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 --- vendor/gmp-6.3.0/mpn/powerpc64/p7/copyd.asm | 128 ++++++++++++++++++++++++ vendor/gmp-6.3.0/mpn/powerpc64/p7/copyi.asm | 129 +++++++++++++++++++++++++ vendor/gmp-6.3.0/mpn/powerpc64/p7/hamdist.asm | 110 +++++++++++++++++++++ vendor/gmp-6.3.0/mpn/powerpc64/p7/popcount.asm | 90 +++++++++++++++++ 4 files changed, 457 insertions(+) create mode 100644 vendor/gmp-6.3.0/mpn/powerpc64/p7/copyd.asm create mode 100644 vendor/gmp-6.3.0/mpn/powerpc64/p7/copyi.asm create mode 100644 vendor/gmp-6.3.0/mpn/powerpc64/p7/hamdist.asm create mode 100644 vendor/gmp-6.3.0/mpn/powerpc64/p7/popcount.asm (limited to 'vendor/gmp-6.3.0/mpn/powerpc64/p7') diff --git a/vendor/gmp-6.3.0/mpn/powerpc64/p7/copyd.asm b/vendor/gmp-6.3.0/mpn/powerpc64/p7/copyd.asm new file mode 100644 index 0000000..f04ca58 --- /dev/null +++ b/vendor/gmp-6.3.0/mpn/powerpc64/p7/copyd.asm @@ -0,0 +1,128 @@ +dnl PowerPC-64 mpn_copyd. + +dnl Copyright 2004, 2005, 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 1.25 +C POWER7 1.09 + +C INPUT PARAMETERS +define(`rp', `r3') +define(`up', `r4') +define(`n', `r5') + +ASM_START() +PROLOGUE(mpn_copyd) + +ifdef(`HAVE_ABI_mode32', +` rldicl n, n, 0,32') + + sldi r0, n, 3 + add up, up, r0 C point at u[] end + add rp, rp, r0 C point at r[] end + + cmpdi cr0, n, 4 + blt L(sml) + + addi r10, n, 4 + srdi r10, r10, 3 + mtctr r10 + + andi. r0, n, 1 + rlwinm r11, n, 0,30,30 + rlwinm r12, n, 0,29,29 + cmpdi cr6, r11, 0 + cmpdi cr7, r12, 0 + + beq cr0, L(xx0) +L(xx1): ld r6, -8(up) + addi up, up, -8 + std r6, -8(rp) + addi rp, rp, -8 + +L(xx0): bne cr6, L(x10) +L(x00): ld r6, -8(up) + ld r7, -16(up) + bne cr7, L(100) +L(000): addi rp, rp, 32 + b L(lo0) +L(100): addi up, up, 32 + b L(lo4) +L(x10): ld r8, -8(up) + ld r9, -16(up) + bne cr7, L(110) +L(010): addi up, up, -16 + addi rp, rp, 16 + b L(lo2) +L(110): addi up, up, 16 + addi rp, rp, 48 + b L(lo6) + +L(sml): cmpdi cr0, n, 0 + beqlr- cr0 + mtctr n +L(t): ld r6, -8(up) + addi up, up, -8 + std r6, -8(rp) + addi rp, rp, -8 + bdnz L(t) + blr + + ALIGN(32) +L(top): std r6, -8(rp) + std r7, -16(rp) +L(lo2): ld r6, -8(up) + ld r7, -16(up) + std r8, -24(rp) + std r9, -32(rp) +L(lo0): ld r8, -24(up) + ld r9, -32(up) + std r6, -40(rp) + std r7, -48(rp) +L(lo6): ld r6, -40(up) + ld r7, -48(up) + std r8, -56(rp) + std r9, -64(rp) + addi rp, rp, -64 +L(lo4): ld r8, -56(up) + ld r9, -64(up) + addi up, up, -64 + bdnz L(top) + +L(end): std r6, -8(rp) + std r7, -16(rp) + std r8, -24(rp) + std r9, -32(rp) + blr +EPILOGUE() diff --git a/vendor/gmp-6.3.0/mpn/powerpc64/p7/copyi.asm b/vendor/gmp-6.3.0/mpn/powerpc64/p7/copyi.asm new file mode 100644 index 0000000..854cf9f --- /dev/null +++ b/vendor/gmp-6.3.0/mpn/powerpc64/p7/copyi.asm @@ -0,0 +1,129 @@ +dnl PowerPC-64 mpn_copyi. + +dnl Copyright 2004, 2005, 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 1.25 +C POWER7 1.09 + +C INPUT PARAMETERS +define(`rp', `r3') +define(`up', `r4') +define(`n', `r5') + +C TODO +C * Try rolling the two loop leading std to the end, allowing the code to +C handle also n = 2. +C * Consider using 4 pointers, schedule ptr update early wrt use. + +ASM_START() +PROLOGUE(mpn_copyi) + +ifdef(`HAVE_ABI_mode32', +` rldicl n, n, 0,32') + + cmpdi cr0, n, 4 + blt L(sml) + + addi r10, n, 4 + srdi r10, r10, 3 + mtctr r10 + + andi. r0, n, 1 + rlwinm r11, n, 0,30,30 + rlwinm r12, n, 0,29,29 + cmpdi cr6, r11, 0 + cmpdi cr7, r12, 0 + + beq cr0, L(xx0) +L(xx1): ld r6, 0(up) + addi up, up, 8 + std r6, 0(rp) + addi