diff options
Diffstat (limited to 'rts/gmp/mpn/powerpc32')
| -rw-r--r-- | rts/gmp/mpn/powerpc32/add_n.asm | 61 | ||||
| -rw-r--r-- | rts/gmp/mpn/powerpc32/addmul_1.asm | 124 | ||||
| -rw-r--r-- | rts/gmp/mpn/powerpc32/aix.m4 | 39 | ||||
| -rw-r--r-- | rts/gmp/mpn/powerpc32/gmp-mparam.h | 66 | ||||
| -rw-r--r-- | rts/gmp/mpn/powerpc32/lshift.asm | 145 | ||||
| -rw-r--r-- | rts/gmp/mpn/powerpc32/mul_1.asm | 86 | ||||
| -rw-r--r-- | rts/gmp/mpn/powerpc32/regmap.m4 | 34 | ||||
| -rw-r--r-- | rts/gmp/mpn/powerpc32/rshift.asm | 60 | ||||
| -rw-r--r-- | rts/gmp/mpn/powerpc32/sub_n.asm | 61 | ||||
| -rw-r--r-- | rts/gmp/mpn/powerpc32/submul_1.asm | 130 | ||||
| -rw-r--r-- | rts/gmp/mpn/powerpc32/umul.asm | 32 |
11 files changed, 838 insertions, 0 deletions
diff --git a/rts/gmp/mpn/powerpc32/add_n.asm b/rts/gmp/mpn/powerpc32/add_n.asm new file mode 100644 index 0000000000..81ed04b162 --- /dev/null +++ b/rts/gmp/mpn/powerpc32/add_n.asm @@ -0,0 +1,61 @@ +dnl PowerPC-32 mpn_add_n -- Add two limb vectors of the same length > 0 and +dnl store sum in a third limb vector. + +dnl Copyright (C) 1995, 1997, 2000 Free Software Foundation, Inc. + +dnl This file is part of the GNU MP Library. + +dnl The GNU MP Library is free software; you can redistribute it and/or modify +dnl it under the terms of the GNU Lesser General Public License as published by +dnl the Free Software Foundation; either version 2.1 of the License, or (at your +dnl option) any later version. + +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 Lesser General Public +dnl License for more details. + +dnl You should have received a copy of the GNU Lesser General Public License +dnl along with the GNU MP Library; see the file COPYING.LIB. If not, write to +dnl the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, +dnl MA 02111-1307, USA. + + +dnl INPUT PARAMETERS +dnl res_ptr r3 +dnl s1_ptr r4 +dnl s2_ptr r5 +dnl size r6 + +include(`../config.m4') + +ASM_START() +PROLOGUE(mpn_add_n) + mtctr r6 C copy size into CTR + addic r0,r0,0 C clear cy + lwz r8,0(r4) C load least significant s1 limb + lwz r0,0(r5) C load least significant s2 limb + addi r3,r3,-4 C offset res_ptr, it's updated before it's used + bdz .Lend C If done, skip loop +.Loop: lwz r9,4(r4) C load s1 limb + lwz r10,4(r5) C load s2 limb + adde r7,r0,r8 C add limbs with cy, set cy + stw r7,4(r3) C store result limb + bdz .Lexit C decrement CTR and exit if done + lwzu r8,8(r4) C load s1 limb and update s1_ptr + lwzu r0,8(r5) C load s2 limb and update s2_ptr + adde r7,r10,r9 C add limbs with cy, set cy + stwu r7,8(r3) C store result limb and update res_ptr + bdnz .Loop C decrement CTR and loop back + +.Lend: adde r7,r0,r8 + stw r7,4(r3) C store ultimate result limb + li r3,0 C load cy into ... + addze r3,r3 C ... return value register + blr +.Lexit: adde r7,r10,r9 + stw r7,8(r3) + li r3,0 C load cy into ... + addze r3,r3 C ... return value register + blr +EPILOGUE(mpn_add_n) diff --git a/rts/gmp/mpn/powerpc32/addmul_1.asm