From a89a14ef5da44684a16b204e7a70460cc8c4922a Mon Sep 17 00:00:00 2001 From: Thomas Voss Date: Fri, 21 Jun 2024 23:36:36 +0200 Subject: Basic constant folding implementation --- .../gmp-6.3.0/mpn/powerpc64/mode64/p7/aormul_2.asm | 135 ++++++++++++++++ .../gmp-6.3.0/mpn/powerpc64/mode64/p7/aors_n.asm | 128 +++++++++++++++ .../mpn/powerpc64/mode64/p7/aorsorrlsh1_n.asm | 43 +++++ .../mpn/powerpc64/mode64/p7/aorsorrlsh2_n.asm | 43 +++++ .../mpn/powerpc64/mode64/p7/aorsorrlshC_n.asm | 129 +++++++++++++++ .../gmp-6.3.0/mpn/powerpc64/mode64/p7/gcd_11.asm | 67 ++++++++ .../gmp-6.3.0/mpn/powerpc64/mode64/p7/gcd_22.asm | 146 +++++++++++++++++ .../gmp-6.3.0/mpn/powerpc64/mode64/p7/gmp-mparam.h | 175 +++++++++++++++++++++ 8 files changed, 866 insertions(+) create mode 100644 vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/aormul_2.asm create mode 100644 vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/aors_n.asm create mode 100644 vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/aorsorrlsh1_n.asm create mode 100644 vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/aorsorrlsh2_n.asm create mode 100644 vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/aorsorrlshC_n.asm create mode 100644 vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/gcd_11.asm create mode 100644 vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/gcd_22.asm create mode 100644 vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/gmp-mparam.h (limited to 'vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7') diff --git a/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/aormul_2.asm b/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/aormul_2.asm new file mode 100644 index 0000000..8731e01 --- /dev/null +++ b/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/aormul_2.asm @@ -0,0 +1,135 @@ +dnl PowerPC-64 mpn_mul_2 and mpn_addmul_2. + +dnl Copyright 2013 Free Software Foundation, Inc. + +dnl This file is part of the GNU MP Library. +dnl +dnl The GNU MP Library is free software; you can redistribute it and/or modify +dnl it under the terms of either: +dnl +dnl * the GNU Lesser General Public License as published by the Free +dnl Software Foundation; either version 3 of the License, or (at your +dnl option) any later version. +dnl +dnl or +dnl +dnl * the GNU General Public License as published by the Free Software +dnl Foundation; either version 2 of the License, or (at your option) any +dnl later version. +dnl +dnl or both in parallel, as here. +dnl +dnl The GNU MP Library is distributed in the hope that it will be useful, but +dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY +dnl or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +dnl for more details. +dnl +dnl You should have received copies of the GNU General Public License and the +dnl GNU Lesser General Public License along with the GNU MP Library. If not, +dnl see https://www.gnu.org/licenses/. + +include(`../config.m4') + +C cycles/limb cycles/limb +C mul_2 addmul_2 +C POWER3/PPC630 ? ? +C POWER4/PPC970 ? ? +C POWER5 ? ? +C POWER6 ? ? +C POWER7-SMT4 3 3 +C POWER7-SMT2 ? ? +C POWER7-SMT1 ? ? + +C INPUT PARAMETERS +define(`rp', `r3') +define(`up', `r4') +define(`n', `r5') +define(`vp', `r6') + +define(`cy0', `r10') +ifdef(`EXTRA_REGISTER', +` define(`cy1', EXTRA_REGISTER)', +` define(`cy1', `r31')') + +ifdef(`OPERATION_mul_2',` + define(`AM', `') + define(`ADDX', `addc') + define(`func', `mpn_mul_2') +') +ifdef(`OPERATION_addmul_2',` + define(`AM', `$1') + define(`ADDX', `adde') + define(`func', `mpn_addmul_2') +') + +MULFUNC_PROLOGUE(mpn_mul_2 