3 11U +#define F 12U +#define F2 13U +#define F3 14U +#define B 15U +#define B2 16U +#define B3 17U + +#define UFr 0 +#define ULr 1 +#define UBr 2 +#define URr 3 +#define DFr 4 +#define DLr 5 +#define DBr 6 +#define DRr 7 +#define RUr 8 +#define RFr 9 +#define RDr 10 +#define RBr 11 +#define LUr 12 +#define LFr 13 +#define LDr 14 +#define LBr 15 +#define FUr 16 +#define FRr 17 +#define FDr 18 +#define FLr 19 +#define BUr 20 +#define BRr 21 +#define BDr 22 +#define BLr 23 + +#define UFm 24 +#define ULm 25 +#define UBm 26 +#define URm 27 +#define DFm 28 +#define DLm 29 +#define DBm 30 +#define DRm 31 +#define RUm 32 +#define RFm 33 +#define RDm 34 +#define RBm 35 +#define LUm 36 +#define LFm 37 +#define LDm 38 +#define LBm 39 +#define FUm 40 +#define FRm 41 +#define FDm 42 +#define FLm 43 +#define BUm 44 +#define BRm 45 +#define BDm 46 +#define BLm 47 + +#define errormove 99U +#define errortrans 99U + +#define _c_ufr 0U +#define _c_ubl 1U +#define _c_dfl 2U +#define _c_dbr 3U +#define _c_ufl 4U +#define _c_ubr 5U +#define _c_dfr 6U +#define _c_dbl 7U + +#define _e_uf 0U +#define _e_ub 1U +#define _e_db 2U +#define _e_df 3U +#define _e_ur 4U +#define _e_ul 5U +#define _e_dl 6U +#define _e_dr 7U +#define _e_fr 8U +#define _e_fl 9U +#define _e_bl 10U +#define _e_br 11U + +#define _eoshift 4U +#define _coshift 5U + +#define _pbits 0xFU +#define _eobit 0x10U +#define _cobits 0xF0U +#define _cobits2 0x60U +#define _ctwist_cw 0x20U +#define _ctwist_ccw 0x40U +#define _eflip 0x10U +#define _error 0xFFU cube_arr_t solvedcube_arr = { .c = {0, 1, 2, 3, 4, 5, 6, 7, 0, 0, 0, 0, 0, 0, 0, 0}, @@ -36,18 +140,133 @@ cube_arr_t zerocube_arr = { .e = {0}, .c = {0} }; _mm256_andnot_si256(cube, _mm256_slli_si256(_c, 0)), \ _mm256_and_si256(_mm256_set1_epi8(p), _mm256_slli_si256(_c, 0))) +#include "_moves_avx2.c" +#include "_trans_avx2.c" + #else #define setsolved(cube) cube = solvedcube_arr #define setzero(cube) cube = zerocube_arr -#define _c _mm256_set_epi64x(0, 0, 0, 0xFF) #define get_edge(cube, i) (cube).e[(i)] #define get_corner(cube, i) (cube).c[(i)] #define set_edge(cube, i, p) (cube).e[(i)] = (p) #define set_corner(cube, i, p) (cube).c[(i)] = (p) +#include "_moves_arr.c" +#include "_trans_arr.c" + #endif + +static char *cornerstr[] = { + [_c_ufr] = "UFR", + [_c_ubl] = "UBL", + [_c_dfl] = "DFL", + [_c_dbr] = "DBR", + [_c_ufl] = "UFL", + [_c_ubr] = "UBR", + [_c_dfr] = "DFR", + [_c_dbl] = "DBL" +}; + +static char *cornerstralt[] = { + [_c_ufr] = "URF", + [_c_ubl] = "ULB", + [_c_dfl] = "DLF", + [_c_dbr] = "DRB", + [_c_ufl] = "ULF", + [_c_ubr] = "URB", + [_c_dfr] = "DRF", + [_c_dbl] = "DLB" +}; + +static char *edgestr[] = { + [_e_uf] = "UF", + [_e_ub] = "UB", + [_e_db] = "DB", + [_e_df] = "DF", + [_e_ur] = "UR", + [_e_ul] = "UL", + [_e_dl] = "DL", + [_e_dr] = "DR", + [_e_fr] = "FR", + [_e_fl] = "FL", + [_e_bl] = "BL", + [_e_br] = "BR" +}; + +static char *movestr[] = { + [U] = "U", + [U2] = "U2", + [U3] = "U'", + [D] = "D", + [D2] = "D2", + [D3] = "D'", + [R] = "R", + [R2] = "R2", + [R3] = "R'", + [L] = "L", + [L2] = "L2", + [L3] = "L'", + [F] = "F", + [F2] = "F2", + [F3] = "F'", + [B] = "B", + [B2] = "B2", + [B3] = "B'", +}; + +static char *transstr[] = { + [UFr] = "rotation UF", + [UFm] = "mirrored UF", + [ULr] = "rotation UL", + [ULm] = "mirrored UL", + [UBr] = "rotation UB", + [UBm] = "mirrored UB", + [URr] = "rotation UR", + [URm] = "mirrored UR", + [DFr] = "rotation DF", + [DFm] = "mirrored DF", + [DLr] = "rotation DL", + [DLm] = "mirrored DL", + [DBr] = "rotation DB", + [DBm] = "mirrored DB", + [DRr] = "rotation DR", + [DRm] = "mirrored DR", + [RUr] = "rotation RU", + [RUm] = "mirrored RU", + [RFr] = "rotation RF", + [RFm] = "mirrored RF", + [RDr] = "rotation RD", + [RDm] = "mirrored RD", + [RBr] = "rotation