b/rts/gmp/mpn/powerpc32/addmul_1.asm new file mode 100644 index 0000000000..3ef75b1532 --- /dev/null +++ b/rts/gmp/mpn/powerpc32/addmul_1.asm @@ -0,0 +1,124 @@ +dnl PowerPC-32 mpn_addmul_1 -- Multiply a limb vector with a limb and add +dnl the result to a second limb vector. + +dnl Copyright (C) 1995, 1997, 1998, 2000 Free Software Foundation, Inc. + +dnl This file is part of the GNU MP Library. + +dnl The GNU MP Library is free software; you can redistribute it and/or modify +dnl it under the terms of the GNU Lesser General Public License as published by +dnl the Free Software Foundation; either version 2.1 of the License, or (at your +dnl option) any later version. + +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 Lesser General Public +dnl License for more details. + +dnl You should have received a copy of the GNU Lesser General Public License +dnl along with the GNU MP Library; see the file COPYING.LIB. If not, write to +dnl the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, +dnl MA 02111-1307, USA. + + +dnl INPUT PARAMETERS +dnl res_ptr r3 +dnl s1_ptr r4 +dnl size r5 +dnl s2_limb r6 + +dnl This is optimized for the PPC604. It has not been tested on PPC601, PPC603 +dnl or PPC750 since I don't have access to any such machines. + +include(`../config.m4') + +ASM_START() +PROLOGUE(mpn_addmul_1) + cmpi cr0,r5,9 C more than 9 limbs? + bgt cr0,.Lbig C branch if more than 9 limbs + + mtctr r5 + lwz r0,0(r4) + mullw r7,r0,r6 + mulhwu r10,r0,r6 + lwz r9,0(r3) + addc r8,r7,r9 + addi r3,r3,-4 + bdz .Lend +.Lloop: + lwzu r0,4(r4) + stwu r8,4(r3) + mullw r8,r0,r6 + adde r7,r8,r10 + mulhwu r10,r0,r6 + lwz r9,4(r3) + addze r10,r10 + addc r8,r7,r9 + bdnz .Lloop +.Lend: stw r8,4(r3) + addze r3,r10 + blr + +.Lbig: stmw r30,-32(r1) + addi r5,r5,-1 + srwi r0,r5,2 + mtctr r0 + + lwz r7,0(r4) + mullw r8,r7,r6 + mulhwu r0,r7,r6 + lwz r7,0(r3) + addc r8,r8,r7 + stw r8,0(r3) + +.LloopU: + lwz r7,4(r4) + lwz r12,8(r4) + lwz r30,12(r4) + lwzu r31,16(r4) + mullw r8,r7,r6 + mullw r9,r12,r6 + mullw r10,r30,r6 + mullw r11,r31,r6 + adde r8,r8,r0 C add cy_limb + mulhwu r0,r7,r6 + lwz r7,4(r3) + adde r9,r9,r0 + mulhwu r0,r12,r6 + lwz r12,8(r3) + adde r10,r10,r0 + mulhwu r0,r30,r6 + lwz r30,12(r3) + adde r11,r11,r0 + mulhwu r0,r31,r6 + lwz r31,16(r3) + addze r0,r0 C new cy_limb + addc r8,r8,r7 + stw r8,4(r3) + adde r9,r9,r12 + stw r9,8(r3) + adde r10,r10,r30 + stw r10,12(r3) + adde r11,r11,r31 + stwu r11,16(r3) + bdnz .LloopU + + andi. r31,r5,3 + mtctr r31 + beq cr0,.Lendx + +.LloopE: + lwzu r7,4(r4) + mullw r8,r7,r6 + adde r8,r8,r0 C add cy_limb + mulhwu r0,r7,r6 + lwz r7,4(r3) + addze r0,r0 C new cy_limb + addc r8,r8,r7 + stwu r8,4(r3) + bdnz .LloopE +.Lendx: + addze r3,r0 + lmw r30,-32(r1) + blr +EPILOGUE(mpn_addmul_1) diff --git a/rts/gmp/mpn/powerpc32/aix.m4 b/rts/gmp/mpn/powerpc32/aix.m4 new file mode 100644 index 0000000000..2bd8425817 --- /dev/null +++ b/rts/gmp/mpn/powerpc32/aix.m4 @@ -0,0 +1,39 @@ +divert(-1) +dnl m4 macros for AIX 