mpn_addmul_2) + +ASM_START() +PROLOGUE(func) + +ifdef(`EXTRA_REGISTER',,` + std r31, -8(r1) +') + andi. r12, n, 1 + addi r0, n, 1 + srdi r0, r0, 1 + mtctr r0 + ld r11, 0(vp) C v0 + li cy0, 0 + ld r12, 8(vp) C v1 + li cy1, 0 + ld r5, 0(up) + beq L(lo0) + addi up, up, -8 + addi rp, rp, -8 + b L(lo1) + + ALIGN(32) +L(top): +AM(` ld r0, -8(rp)') + ld r5, 0(up) +AM(` addc r6, r6, r0') + ADDX r7, r7, r8 + addze r9, r9 + addc r6, r6, cy0 + adde cy0, r7, cy1 + std r6, -8(rp) + addze cy1, r9 +L(lo0): mulld r6, r11, r5 C v0 * u[i] weight 0 + mulhdu r7, r11, r5 C v0 * u[i] weight 1 + mulld r8, r12, r5 C v1 * u[i] weight 1 + mulhdu r9, r12, r5 C v1 * u[i] weight 2 +AM(` ld r0, 0(rp)') + ld r5, 8(up) +AM(` addc r6, r6, r0') + ADDX r7, r7, r8 + addze r9, r9 + addc r6, r6, cy0 + adde cy0, r7, cy1 + std r6, 0(rp) + addze cy1, r9 +L(lo1): mulld r6, r11, r5 C v0 * u[i] weight 0 + mulhdu r7, r11, r5 C v0 * u[i] weight 1 + addi up, up, 16 + addi rp, rp, 16 + mulld r8, r12, r5 C v1 * u[i] weight 1 + mulhdu r9, r12, r5 C v1 * u[i] weight 2 + bdnz L(top) + +L(end): +AM(` ld r0, -8(rp)') +AM(` addc r6, r6, r0') + ADDX r7, r7, r8 + addze r9, r9 + addc r6, r6, cy0 + std r6, -8(rp) + adde cy0, r7, cy1 + addze cy1, r9 + std cy0, 0(rp) + mr r3, cy1 + +ifdef(`EXTRA_REGISTER',,` + ld r31, -8(r1) +') + blr +EPILOGUE() diff --git a/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/aors_n.asm b/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/aors_n.asm new file mode 100644 index 0000000..857c701 --- /dev/null +++ b/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/aors_n.asm @@ -0,0 +1,128 @@ +dnl PowerPC-64 mpn_add_n, mpn_sub_n optimised for POWER7. + +dnl Copyright 2013 Free Software Foundation, Inc. + +dnl This file is part of the GNU MP Library. +dnl +dnl The GNU MP Library is free software; you can redistribute it and/or modify +dnl it under the terms of either: +dnl +dnl * the GNU Lesser General Public License as published by the Free +dnl Software Foundation; either version 3 of the License, or (at your +dnl option) any later version. +dnl +dnl or +dnl +dnl * the GNU General Public License as published by the Free Software +dnl Foundation; either version 2 of the License, or (at your option) any +dnl later version. +dnl +dnl or both in parallel, as here. +dnl +dnl The GNU MP Library is distributed in the hope that it will be useful, but +dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY +dnl or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +dnl for more details. +dnl +dnl You should have received copies of the GNU General Public License and the +dnl GNU Lesser General Public License along with the GNU MP Library. If not, +dnl see https://www.gnu.org/licenses/. + +include(`../config.m4') + +C cycles/limb +C POWER3/PPC630 ? +C POWER4/PPC970 ? +C POWER5 ? +C POWER6 ? +C POWER7 2.18 + +C This is a tad bit slower than the cnd_aors_n.asm code, which is of course an +C anomaly. + +ifdef(`OPERATION_add_n',` + define(ADDSUBC, adde) + define(ADDSUB, addc) + define(func, mpn_add_n) + define(func_nc, mpn_add_nc) + define(GENRVAL, `addi r3, r3, 1') + define(SETCBR, `addic r0, $1, -1') + define(CLRCB, `addic r0, r0, 