RB", + [RBm] = "mirrored RB", + [LUr] = "rotation LU", + [LUm] = "mirrored LU", + [LFr] = "rotation LF", + [LFm] = "mirrored LF", + [LDr] = "rotation LD", + [LDm] = "mirrored LD", + [LBr] = "rotation LB", + [LBm] = "mirrored LB", + [FUr] = "rotation FU", + [FUm] = "mirrored FU", + [FRr] = "rotation FR", + [FRm] = "mirrored FR", + [FDr] = "rotation FD", + [FDm] = "mirrored FD", + [FLr] = "rotation FL", + [FLm] = "mirrored FL", + [BUr] = "rotation BU", + [BUm] = "mirrored BU", + [BRr] = "rotation BR", + [BRm] = "mirrored BR", + [BDr] = "rotation BD", + [BDm] = "mirrored BD", + [BLr] = "rotation BL", + [BLm] = "mirrored BL", +}; + static bool isconsistent(cube_t); static cube_t flipallcorners(cube_t); static uint8_t readco(char *); @@ -187,9 +406,9 @@ readcube(format_t format, char *buf) ret = readcube_H48(buf); break; default: - #ifdef DEBUG +#ifdef DEBUG fprintf(stderr, "Cannot read cube in the given format\n"); - #endif +#endif setzero(ret); } @@ -421,18 +640,6 @@ writetrans(trans_t t, char *buf) buf[11] = '\0'; } -move_t -inverse_move(move_t m) -{ - return inverse_move_arr[m]; -} - -trans_t -inverse_trans(trans_t t) -{ - return inverse_trans_arr[t]; -} - static int permsign(uint8_t *a, int n) { @@ -617,11 +824,49 @@ move(cube_t c, move_t m) } #endif -#ifdef CUBE_AVX2 -#include "_move_logic_avx2.c" -#else -#include "_move_logic_arr.c" + switch (m) { + case U: + return inline_move_U(c); + case U2: + return inline_move_U2(c); + case U3: + return inline_move_U3(c); + case D: + return inline_move_D(c); + case D2: + return inline_move_D2(c); + case D3: + return inline_move_D3(c); + case R: + return inline_move_R(c); + case R2: + return inline_move_R2(c); + case R3: + return inline_move_R3(c); + case L: + return inline_move_L(c); + case L2: + return inline_move_L2(c); + case L3: + return inline_move_L3(c); + case F: + return inline_move_F(c); + case F2: + return inline_move_F2(c); + case F3: + return inline_move_F3(c); + case B: + return inline_move_B(c); + case B2: + return inline_move_B2(c); + case B3: + return inline_move_B3(c); + default: +#ifdef DEBUG + fprintf(stderr, "mover error, unknown move\n"); #endif + goto move_error; + } move_error: setzero(err); @@ -696,56 +941,123 @@ compose(cube_t c1, cube_t c2) return ret; } -static cube_t -flipallcorners(cube_t c) -{ -/* TODO: optimize for avx2, can be a couple of instructions */ - - uint8_t i, piece, orien; - cube_t ret; - - ret = c; - for (i = 0; i < 8; i++) { - piece = get_corner(c, i); - orien = ((piece << 1) | (piece >> 1)) & _cobits2; - set_corner(ret, i, (piece & _pbits) | orien); - } - - return ret; -} - cube_t transform(cube_t c, trans_t t) { - cube_t ret, solved; + cube_t ret; #ifdef DEBUG - setzero(ret); if (!isconsistent(c)) { fprintf(stderr, "transform error, inconsistent cube\n"); - return ret; - } - if (t >= 48) { - fprintf(stderr, "transform error, unknown transformation\n"); - return ret; + goto trans_error; } #endif -#ifdef CUBE_AVX2 -#include "_trans_move_avx2.c" -#else -#include "_trans_move_arr.c" + switch (t) { + case UFr: + return inline_trans_UFr(c); + case ULr: + return inline_trans_ULr(c); + case UBr: + return inline_trans_UBr(c); + case URr: + return inline_trans_URr(c); + case DFr: + return inline_trans_DFr(c); + case DLr: + return inline_trans_DLr(c); + case DBr: + return inline_trans_DBr(c); + case DRr: + return inline_trans_DRr(c); + case RUr: + return inline_trans_RUr(c); + case RFr: + return inline_trans_RFr(c); + case RDr: + return inline_trans_RDr(c); + case RBr: + return inline_trans_RBr(c); + case LUr: + return inline_trans_LUr(c); + case LFr: + return inline_trans_LFr(c); + case LDr: + return inline_trans_LDr(c); + case LBr: + return inline_trans_LBr(c); + case FUr: + return inline_trans_FUr(c); + case