32-bit assembly. + +dnl Copyright (C) 2000 Free Software Foundation, Inc. +dnl +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 +dnl modify it under the terms of the GNU Lesser General Public License as +dnl published by the Free Software Foundation; either version 2.1 of the +dnl License, or (at your option) any later version. +dnl +dnl The GNU MP Library is distributed in the hope that it will be useful, +dnl but WITHOUT ANY WARRANTY; without even the implied warranty of +dnl MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +dnl Lesser General Public License for more details. +dnl +dnl You should have received a copy of the GNU Lesser General Public +dnl License along with the GNU MP Library; see the file COPYING.LIB. If +dnl not, write to the Free Software Foundation, Inc., 59 Temple Place - +dnl Suite 330, Boston, MA 02111-1307, USA. + +define(`ASM_START', + `.toc') + +define(`PROLOGUE', + ` + .globl $1 + .globl .$1 + .csect $1[DS],2 +$1: + .long .$1, TOC[tc0], 0 + .csect .text[PR] + .align 2 +.$1:') + +define(`EPILOGUE', `') + +divert diff --git a/rts/gmp/mpn/powerpc32/gmp-mparam.h b/rts/gmp/mpn/powerpc32/gmp-mparam.h new file mode 100644 index 0000000000..b283185789 --- /dev/null +++ b/rts/gmp/mpn/powerpc32/gmp-mparam.h @@ -0,0 +1,66 @@ +/* gmp-mparam.h -- Compiler/machine parameter header file. + +Copyright (C) 1991, 1993, 1994, 1999, 2000 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 the GNU Lesser General Public License as published by +the Free Software Foundation; either version 2.1 of the License, or (at your +option) any later version. + +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 Lesser General Public +License for more details. + +You should have received a copy of the GNU Lesser General Public License +along with the GNU MP Library; see the file COPYING.LIB. If not, write to +the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, +MA 02111-1307, USA. */ + +#define BITS_PER_MP_LIMB 32 +#define BYTES_PER_MP_LIMB 4 +#define BITS_PER_LONGINT 32 +#define BITS_PER_INT 32 +#define BITS_PER_SHORTINT 16 +#define BITS_PER_CHAR 8 + +/* These values are for the 604. Presumably, these should be considerably + different for the 603 and 750 that have much slower multiply + instructions. */ + +/* Generated by tuneup.c, 2000-05-26. */ + +#ifndef KARATSUBA_MUL_THRESHOLD +#define KARATSUBA_MUL_THRESHOLD 26 /* tuneup says 20 */ +#endif +#ifndef TOOM3_MUL_THRESHOLD +#define TOOM3_MUL_THRESHOLD 228 +#endif + +#ifndef KARATSUBA_SQR_THRESHOLD +#define KARATSUBA_SQR_THRESHOLD 46 /* tuneup says 44 */ +#endif +#ifndef TOOM3_SQR_THRESHOLD +#define TOOM3_SQR_THRESHOLD 262 +#endif + +#ifndef BZ_THRESHOLD +#define BZ_THRESHOLD 52 +#endif + +#ifndef FIB_THRESHOLD +#define FIB_THRESHOLD 86 +#endif + +#ifndef POWM_THRESHOLD +#define POWM_THRESHOLD 23 +#endif + +#ifndef GCD_ACCEL_THRESHOLD +#define GCD_ACCEL_THRESHOLD 7 +#endif +#ifndef GCDEXT_THRESHOLD +#define GCDEXT_THRESHOLD 53 +#endif diff --git