0') +') +ifdef(`OPERATION_sub_n',` + define(ADDSUBC, subfe) + define(ADDSUB, subfc) + define(func, mpn_sub_n) + define(func_nc, mpn_sub_nc) + define(GENRVAL, `neg r3, r3') + define(SETCBR, `subfic r0, $1, 0') + define(CLRCB, `addic r0, r1, -1') +') + +MULFUNC_PROLOGUE(mpn_add_n mpn_add_nc mpn_sub_n mpn_sub_nc) + +C INPUT PARAMETERS +define(`rp', `r3') +define(`up', `r4') +define(`vp', `r5') +define(`n', `r6') + +ASM_START() +PROLOGUE(func_nc) + SETCBR(r7) + b L(ent) +EPILOGUE() + +PROLOGUE(func) + CLRCB +L(ent): + andi. r7, n, 1 + beq L(bx0) + +L(bx1): ld r7, 0(up) + ld r9, 0(vp) + ADDSUBC r11, r9, r7 + std r11, 0(rp) + cmpldi cr6, n, 1 + beq cr6, L(end) + addi up, up, 8 + addi vp, vp, 8 + addi rp, rp, 8 + +L(bx0): addi r0, n, 2 C compute branch... + srdi r0, r0, 2 C ...count + mtctr r0 + + andi. r7, n, 2 + bne L(mid) + + addi up, up, 16 + addi vp, vp, 16 + addi rp, rp, 16 + + ALIGN(32) +L(top): ld r6, -16(up) + ld r7, -8(up) + ld r8, -16(vp) + ld r9, -8(vp) + ADDSUBC r10, r8, r6 + ADDSUBC r11, r9, r7 + std r10, -16(rp) + std r11, -8(rp) +L(mid): ld r6, 0(up) + ld r7, 8(up) + ld r8, 0(vp) + ld r9, 8(vp) + ADDSUBC r10, r8, r6 + ADDSUBC r11, r9, r7 + std r10, 0(rp) + std r11, 8(rp) + addi up, up, 32 + addi vp, vp, 32 + addi rp, rp, 32 + bdnz L(top) + +L(end): subfe r3, r0, r0 C -cy + GENRVAL + blr +EPILOGUE() diff --git a/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/aorsorrlsh1_n.asm b/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/aorsorrlsh1_n.asm new file mode 100644 index 0000000..ddf5fd8 --- /dev/null +++ b/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/aorsorrlsh1_n.asm @@ -0,0 +1,43 @@ +dnl PowerPC-64 mpn_addlsh1_n, mpn_sublsh1_n, mpn_rsblsh1_n. + +dnl Copyright 2003, 2005, 2009, 2010, 2013 Free Software Foundation, Inc. + +dnl This file is part of the GNU MP Library. +dnl +dnl The GNU MP Library is free software; you can redistribute it and/or modify +dnl it under the terms of either: +dnl +dnl * the GNU Lesser General Public License as published by the Free +dnl Software Foundation; either version 3 of the License, or (at your +dnl option) any later version. +dnl +dnl or +dnl +dnl * the GNU General Public License as published by the Free Software +dnl Foundation; either version 2 of the License, or (at your option) any +dnl later version. +dnl +dnl or both in parallel, as here. +dnl +dnl The GNU MP Library is distributed in the hope that it will be useful, but +dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY +dnl or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +dnl for more details. +dnl +dnl You should have received copies of the GNU General Public License and the +dnl GNU Lesser General Public License along with the GNU MP Library. If not, +dnl see https://www.gnu.org/licenses/. + +include(`../config.m4') + + +define(LSH, 1) +define(RSH, 63) + +ifdef(`OPERATION_addlsh1_n',`define(`DO_add')') +ifdef(`OPERATION_sublsh1_n',`define(`DO_sub')') +ifdef(`OPERATION_rsblsh1_n',`define(`DO_rsb')') + +MULFUNC_PROLOGUE(mpn_addlsh1_n mpn_sublsh1_n mpn_rsblsh1_n) + +include_mpn(`powerpc64/mode64/p7/aorsorrlshC_n.asm') diff --git a/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/aorsorrlsh2_n.asm b/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/aorsorrlsh2_n.asm new file mode 100644 index 0000000..3f9d88d --- /dev/null +++ b/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/aorsorrlsh2_n.asm @@ -0,0 +1,43 @@ +dnl PowerPC-64 mpn_addlsh2_n, mpn_sublsh2_n, mpn_rsblsh2_n. + +dnl