FRr: + return inline_trans_FRr(c); + case FDr: + return inline_trans_FDr(c); + case FLr: + return inline_trans_FLr(c); + case BUr: + return inline_trans_BUr(c); + case BRr: + return inline_trans_BRr(c); + case BDr: + return inline_trans_BDr(c); + case BLr: + return inline_trans_BLr(c); + case UFm: + return inline_trans_UFm(c); + case ULm: + return inline_trans_ULm(c); + case UBm: + return inline_trans_UBm(c); + case URm: + return inline_trans_URm(c); + case DFm: + return inline_trans_DFm(c); + case DLm: + return inline_trans_DLm(c); + case DBm: + return inline_trans_DBm(c); + case DRm: + return inline_trans_DRm(c); + case RUm: + return inline_trans_RUm(c); + case RFm: + return inline_trans_RFm(c); + case RDm: + return inline_trans_RDm(c); + case RBm: + return inline_trans_RBm(c); + case LUm: + return inline_trans_LUm(c); + case LFm: + return inline_trans_LFm(c); + case LDm: + return inline_trans_LDm(c); + case LBm: + return inline_trans_LBm(c); + case FUm: + return inline_trans_FUm(c); + case FRm: + return inline_trans_FRm(c); + case FDm: + return inline_trans_FDm(c); + case FLm: + return inline_trans_FLm(c); + case BUm: + return inline_trans_BUm(c); + case BRm: + return inline_trans_BRm(c); + case BDm: + return inline_trans_BDm(c); + case BLm: + return inline_trans_BLm(c); + default: +#ifdef DEBUG + fprintf(stderr, "transform error, unknown transformation\n"); #endif + goto trans_error; + } - setsolved(solved); - - ret = compose(solved, trans_move_cube[t]); - ret = compose(ret, c); - ret = compose(ret, trans_move_cube_inverse[t]); - - /* TODO: work out a better way to do this */ - if (t >= 24) - ret = flipallcorners(ret); - +trans_error: + setzero(ret); return ret; } diff --git a/src/cube.h b/src/cube.h @@ -16,9 +16,6 @@ void writemoves(move_t *, int, char *); trans_t readtrans(char *); void writetrans(trans_t, char *); -move_t inverse_move(move_t); -trans_t inverse_trans(trans_t); - typedef enum {H48, SRC} format_t; cube_t readcube(format_t, char *); /* Supports: H48 */ void writecube(format_t, cube_t, char *); /* Supports: H48, SRC */ diff --git a/utils/gentranscode.sh b/utils/gentranscode.sh @@ -3,15 +3,42 @@ gcc -DDEBUG h48_to_src.c ../src/cube.c -o h48_to_src gcc -DDEBUG invert.c ../src/cube.c -o invert -for f in transform_??_???.txt; do - trans="$(echo $f | sed 's/.*_// ; s/\.txt//')" - printf '[%s] = ' "$trans" - if [ "$1" = "-i" ]; then - ./invert <"$f" | ./h48_to_src - else - ./h48_to_src <"$f" - fi - printf ',\n' -done +# Old version +genarray() { + for f in transform_??_???.txt; do + trans="$(echo $f | sed 's/.*_// ; s/\.txt//')" + printf '[%s] = ' "$trans" + if [ "$1" = "-i" ]; then + ./invert <"$f" | ./h48_to_src + else + ./h48_to_src <"$f" + fi + printf ',\n' + done +} +genfuncs() { + for f in transform_??_???.txt; do + trans="$(echo $f | sed 's/.*_// ; s/\.txt//')" + printf 'static inline cube_t\ninline_trans_%s' "$trans" + [ "$1" = "-i" ] && printf '_inverse' + printf '(cube_t c)\n{\n' + printf '\tcube_t ret;\n' + printf '\tcube_t tn = {\n' + ./h48_to_src <"$f" | sed 's/^/\t/' | \ + tail -n 3 | head -n 2 + printf '\t};\n\tcube_t ti = {\n' + ./invert <"$f" | ./h48_to_src | sed 's/^/\t/' | \ + tail -n 3 | head -n 2 + printf '\t};\n\n' + printf '\tret = compose(tn, c);\n' + printf '\tret = compose(ret, ti);\n' + if [ -n "$(echo "$trans" | grep "m")" ]; then + printf '\tret = flipallcorners(ret);\n' + fi + printf '\n\treturn ret;\n}\n\n' + done +} + +genfuncs rm -f h48_to_src invert diff --git a/utils/gentransswitch.sh b/utils/gentransswitch.sh @@ -0,0 +1,8 @@ +#!/bin/sh + +printf '\tswitch (t) {\n' +for f in transform_??_???.txt; do + t="$(echo $f | sed 's/.*_// ; s/\.txt//')" + printf '\tcase %s:\n\t\treturn inline_trans_%s(c);\n' "$t" "$t" +done +printf '\t}\n'