a/rts/gmp/mpn/powerpc32/lshift.asm b/rts/gmp/mpn/powerpc32/lshift.asm new file mode 100644 index 0000000000..73a85430ab --- /dev/null +++ b/rts/gmp/mpn/powerpc32/lshift.asm @@ -0,0 +1,145 @@ +dnl PowerPC-32 mpn_lshift -- Shift a number left. + +dnl Copyright (C) 1995, 1998, 2000 Free Software Foundation, Inc. + +dnl This file is part of the GNU MP Library. + +dnl The GNU MP Library is free software; you can redistribute it and/or modify +dnl it under the terms of the GNU Lesser General Public License as published by +dnl the Free Software Foundation; either version 2.1 of the License, or (at your +dnl option) any later version. + +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 Lesser General Public +dnl License for more details. + +dnl You should have received a copy of the GNU Lesser General Public License +dnl along with the GNU MP Library; see the file COPYING.LIB. If not, write to +dnl the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, +dnl MA 02111-1307, USA. + + +dnl INPUT PARAMETERS +dnl res_ptr r3 +dnl s1_ptr r4 +dnl size r5 +dnl cnt r6 + +include(`../config.m4') + +ASM_START() +PROLOGUE(mpn_lshift) + cmpi cr0,r5,12 C more than 12 limbs? + slwi r0,r5,2 + add r4,r4,r0 C make r4 point at end of s1 + add r7,r3,r0 C make r7 point at end of res + bgt .LBIG C branch if more than 12 limbs + + mtctr r5 C copy size into CTR + subfic r8,r6,32 + lwzu r11,-4(r4) C load first s1 limb + srw r3,r11,r8 C compute function return value + bdz .Lend1 + +.Loop: lwzu r10,-4(r4) + slw r9,r11,r6 + srw r12,r10,r8 + or r9,r9,r12 + stwu r9,-4(r7) + bdz .Lend2 + lwzu r11,-4(r4) + slw r9,r10,r6 + srw r12,r11,r8 + or r9,r9,r12 + stwu r9,-4(r7) + bdnz .Loop + +.Lend1: slw r0,r11,r6 + stw r0,-4(r7) + blr +.Lend2: slw r0,r10,r6 + stw r0,-4(r7) + blr + +.LBIG: + stmw r24,-32(r1) C save registers we are supposed to preserve + lwzu r9,-4(r4) + subfic r8,r6,32 + srw r3,r9,r8 C compute function return value + slw r0,r9,r6 + addi r5,r5,-1 + + andi. r10,r5,3 C count for spill loop + beq .Le + mtctr r10 + lwzu r28,-4(r4) + bdz .Lxe0 + +.Loop0: slw r12,r28,r6 + srw r24,r28,r8 + lwzu r28,-4(r4) + or r24,r0,r24 + stwu r24,-4(r7) + mr r0,r12 + bdnz .Loop0 C taken at most once! + +.Lxe0: slw r12,r28,r6 + srw r24,r28,r8 + or r24,r0,r24 + stwu r24,-4(r7) + mr r0,r12 + +.Le: srwi r5,r5,2 C count for unrolled loop + addi r5,r5,-1 + mtctr r5 + lwz r28,-4(r4) + lwz r29,-8(r4) + lwz r30,-12(r4) + lwzu r31,-16(r4) + +.LoopU: slw r9,r28,r6 + srw r24,r28,r8 + lwz r28,-4(r4) + slw r10,r29,r6 + srw r25,r29,r8 + lwz r29,-8(r4) + slw r11,r30,r6 + srw r26,r30,r8 + lwz r30,-12(r4) + slw r12,r31,r6 + srw r27,r31,r8 + lwzu r31,-16(r4) + or r24,r0,r24 + stw r24,-4(r7) + or r25,r9,r25 + stw r25,-8(r7) + or r26,r10,r26 + stw r26,-12(r7) + or r27,r11,r27 + stwu r27,-16(r7) + mr r0,r12 + bdnz .LoopU + + slw r9,r28,r6 + srw r24,r28,r8 + slw r10,r29,r6 + srw r25,r29,r8 + slw r11,r30,r6 + srw r26,r30,r8 + slw r12,r31,r6 + srw r27,r31,r8 + or r24,r0,r24 + stw r24,-4(r7) + or r25,r9,r25 + stw