Copyright 2003, 2005, 2009, 2010, 2013 Free Software Foundation, Inc. + +dnl This file is part of the GNU MP Library. +dnl +dnl The GNU MP Library is free software; you can redistribute it and/or modify +dnl it under the terms of either: +dnl +dnl * the GNU Lesser General Public License as published by the Free +dnl Software Foundation; either version 3 of the License, or (at your +dnl option) any later version. +dnl +dnl or +dnl +dnl * the GNU General Public License as published by the Free Software +dnl Foundation; either version 2 of the License, or (at your option) any +dnl later version. +dnl +dnl or both in parallel, as here. +dnl +dnl The GNU MP Library is distributed in the hope that it will be useful, but +dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY +dnl or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +dnl for more details. +dnl +dnl You should have received copies of the GNU General Public License and the +dnl GNU Lesser General Public License along with the GNU MP Library. If not, +dnl see https://www.gnu.org/licenses/. + +include(`../config.m4') + + +define(LSH, 2) +define(RSH, 62) + +ifdef(`OPERATION_addlsh2_n',`define(`DO_add')') +ifdef(`OPERATION_sublsh2_n',`define(`DO_sub')') +ifdef(`OPERATION_rsblsh2_n',`define(`DO_rsb')') + +MULFUNC_PROLOGUE(mpn_addlsh2_n mpn_sublsh2_n mpn_rsblsh2_n) + +include_mpn(`powerpc64/mode64/p7/aorsorrlshC_n.asm') diff --git a/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/aorsorrlshC_n.asm b/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/aorsorrlshC_n.asm new file mode 100644 index 0000000..5251202 --- /dev/null +++ b/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/aorsorrlshC_n.asm @@ -0,0 +1,129 @@ +dnl PowerPC-64 mpn_addlshC_n, mpn_sublshC_n, mpn_rsblshC_n. + +dnl Copyright 2003, 2005, 2009, 2010, 2013 Free Software Foundation, Inc. + +dnl This file is part of the GNU MP Library. +dnl +dnl The GNU MP Library is free software; you can redistribute it and/or modify +dnl it under the terms of either: +dnl +dnl * the GNU Lesser General Public License as published by the Free +dnl Software Foundation; either version 3 of the License, or (at your +dnl option) any later version. +dnl +dnl or +dnl +dnl * the GNU General Public License as published by the Free Software +dnl Foundation; either version 2 of the License, or (at your option) any +dnl later version. +dnl +dnl or both in parallel, as here. +dnl +dnl The GNU MP Library is distributed in the hope that it will be useful, but +dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY +dnl or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +dnl for more details. +dnl +dnl You should have received copies of the GNU General Public License and the +dnl GNU Lesser General Public License along with the GNU MP Library. If not, +dnl see https://www.gnu.org/licenses/. + +C cycles/limb +C POWER3/PPC630 ? +C POWER4/PPC970 ? +C POWER5 ? +C POWER6 ? +C POWER7 2.5 + +C INPUT PARAMETERS +define(`rp', `r3') +define(`up', `r4') +define(`vp', `r5') +define(`n', `r6') + +ifdef(`DO_add', ` + define(`ADDSUBC', `addc $1, $2, $3') + define(`ADDSUBE', `adde $1, $2, $3') + define(INITCY, `addic $1, r1, 0') + define(RETVAL, `addze r3, $1') + define(`func', mpn_addlsh`'LSH`'_n)') +ifdef(`DO_sub', ` + define(`ADDSUBC', `subfc $1, $2, $3') + define(`ADDSUBE', `subfe $1, $2, $3') + define(INITCY, `addic $1, r1, -1') + define(RETVAL, `subfze