rp, rp, 8 + +L(xx0): bne cr6, L(x10) +L(x00): ld r6, 0(up) + ld r7, 8(up) + bne cr7, L(100) +L(000): addi rp, rp, -32 + b L(lo0) +L(100): addi up, up, -32 + b L(lo4) +L(x10): ld r8, 0(up) + ld r9, 8(up) + bne cr7, L(110) +L(010): addi up, up, 16 + addi rp, rp, -16 + b L(lo2) +L(110): addi up, up, -16 + addi rp, rp, -48 + b L(lo6) + +L(sml): cmpdi cr0, n, 0 + beqlr- cr0 + mtctr n +L(t): ld r6, 0(up) + addi up, up, 8 + std r6, 0(rp) + addi rp, rp, 8 + bdnz L(t) + blr + + ALIGN(32) +L(top): std r6, 0(rp) + std r7, 8(rp) +L(lo2): ld r6, 0(up) + ld r7, 8(up) + std r8, 16(rp) + std r9, 24(rp) +L(lo0): ld r8, 16(up) + ld r9, 24(up) + std r6, 32(rp) + std r7, 40(rp) +L(lo6): ld r6, 32(up) + ld r7, 40(up) + std r8, 48(rp) + std r9, 56(rp) + addi rp, rp, 64 +L(lo4): ld r8, 48(up) + ld r9, 56(up) + addi up, up, 64 + bdnz L(top) + +L(end): std r6, 0(rp) + std r7, 8(rp) + std r8, 16(rp) + std r9, 24(rp) + blr +EPILOGUE() diff --git a/vendor/gmp-6.3.0/mpn/powerpc64/p7/hamdist.asm b/vendor/gmp-6.3.0/mpn/powerpc64/p7/hamdist.asm new file mode 100644 index 0000000..960b3bc --- /dev/null +++ b/vendor/gmp-6.3.0/mpn/powerpc64/p7/hamdist.asm @@ -0,0 +1,110 @@ +dnl PowerPC-64 mpn_hamdist. + +dnl Copyright 2012, 2017 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 2.87 + +define(`up', r3) +define(`vp', r4) +define(`n', r5) + +ASM_START() +PROLOGUE(mpn_hamdist) + std r30, -16(r1) + std r31, -8(r1) + + addi r0, n, 1 +ifdef(`HAVE_ABI_mode32', +` rldicl r0, r0, 63,33', C ...branch count +` srdi r0, r0, 1') C ...for ctr + mtctr r0 + + andi. r0, n, 1 + + li r0, 0 + li r12, 0 + + beq L(evn) + +L(odd): ld r6, 0(up) + addi up, up, 8 + ld r8, 0(vp) + addi vp, vp, 8 + xor r10, r6, r8 + popcntd(r0, r10) + bdz L(e1) + +L(evn): ld r6, 0(up) + ld r8, 0(vp) + ld r7, 8(up) + ld r9, 8(vp) + xor r10, r6, r8 + addi up, up, 16 + addi vp, vp, 16 + li r30, 0 + li r31, 0 + bdz L(end) + + nop + nop +C ALIGN(16) +L(top): add r0, r0, r30 + ld r6, 0(up) + ld r8, 0(vp) + xor r11, r7, r9 + popcntd(r30, r10) + add r12, r12, r31 + ld r7, 8(up) + ld r9, 8(vp) + xor r10, r6, r8 + popcntd(r31, r11) + addi up, up, 16 + addi vp, vp, 16 + bdnz L(top) + +L(end): add r0, r0, r30 + xor r11, r7, r9 + popcntd(r30, r10) + add r12, r12, r31 + popcntd(r31, r11) + + add r0, r0, r30 + add r12, r12, r31 +L(e1): add r3, r0, r12 + ld r30, -16(r1) + ld r31, -8(r1) + blr +EPILOGUE() diff --git a/vendor/gmp-6.3.0/mpn/powerpc64/p7/popcount.asm b/vendor/gmp-6.3.0/mpn/powerpc64/p7/popcount.asm new file mode 100644 index 0000000..129ffef --- /dev/null +++ b/vendor/gmp-6.3.0/mpn/powerpc64/p7/popcount.asm @@ -0,0 +1,90 @@ +dnl PowerPC-64 mpn_popcount. + +dnl Copyright 2012, 2017 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 + +define(`up', r3) +define(`n', r4) + +ASM_START() +PROLOGUE(mpn_popcount) + addi r0, n, 1 +ifdef(`HAVE_ABI_mode32', +` rldicl r0, r0, 63,33', C ...branch count +` srdi r0, r0, 1') C ...for ctr + mtctr r0 + + andi. r0, n, 1 + + li r0, 0 + li r12, 0 + beq L(evn) + +L(odd): ld r4, 0(up) + addi up, up, 8 + popcntd(r0, r4) + bdz L(e1) + +L(evn): ld r4, 0(up) + ld r5, 8(up) + popcntd(r8, r4) + popcntd(r9, r5) + bdz L(e2) + + ld r4, 16(up) + ld r5, 24(up) + bdz L(e4) + addi up, up, 32 + +L(top): add r0, r0, r8 + popcntd(r8, r4) + ld r4, 0(up) + add r12, r12, r9 + popcntd(r9, r5) + ld r5, 8(up) + addi up, up, 16 + bdnz L(top) + +L(e4): add r0, r0, r8 + popcntd(r8, r4) + add r12, r12, r9 + popcntd(r9, r5) +L(e2): add r0, r0, r8 + add r12, r12, r9 +L(e1): add r3, r0, r12 + blr +EPILOGUE() -- cgit v1.2.3