r25,-8(r7) + or r26,r10,r26 + stw r26,-12(r7) + or r27,r11,r27 + stwu r27,-16(r7) + mr r0,r12 + + stw r0,-4(r7) + lmw r24,-32(r1) C restore registers + blr +EPILOGUE(mpn_lshift) diff --git a/rts/gmp/mpn/powerpc32/mul_1.asm b/rts/gmp/mpn/powerpc32/mul_1.asm new file mode 100644 index 0000000000..ec878b54d5 --- /dev/null +++ b/rts/gmp/mpn/powerpc32/mul_1.asm @@ -0,0 +1,86 @@ +dnl PowerPC-32 mpn_mul_1 -- Multiply a limb vector with a limb and store +dnl the result in a second limb vector. + +dnl Copyright (C) 1995, 1997, 2000 Free Software Foundation, Inc. + +dnl This file is part of the GNU MP Library. + +dnl The GNU MP Library is free software; you can redistribute it and/or modify +dnl it under the terms of the GNU Lesser General Public License as published by +dnl the Free Software Foundation; either version 2.1 of the License, or (at your +dnl option) any later version. + +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 Lesser General Public +dnl License for more details. + +dnl You should have received a copy of the GNU Lesser General Public License +dnl along with the GNU MP Library; see the file COPYING.LIB. If not, write to +dnl the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, +dnl MA 02111-1307, USA. + + +dnl INPUT PARAMETERS +dnl res_ptr r3 +dnl s1_ptr r4 +dnl size r5 +dnl s2_limb r6 + +dnl This is optimized for the PPC604 but it runs decently even on PPC601. It +dnl has not been tested on a PPC603 since I don't have access to any such +dnl machines. + +include(`../config.m4') + +ASM_START() +PROLOGUE(mpn_mul_1) + mtctr r5 + addi r3,r3,-4 C adjust res_ptr, it's offset before it's used + li r12,0 C clear upper product reg + addic r0,r0,0 C clear cy +C Start software pipeline + lwz r8,0(r4) + bdz .Lend3 + stmw r30,-8(r1) C save registers we are supposed to preserve + lwzu r9,4(r4) + mullw r11,r8,r6 + mulhwu r0,r8,r6 + bdz .Lend1 +C Software pipelined main loop +.Loop: lwz r8,4(r4) + mullw r10,r9,r6 + adde r30,r11,r12 + mulhwu r12,r9,r6 + stw r30,4(r3) + bdz .Lend2 + lwzu r9,8(r4) + mullw r11,r8,r6 + adde r31,r10,r0 + mulhwu r0,r8,r6 + stwu r31,8(r3) + bdnz .Loop +C Finish software pipeline +.Lend1: mullw r10,r9,r6 + adde r30,r11,r12 + mulhwu r12,r9,r6 + stw r30,4(r3) + adde r31,r10,r0 + stwu r31,8(r3) + addze r3,r12 + lmw r30,-8(r1) C restore registers from stack + blr +.Lend2: mullw r11,r8,r6 + adde r31,r10,r0 + mulhwu r0,r8,r6 + stwu r31,8(r3) + adde r30,r11,r12 + stw r30,4(r3) + addze r3,r0 + lmw r30,-8(r1) C restore registers from stack + blr +.Lend3: mullw r11,r8,r6 + stw r11,4(r3) + mulhwu r3,r8,r6 + blr +EPILOGUE(mpn_mul_1) diff --git a/rts/gmp/mpn/powerpc32/regmap.m4 b/rts/gmp/mpn/powerpc32/regmap.m4 new file mode 100644 index 0000000000..978f18902a --- /dev/null +++ b/rts/gmp/mpn/powerpc32/regmap.m4 @@ -0,0 +1,34 @@ +divert(-1) + +dnl Copyright (C) 2000 Free Software Foundation, Inc. +dnl +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 +dnl modify it under the terms of the GNU Lesser General