r3, $1 + neg r3, r3') + define(`func', mpn_sublsh`'LSH`'_n)') +ifdef(`DO_rsb', ` + define(`ADDSUBC', `subfc $1, $3, $2') + define(`ADDSUBE', `subfe $1, $3, $2') + define(INITCY, `addic $1, r1, -1') + define(RETVAL, `addme r3, $1') + define(`func', mpn_rsblsh`'LSH`'_n)') + +define(`s0', `r0') define(`s1', `r9') +define(`u0', `r6') define(`u1', `r7') +define(`v0', `r10') define(`v1', `r11') + + +ASM_START() +PROLOGUE(func) + rldic r7, n, 3, 59 + add up, up, r7 + add vp, vp, r7 + add rp, rp, r7 + +ifdef(`DO_add', ` + addic r0, n, 3 C set cy flag as side effect +',` + subfc r0, r0, r0 C set cy flag + addi r0, n, 3 +') + srdi r0, r0, 2 + mtctr r0 + + andi. r0, n, 1 + beq L(bx0) + +L(bx1): andi. r0, n, 2 + li s0, 0 + bne L(lo3) + b L(lo1) + +L(bx0): andi. r0, n, 2 + li s1, 0 + bne L(lo2) + + ALIGN(32) +L(top): addi rp, rp, 32 + ld v0, 0(vp) + addi vp, vp, 32 + rldimi s1, v0, LSH, 0 + ld u0, 0(up) + addi up, up, 32 + srdi s0, v0, RSH + ADDSUBE(s1, s1, u0) + std s1, -32(rp) +L(lo3): ld v1, -24(vp) + rldimi s0, v1, LSH, 0 + ld u1, -24(up) + srdi s1, v1, RSH + ADDSUBE(s0, s0, u1) + std s0, -24(rp) +L(lo2): ld v0, -16(vp) + rldimi s1, v0, LSH, 0 + ld u0, -16(up) + srdi s0, v0, RSH + ADDSUBE(s1, s1, u0) + std s1, -16(rp) +L(lo1): ld v1, -8(vp) + rldimi s0, v1, LSH, 0 + ld u1, -8(up) + srdi s1, v1, RSH + ADDSUBE(s0, s0, u1) + std s0, -8(rp) + bdnz L(top) C decrement CTR and loop back + + RETVAL( s1) + blr +EPILOGUE() diff --git a/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/gcd_11.asm b/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/gcd_11.asm new file mode 100644 index 0000000..f04e896 --- /dev/null +++ b/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/gcd_11.asm @@ -0,0 +1,67 @@ +dnl PowerPC-64 mpn_gcd_11. + +dnl Copyright 2000-2002, 2005, 2009, 2011-2013 Free Software Foundation, Inc. + +dnl This file is part of the GNU MP Library. +dnl +dnl The GNU MP Library is free software; you can redistribute it and/or modify +dnl it under the terms of either: +dnl +dnl * the GNU Lesser General Public License as published by the Free +dnl Software Foundation; either version 3 of the License, or (at your +dnl option) any later version. +dnl +dnl or +dnl +dnl * the GNU General Public License as published by the Free Software +dnl Foundation; either version 2 of the License, or (at your option) any +dnl later version. +dnl +dnl or both in parallel, as here. +dnl +dnl The GNU MP Library is distributed in the hope that it will be useful, but +dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY +dnl or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +dnl for more details. +dnl +dnl You should have received copies of the GNU General Public License and the +dnl GNU Lesser General Public License along with the GNU MP Library. If not, +dnl see https://www.gnu.org/licenses/. + +include(`../config.m4') + +C cycles/bit (approx) +C POWER3/PPC630 - +C POWER4/PPC970 - +C POWER5 - +C POWER6 - +C POWER7 7.6 obsolete +C POWER8 ? +C POWER9 ? +C Numbers measured with: speed -CD -s16-64 -t48 mpn_gcd_1 + +C INPUT PARAMETERS +define(`u0', `r3') +define(`v0', `r4') + +define(`cnt', `r9')dnl + +ASM_START() +PROLOGUE(mpn_gcd_11) + li r12, 63 + b L(odd) + + ALIGN(16) +L(top): and r8, r11, r10 C isolate lsb + cntlzd cnt, r8 + isel v0, u0, v0, 29 C v = min(u,v) + isel u0, r10, r11, 29 C u = |u - v| + subf cnt, cnt, r12 C cnt = 63-cnt + srd u0, u0, cnt +L(odd): cmpld cr7, v0, u0 + subf r10, u0, v0 C r10 = v - u + subf r11, v0, u0 C r11 = u - v + bne cr7, L(top) + +L(end): blr +EPILOGUE() diff --git a/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/gcd_22.asm b/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/gcd_22.asm new file mode 100644 index 0000000..ade30e4 --- /dev/null +++ b/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/gcd_22.asm @@ -0,0 +1,146 @@ +dnl PowerPC-64 mpn_gcd_22 optimised for POWER7 and POWER8. + +dnl Copyright 2000-2002, 2005, 2009, 2011-2013, 2019 Free Software Foundation, +dnl Inc. + +dnl This file is part of the GNU MP Library. +dnl +dnl The GNU MP Library is free software; you can redistribute it and/or modify +dnl it under the terms of either: +dnl +dnl * the GNU Lesser General Public License as published by the Free +dnl Software Foundation; either version 3 of the License, or (at your +dnl option) any later version. +dnl +dnl or +dnl +dnl * the GNU General Public License as published by the Free Software +dnl Foundation; either version 2 of the License, or (at your option) any +dnl later version. +dnl +dnl or both in parallel, as here. +dnl +dnl The GNU MP Library is distributed in the hope that it will be useful, but +dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY +dnl or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +dnl for more details. +dnl +dnl You should have received copies of the GNU General Public License and the +dnl GNU Lesser General Public License along with the GNU MP Library. If not, +dnl see https://www.gnu.org/licenses/. + +include(`../config.m4') + +C cycles/bit (approx) +C POWER3/PPC630 - +C POWER4/PPC970 - +C POWER5 - +C POWER6 - +C POWER7 12.3 +C POWER8 13.4 +C POWER9 10.6 + +C We define SLOW if this target uses a slow struct return mechanism, with +C r3 as an implicit parameter for the struct pointer. +undefine(`SLOW')dnl +ifdef(`AIX',`define(`SLOW',`due to AIX')',` + ifdef(`DARWIN',,` + ifdef(`ELFv2_ABI',,`define(`SLOW',`due to ELFv1')')dnl + ') +') + +ifdef(`SLOW',` +define(`IFSLOW', `$1') +define(`u1', `r4') +define(`u0', `r5') +define(`v1', `r6') +define(`v0', `r7') +',` +define(`IFSLOW', `') +define(`u1', `r3') +define(`u0', `r4') +define(`v1', `r5') +define(`v0', `r6') +') + +define(`tmp', `r0') +define(`t0', `r8') +define(`t1', `r9') +define(`s0', `r10') +define(`s1', `r11') +define(`cnt', `r12') + +ASM_START() +PROLOGUE(mpn_gcd_22) +L(top): subfc. t0, v0, u0 C 0 12 + beq cr0, L(lowz) + subfe t1, v1, u1 C 2 14 + subfe. tmp, tmp, tmp C 4 set cr0 from the carry bit + subfc s0, u0, v0 C 0 + subfe s1, u1, v1 C 2 + +L(bck): and tmp, s0, t0 C 2 + cntlzd cnt, tmp C 4 + addi tmp, cnt, 1 C 6 + subfic cnt, cnt, 63 C 6 + + isel v0, v0, u0, 2 C 6 use condition set by subfe + isel v1, v1, u1, 2 C 6 + isel u0, t0, s0, 2 C 6 + isel u1, t1, s1, 2 C 6 + + srd u0, u0, cnt C 8 + sld tmp, u1, tmp C 8 + srd u1, u1, cnt C 8 + or u0, u0, tmp C 10 + + or. r0, u1, v1 C 10 + bne L(top) + + + li r0, 63 + b L(odd) + ALIGN(16) +L(top1):isel v0, u0, v0, 29 C v = min(u,v) + isel u0, r10, r11, 29 C u = |u - v| + subf cnt, cnt, r0 C cnt = 63-cnt + srd u0, u0, cnt +L(odd): subf r10, u0, v0 C r10 = v - u + subf r11, v0, u0 C r11 = u - v + cmpld cr7, v0, u0 + and r8, r11, r10 C isolate lsb + cntlzd cnt, r8 + bne cr7, L(top1) + +ifdef(`SLOW',` + std v0, 0(r3) + std r10, 8(r3) C zero +',` + mr r3, v0 + li r4, 0 +') + blr + + +L(lowz):C We come here when v0 - u0 = 