Public License as +dnl published by the Free Software Foundation; either version 2.1 of the +dnl License, or (at your option) any later version. +dnl +dnl The GNU MP Library is distributed in the hope that it will be useful, +dnl but WITHOUT ANY WARRANTY; without even the implied warranty of +dnl MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +dnl Lesser General Public License for more details. +dnl +dnl You should have received a copy of the GNU Lesser General Public +dnl License along with the GNU MP Library; see the file COPYING.LIB. If +dnl not, write to the Free Software Foundation, Inc., 59 Temple Place - +dnl Suite 330, Boston, MA 02111-1307, USA. + + +dnl Map register names r0, r1, etc, to just `0', `1', etc. +dnl This is needed on all systems but NeXT, Rhapsody, and MacOS-X +forloop(i,0,31, +`define(`r'i,i)' +) + +dnl Likewise for cr0, cr1, etc. +forloop(i,0,7, +`define(`cr'i,i)' +) + +divert diff --git a/rts/gmp/mpn/powerpc32/rshift.asm b/rts/gmp/mpn/powerpc32/rshift.asm new file mode 100644 index 0000000000..a09ba04938 --- /dev/null +++ b/rts/gmp/mpn/powerpc32/rshift.asm @@ -0,0 +1,60 @@ +dnl PowerPC-32 mpn_rshift -- Shift a number right. + +dnl Copyright (C) 1995, 2000 Free Software Foundation, Inc. + +dnl This file is part of the GNU MP Library. + +dnl The GNU MP Library is free software; you can redistribute it and/or modify +dnl it under the terms of the GNU Lesser General Public License as published by +dnl the Free Software Foundation; either version 2.1 of the License, or (at your +dnl option) any later version. + +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 Lesser General Public +dnl License for more details. + +dnl You should have received a copy of the GNU Lesser General Public License +dnl along with the GNU MP Library; see the file COPYING.LIB. If not, write to +dnl the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, +dnl MA 02111-1307, USA. + + +dnl INPUT PARAMETERS +dnl res_ptr r3 +dnl s1_ptr r4 +dnl size r5 +dnl cnt r6 + +include(`../config.m4') + +ASM_START() +PROLOGUE(mpn_rshift) + mtctr r5 C copy size into CTR + addi r7,r3,-4 C move adjusted res_ptr to free return reg + subfic r8,r6,32 + lwz r11,0(r4) C load first s1 limb + slw r3,r11,r8 C compute function return value + bdz .Lend1 + +.Loop: lwzu r10,4(r4) + srw r9,r11,r6 + slw r12,r10,r8 + or r9,r9,r12 + stwu r9,4(r7) + bdz .Lend2 + lwzu r11,4(r4) + srw r9,r10,r6 + slw r12,r11,r8 + or r9,r9,r12 + stwu r9,4(r7) + bdnz .Loop + +.Lend1: srw r0,r11,r6 + stw r0,4(r7) + blr + +.Lend2: srw r0,r10,r6 + stw r0,4(r7) + blr +EPILOGUE(mpn_rshift) diff --git a/rts/gmp/mpn/powerpc32/sub_n.asm b/rts/gmp/mpn/powerpc32/sub_n.asm new file mode 100644 index 0000000000..b04b4192ef --- /dev/null +++ b/rts/gmp/mpn/powerpc32/sub_n.asm @@ -0,0 +1,61 @@ +dnl PowerPC-32 mpn_sub_n -- Subtract two limb vectors of the same length > 0 +dnl and store difference in a third limb vector. + +dnl Copyright (C) 1995, 1997, 2000 Free Software Foundation, Inc. + +dnl This file is part of the GNU MP Library. + +dnl The GNU MP Library is free software; you can redistribute it and/or modify +dnl it under the terms of the GNU Lesser General Public License as published by +dnl the Free Software Foundation; either version 2.1 of the License, or (at your +dnl option) any later version. + +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 Lesser General Public +dnl License for more details. + +dnl You should have received a copy of the GNU Lesser General Public License +dnl along with the GNU MP Library; see the file COPYING.LIB. If not, write to +dnl the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, +dnl MA 02111-1307, USA. + + +dnl INPUT PARAMETERS +dnl res_ptr r3 +dnl s1_ptr r4 +dnl s2_ptr r5 +dnl size r6 + +include(`../config.m4') + +ASM_START() +PROLOGUE(mpn_sub_n) + mtctr r6 C copy size into CTR + addic r0,r6,-1 C set cy + lwz r8,0(r4) C load least significant s1 limb + lwz r0,0(r5) C load least significant s2 limb + addi r3,r3,-4 C offset res_ptr, it's updated before it's used + bdz .Lend C If done, skip loop +.Loop: lwz r9,4(r4) C load s1 limb + lwz r10,4(r5) C load s2 limb + subfe r7,r0,r8 C subtract limbs with cy, set cy + stw r7,4(r3) C store result limb + bdz .Lexit C decrement CTR and exit if done + lwzu r8,8(r4) C load s1 limb and update s1_ptr + lwzu r0,8(r5) C load s2 limb and update s2_ptr + subfe r7,r10,r9 C subtract limbs with cy, set cy + stwu r7,8(r3) C store result limb and update res_ptr + bdnz .Loop C decrement CTR and loop back + +.Lend: subfe r7,r0,r8 + stw r7,4(r3) C store ultimate result limb + subfe r3,r0,r0 C load !cy into ... + subfic r3,r3,0 C ... return value register + blr +.Lexit: subfe r7,r10,r9 + stw r7,8(r3) + subfe r3,r0,r0 C load !cy into ... + subfic r3,r3,0 C ... return value register + blr +EPILOGUE(mpn_sub_n) diff --git a/rts/gmp/mpn/powerpc32/submul_1.asm b/rts/gmp/mpn/powerpc32/submul_1.asm new file mode 100644 index 0000000000..a129e9f9ea --- /dev/null +++ b/rts/gmp/mpn/powerpc32/submul_1.asm @@ -0,0 +1,130 @@ +dnl PowerPC-32 mpn_submul_1 -- Multiply a limb vector with a limb and subtract +dnl the result from a second limb vector. + +dnl Copyright (C) 1995, 1997, 1998, 2000 Free Software Foundation, Inc. + +dnl This file is part of the GNU MP Library. + +dnl The GNU MP Library is free software; you can redistribute it and/or modify +dnl it under the terms of the GNU Lesser General Public License as published by +dnl the Free Software Foundation; either version 2.1 of the License, or (at your +dnl option) any later version. + +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 Lesser General Public +dnl License for more details. + +dnl You should have received a copy of the GNU Lesser General Public License +dnl along with the GNU MP Library; see the file COPYING.LIB. If not, write to +dnl the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, +dnl MA 02111-1307, USA. + + +dnl INPUT PARAMETERS +dnl res_ptr r3 +dnl s1_ptr r4 +dnl