0 + C 1. If v1 - u1 = 0, then gcd is u = v. + C 2. Else compute gcd_21({v1,v0}, |u1-v1|) + subfc. t0, v1, u1 C 2 8 + beq L(end) + li t1, 0 + subfe. tmp, tmp, tmp C 4 set cr0 from the carry bit + subf s0, u1, v1 C 2 + li s1, 0 + b L(bck) + +L(end): +ifdef(`SLOW',` + std v0, 0(r3) + std v1, 8(r3) + blr +',` + mr r3, v0 + mr r4, v1 + blr +') +EPILOGUE() diff --git a/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/gmp-mparam.h b/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/gmp-mparam.h new file mode 100644 index 0000000..9da4080 --- /dev/null +++ b/vendor/gmp-6.3.0/mpn/powerpc64/mode64/p7/gmp-mparam.h @@ -0,0 +1,175 @@ +/* POWER7 gmp-mparam.h -- Compiler/machine parameter header file. + +Copyright 2019 Free Software Foundation, Inc. + +This file is part of the GNU MP Library. + +The GNU MP Library is free software; you can redistribute it and/or modify +it under the terms of either: + + * the GNU Lesser General Public License as published by the Free + Software Foundation; either version 3 of the License, or (at your + option) any later version. + +or + + * the GNU General Public License as published by the Free Software + Foundation; either version 2 of the License, or (at your option) any + later version. + +or both in parallel, as here. + +The GNU MP Library is distributed in the hope that it will be useful, but +WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY +or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License +for more details. + +You should have received copies of the GNU General Public License and the +GNU Lesser General Public License along with the GNU MP Library. If not, +see https://www.gnu.org/licenses/. */ + +#define GMP_LIMB_BITS 64 +#define GMP_LIMB_BYTES 8 + +/* 3720 MHz POWER7/SMT4 */ +/* FFT tuning limit = 0.5 M */ +/* Generated by tuneup.c, 2019-10-02, gcc 4.8 */ + +#define MOD_1_NORM_THRESHOLD 0 /* always */ +#define MOD_1_UNNORM_THRESHOLD 0 /* always */ +#define MOD_1N_TO_MOD_1_1_THRESHOLD 6 +#define MOD_1U_TO_MOD_1_1_THRESHOLD 4 +#define MOD_1_1_TO_MOD_1_2_THRESHOLD 8 +#define MOD_1_2_TO_MOD_1_4_THRESHOLD 16 +#define PREINV_MOD_1_TO_MOD_1_THRESHOLD 13 +#define USE_PREINV_DIVREM_1 0 +/* From gcc110.osuosl.org, 2023-07-27 */ +#define DIV_QR_1N_PI1_METHOD 3 /* 8.45% faster than 4 */ +#define DIV_QR_1_NORM_THRESHOLD 1 +#define DIV_QR_1_UNNORM_THRESHOLD 1 +#define DIV_QR_2_PI2_THRESHOLD MP_SIZE_T_MAX /* never */ +#define DIVEXACT_1_THRESHOLD 0 /* always (native) */ +#define BMOD_1_TO_MOD_1_THRESHOLD 27 + +#define DIV_1_VS_MUL_1_PERCENT 341 + +#define MUL_TOOM22_THRESHOLD 22 +#define MUL_TOOM33_THRESHOLD 71 +#define MUL_TOOM44_THRESHOLD 196 +#define MUL_TOOM6H_THRESHOLD 298 +#define MUL_TOOM8H_THRESHOLD 406 + +#define MUL_TOOM32_TO_TOOM43_THRESHOLD 81 +#define MUL_TOOM32_TO_TOOM53_THRESHOLD 140 +#define MUL_TOOM42_TO_TOOM53_THRESHOLD 132 +#define MUL_TOOM42_TO_TOOM63_THRESHOLD 139 +#define MUL_TOOM43_TO_TOOM54_THRESHOLD 120 + +#define SQR_BASECASE_THRESHOLD 0 /* always (native) */ +#define SQR_TOOM2_THRESHOLD 32 +#define SQR_TOOM3_THRESHOLD 105 +#define SQR_TOOM4_THRESHOLD 190 +#define SQR_TOOM6_THRESHOLD 318 +#define SQR_TOOM8_THRESHOLD 547 + +#define MULMID_TOOM42_THRESHOLD 56 + +#define MULMOD_BNM1_THRESHOLD 18 +#define SQRMOD_BNM1_THRESHOLD 20 + +#define MUL_FFT_MODF_THRESHOLD 436 /* k = 5 */ +#define MUL_FFT_TABLE3 \ + { { 436, 5}, { 21, 6}, { 21, 7}, { 11, 6}, \ + { 23, 7}, { 12, 6}, { 25, 7}, { 21, 8}, \ + { 11, 7}, { 25, 8}, { 13, 7}, { 28, 8}, \ + { 15, 7}, { 33, 8}, { 17, 7}, { 35, 8}, \ + { 19, 7}, { 39, 8}, { 21, 9}, { 11, 8}, \ + { 29, 9}, { 15, 8}, { 35, 9}, { 19, 8}, \ + { 41, 9}, { 23, 8}, { 49, 9}, { 27,10}, \ + { 15, 9}, { 31, 8}, { 63, 9}, { 43,10}, \ + { 23, 9}, { 55,11}, { 15,10}, { 31, 9}, \ + { 67,10}, { 39, 9}, { 79,10}, { 47, 9}, \ + { 95,10}, { 55,11}, { 31,10}, { 63, 9}, \ + { 127,10}, { 79,11}, { 47,10}, { 103,12}, \ + { 31,11}, { 63,10}, { 135,11}, { 79,10}, \ + { 159,11}, { 95,10}, { 191, 9}, { 383,11}, \ + { 111,12}, { 63,11}, { 127,10}, { 255, 9}, \ + { 511,11}, { 143,10}, { 287, 9}, { 575,11}, \ + { 159,10}, { 319,12}, { 95,11}, { 191,10}, \ + { 383, 9}, { 767,11}, { 207,10}, { 415,13}, \ + { 8192,14}, { 16384,15}, { 32768,16}, { 65536,17}, \ + { 131072,18}, { 262144,19}, { 524288,20}, {1048576,21}, \ + {2097152,22}, {4194304,23}, {8388608,24} } +#define MUL_FFT_TABLE3_SIZE 83 +#define MUL_FFT_THRESHOLD 4736 + +#define SQR_FFT_MODF_THRESHOLD 368 /* k = 5 */ +#define SQR_FFT_TABLE3 \ + { { 368, 5}, { 19, 6}, { 10, 5}, { 21, 6}, \ + { 21, 7}, { 11, 6}, { 23, 7}, { 12, 6}, \ + { 25, 7}, { 13, 6}, { 27, 7}, { 25, 8}, \ + { 13, 7}, { 28, 8}, { 15, 7}, { 32, 8}, \ + { 17, 7}, { 35, 8}, { 19, 7}, { 39, 8}, \ + { 21, 9}, { 11, 8}, { 29, 9}, { 15, 8}, \ + { 35, 9}, { 19, 8}, { 41, 9}, { 23, 8}, \ + { 47, 9}, { 27,10}, { 15, 9}, { 31, 8}, \ + { 63, 9}, { 39,10}, { 23, 9}, { 51,11}, \ + { 15,10}, { 31, 9}, { 67,10}, { 39, 9}, \ + { 79,10}, { 47, 9}, { 95,10}, { 55,11}, \ + { 31,10}, { 79,11}, { 47,10}, { 95,12}, \ + { 31,11}, { 63,10}, { 135,11}, { 79,10}, \ + { 159, 9}, { 319,11}, { 95,10}, { 191, 9}, \ + { 383,11}, { 111,12}, { 63,11}, { 127,10}, \ + { 255, 9}, { 511,11}, { 143,10}, { 287, 9}, \ + { 575,10}, { 303,11}, { 159,10}, { 319, 9}, \ + { 639,12}, { 95,11}, { 191,10}, { 383, 9}, \ + { 767,13}, { 8192,14}, { 16384,15}, { 32768,16}, \ + { 65536,17}, { 131072,18}, { 262144,19}, { 524288,20}, \ + {1048576,21}, {2097152,22}, {4194304,23}, {8388608,24} } +#define SQR_FFT_TABLE3_SIZE 84 +#define SQR_FFT_THRESHOLD 3264 + +#define MULLO_BASECASE_THRESHOLD 3 +#define MULLO_DC_THRESHOLD 35 +#define MULLO_MUL_N_THRESHOLD 9449 +#define SQRLO_BASECASE_THRESHOLD 3 +#define SQRLO_DC_THRESHOLD 119 +#define SQRLO_SQR_THRESHOLD 6440 + +#define DC_DIV_QR_THRESHOLD 33 +#define DC_DIVAPPR_Q_THRESHOLD 124 +#define DC_BDIV_QR_THRESHOLD 62 +#define DC_BDIV_Q_THRESHOLD 144 + +#define INV_MULMOD_BNM1_THRESHOLD 67 +#define INV_NEWTON_THRESHOLD 123 +#define INV_APPR_THRESHOLD 123 + +#define BINV_NEWTON_THRESHOLD 284 +#define REDC_1_TO_REDC_2_THRESHOLD 18 +#define REDC_2_TO_REDC_N_THRESHOLD 109 + +#define MU_DIV_QR_THRESHOLD 1387 +#define MU_DIVAPPR_Q_THRESHOLD 1334 +#define MUPI_DIV_QR_THRESHOLD 50 +#define MU_BDIV_QR_THRESHOLD 1308 +#define MU_BDIV_Q_THRESHOLD 1499 + +#define POWM_SEC_TABLE 1,23,121,579,642 + +#define GET_STR_DC_THRESHOLD 11 +#define GET_STR_PRECOMPUTE_THRESHOLD 18 +#define SET_STR_DC_THRESHOLD 1562 +#define SET_STR_PRECOMPUTE_THRESHOLD 3100 + +#define FAC_DSC_THRESHOLD 774 +#define FAC_ODD_THRESHOLD 25 + +#define MATRIX22_STRASSEN_THRESHOLD 18 +#define HGCD2_DIV1_METHOD 5 /* 3.27% faster than 3 */ +#define HGCD_THRESHOLD 118 +#define HGCD_APPR_THRESHOLD 150 +#define HGCD_REDUCE_THRESHOLD 3014 +#define GCD_DC_THRESHOLD 386 +#define GCDEXT_DC_THRESHOLD 365 +#define JACOBI_BASE_METHOD 4 /* 27.64% faster than 1 */ -- cgit v1.2.3