size r5 +dnl s2_limb r6 + +dnl This is optimized for the PPC604. It has not been tested on PPC601, PPC603 +dnl or PPC750 since I don't have access to any such machines. + +include(`../config.m4') + +ASM_START() +PROLOGUE(mpn_submul_1) + cmpi cr0,r5,9 C more than 9 limbs? + bgt cr0,.Lbig C branch if more than 9 limbs + + mtctr r5 + lwz r0,0(r4) + mullw r7,r0,r6 + mulhwu r10,r0,r6 + lwz r9,0(r3) + subfc r8,r7,r9 + addc r7,r7,r8 C invert cy (r7 is junk) + addi r3,r3,-4 + bdz .Lend +.Lloop: + lwzu r0,4(r4) + stwu r8,4(r3) + mullw r8,r0,r6 + adde r7,r8,r10 + mulhwu r10,r0,r6 + lwz r9,4(r3) + addze r10,r10 + subfc r8,r7,r9 + addc r7,r7,r8 C invert cy (r7 is junk) + bdnz .Lloop +.Lend: stw r8,4(r3) + addze r3,r10 + blr + +.Lbig: stmw r30,-32(r1) + addi r5,r5,-1 + srwi r0,r5,2 + mtctr r0 + + lwz r7,0(r4) + mullw r8,r7,r6 + mulhwu r0,r7,r6 + lwz r7,0(r3) + subfc r7,r8,r7 + addc r8,r8,r7 + stw r7,0(r3) + +.LloopU: + lwz r7,4(r4) + lwz r12,8(r4) + lwz r30,12(r4) + lwzu r31,16(r4) + mullw r8,r7,r6 + mullw r9,r12,r6 + mullw r10,r30,r6 + mullw r11,r31,r6 + adde r8,r8,r0 C add cy_limb + mulhwu r0,r7,r6 + lwz r7,4(r3) + adde r9,r9,r0 + mulhwu r0,r12,r6 + lwz r12,8(r3) + adde r10,r10,r0 + mulhwu r0,r30,r6 + lwz r30,12(r3) + adde r11,r11,r0 + mulhwu r0,r31,r6 + lwz r31,16(r3) + addze r0,r0 C new cy_limb + subfc r7,r8,r7 + stw r7,4(r3) + subfe r12,r9,r12 + stw r12,8(r3) + subfe r30,r10,r30 + stw r30,12(r3) + subfe r31,r11,r31 + stwu r31,16(r3) + subfe r11,r11,r11 C invert ... + addic r11,r11,1 C ... carry + bdnz .LloopU + + andi. r31,r5,3 + mtctr r31 + beq cr0,.Lendx + +.LloopE: + lwzu r7,4(r4) + mullw r8,r7,r6 + adde r8,r8,r0 C add cy_limb + mulhwu r0,r7,r6 + lwz r7,4(r3) + addze r0,r0 C new cy_limb + subfc r7,r8,r7 + addc r8,r8,r7 + stwu r7,4(r3) + bdnz .LloopE +.Lendx: + addze r3,r0 + lmw r30,-32(r1) + blr +EPILOGUE(mpn_submul_1) diff --git a/rts/gmp/mpn/powerpc32/umul.asm b/rts/gmp/mpn/powerpc32/umul.asm new file mode 100644 index 0000000000..eeaa0a4dc8 --- /dev/null +++ b/rts/gmp/mpn/powerpc32/umul.asm @@ -0,0 +1,32 @@ +dnl PowerPC-32 umul_ppmm -- support for longlong.h + +dnl Copyright (C) 2000 Free Software Foundation, Inc. + +dnl This file is part of the GNU MP Library. + +dnl The GNU MP Library is free software; you can redistribute it and/or modify +dnl it under the terms of the GNU Lesser General Public License as published by +dnl the Free Software Foundation; either version 2.1 of the License, or (at your +dnl option) any later version. + +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 Lesser General Public +dnl License for more details. + +dnl You should have received a copy of the GNU Lesser General Public License +dnl along with the GNU MP Library; see the file COPYING.LIB. If not, write to +dnl the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, +dnl MA 02111-1307, USA. + + +include(`../config.m4') + +ASM_START() +PROLOGUE(mpn_umul_ppmm) + mullw 0,4,5 + mulhwu 9,4,5 + stw 0,0(3) + mr 3,9 + blr +EPILOGUE(mpn_umul_ppmm) |