diff options
Diffstat (limited to 'mpi/mips3')
| -rw-r--r-- | mpi/mips3/README | 23 | ||||
| -rw-r--r-- | mpi/mips3/distfiles | 10 | ||||
| -rw-r--r-- | mpi/mips3/mpi-asm-defs.h | 10 | ||||
| -rw-r--r-- | mpi/mips3/mpih-add1.S | 124 | ||||
| -rw-r--r-- | mpi/mips3/mpih-lshift.S | 97 | ||||
| -rw-r--r-- | mpi/mips3/mpih-mul1.S | 89 | ||||
| -rw-r--r-- | mpi/mips3/mpih-mul2.S | 101 | ||||
| -rw-r--r-- | mpi/mips3/mpih-mul3.S | 101 | ||||
| -rw-r--r-- | mpi/mips3/mpih-rshift.S | 95 | ||||
| -rw-r--r-- | mpi/mips3/mpih-sub1.S | 125 |
10 files changed, 775 insertions, 0 deletions
diff --git a/mpi/mips3/README b/mpi/mips3/README new file mode 100644 index 000000000..e94b2c746 --- /dev/null +++ b/mpi/mips3/README @@ -0,0 +1,23 @@ +This directory contains mpn functions optimized for MIPS3. Example of +processors that implement MIPS3 are R4000, R4400, R4600, R4700, and R8000. + +RELEVANT OPTIMIZATION ISSUES + +1. On the R4000 and R4400, branches, both the plain and the "likely" ones, + take 3 cycles to execute. (The fastest possible loop will take 4 cycles, + because of the delay insn.) + + On the R4600, branches takes a single cycle + + On the R8000, branches often take no noticable cycles, as they are + executed in a separate function unit.. + +2. The R4000 and R4400 have a load latency of 4 cycles. + +3. On the R4000 and R4400, multiplies take a data-dependent number of + cycles, contrary to the SGI documentation. There seem to be 3 or 4 + possible latencies. + +STATUS + +Good... diff --git a/mpi/mips3/distfiles b/mpi/mips3/distfiles new file mode 100644 index 000000000..85260fc8e --- /dev/null +++ b/mpi/mips3/distfiles @@ -0,0 +1,10 @@ +README +mpih-add1.S +mpih-sub1.S +mpih-mul1.S +mpih-mul2.S +mpih-mul3.S +mpih-lshift.S +mpih-rshift.S +mpi-asm-defs.h + diff --git a/mpi/mips3/mpi-asm-defs.h b/mpi/mips3/mpi-asm-defs.h new file mode 100644 index 000000000..2d9a9c1f2 --- /dev/null +++ b/mpi/mips3/mpi-asm-defs.h @@ -0,0 +1,10 @@ +/* This file defines some basic constants for the MPI machinery. We + * need to define the types on a per-CPU basis, so it is done with + * this file here. */ +#define BYTES_PER_MPI_LIMB 8 + + + + + + diff --git a/mpi/mips3/mpih-add1.S b/mpi/mips3/mpih-add1.S new file mode 100644 index 000000000..a998420d7 --- /dev/null +++ b/mpi/mips3/mpih-add1.S @@ -0,0 +1,124 @@ +/* mips3 add_n -- Add two limb vectors of the same length > 0 and store + * sum in a third limb vector. + * + * Copyright (C) 1995, 1998, 2000 + * 2001 Free Software Foundation, Inc. + * + * This file is part of GnuPG. + * + * GnuPG is free software; you can redistribute it and/or modify + * it under the terms of 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. + * + * GnuPG 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 a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA + */ + + +/******************* + * mpi_limb_t + * mpihelp_add_n( mpi_ptr_t res_ptr, ($4) + * mpi_ptr_t s1_ptr, ($5) + * mpi_ptr_t s2_ptr, ($6) + * mpi_size_t size) ($7) + */ + + .text + .align 2 + .globl mpihelp_add_n + .ent mpihelp_add_n +mpihelp_add_n: + .set noreorder + .set nomacro + + ld $10,0($5) + ld $11,0($6) + + daddiu $7,$7,-1 + and $9,$7,4-1 # number of limbs in first loop + beq $9,$0,.L0 # if multiple of 4 limbs, skip first loop + move $2,$0 + + dsubu $7,$7,$9 + +.Loop0: daddiu $9,$9,-1 + ld $12,8($5) + daddu $11,$11,$2 + ld $13,8($6) + sltu $8,$11,$2 + daddu $11,$10,$11 + sltu $2,$11,$10 + sd $11,0($4) + or $2,$2,$8 + + daddiu $5,$5,8 + daddiu $6,$6,8 + move $10,$12 + move $11,$13 + bne $9,$0,.Loop0 + daddiu $4,$4,8 + +.L0: beq $7,$0,.Lend + nop + +.Loop: daddiu $7,$7,-4 + + ld $12,8($5) + daddu $11,$11,$2 + ld $13,8($6) + sltu $8,$11,$2 + daddu $11,$10,$11 + sltu $2,$11,$10 + sd $11,0($4) + or $2,$2,$8 + + ld $10,16($5) + daddu $13,$13,$2 + ld $11,16($6) + sltu $8,$13,$2 + daddu $13,$12,$13 + sltu $2,$13,$12 + sd $13,8($4) + or $2,$2,$8 + + ld $12,24($5) + daddu $11,$11,$2 + ld $13,24($6) + sltu $8,$11,$2 + daddu $11,$10,$11 + sltu $2,$11,$10 + sd $11,16($4) + or $2,$2,$8 + + ld $10,32($5) + daddu $13,$13,$2 + ld $11,32($6) + sltu $8,$13,$2 + daddu $13,$12,$13 + sltu $2,$13,$12 + sd $13,24($4) + or $2,$2,$8 + + daddiu $5,$5,32 + daddiu $6,$6,32 + + bne $7,$0,.Loop + daddiu $4,$4,32 + +.Lend: daddu $11,$11,$2 + sltu $8,$11,$2 + daddu $11,$10,$11 + sltu $2,$11,$10 + sd $11,0($4) + j $31 + or $2,$2,$8 + + .end mpihelp_add_n + diff --git a/mpi/mips3/mpih-lshift.S b/mpi/mips3/mpih-lshift.S new file mode 100644 index 000000000..112383f90 --- /dev/null +++ b/mpi/mips3/mpih-lshift.S @@ -0,0 +1,97 @@ +/* mips3 lshift + * + * Copyright (C) 1995, 1998, 2000, + * 2001 Free Software Foundation, Inc. + * + * This file is part of GnuPG. + * + * GnuPG is free software; you can redistribute it and/or modify + * it under the terms of 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. + * + * GnuPG 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 a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA + */ + +/******************* + * mpi_limb_t + * mpihelp_lshift( mpi_ptr_t wp, ($4) + * mpi_ptr_t up, ($5) + * mpi_size_t usize, ($6) + * unsigned cnt) ($7) + */ + + .text + .align 2 + .globl mpihelp_lshift + .ent mpihelp_lshift +mpihelp_lshift: + .set noreorder + .set nomacro + + dsll $2,$6,3 + daddu $5,$5,$2 # make r5 point at end of src + ld $10,-8($5) # load first limb + dsubu $13,$0,$7 + daddu $4,$4,$2 # make r4 point at end of res + daddiu $6,$6,-1 + and $9,$6,4-1 # number of limbs in first loop + beq $9,$0,.L0 # if multiple of 4 limbs, skip first loop + dsrl $2,$10,$13 # compute function result + + dsubu $6,$6,$9 + +.Loop0: ld $3,-16($5) + daddiu $4,$4,-8 + daddiu $5,$5,-8 + daddiu $9,$9,-1 + dsll $11,$10,$7 + dsrl $12,$3,$13 + move $10,$3 + or $8,$11,$12 + bne $9,$0,.Loop0 + sd $8,0($4) + +.L0: beq $6,$0,.Lend + nop + +.Loop: ld $3,-16($5) + daddiu $4,$4,-32 + daddiu $6,$6,-4 + dsll $11,$10,$7 + dsrl $12,$3,$13 + + ld $10,-24($5) + dsll $14,$3,$7 + or $8,$11,$12 + sd $8,24($4) + dsrl $9,$10,$13 + + ld $3,-32($5) + dsll $11,$10,$7 + or $8,$14,$9 + sd $8,16($4) + dsrl $12,$3,$13 + + ld $10,-40($5) + dsll $14,$3,$7 + or $8,$11,$12 + sd $8,8($4) + dsrl $9,$10,$13 + + daddiu $5,$5,-32 + or $8,$14,$9 + bgtz $6,.Loop + sd $8,0($4) + +.Lend: dsll $8,$10,$7 + j $31 + sd $8,-8($4) + .end mpihelp_lshift diff --git a/mpi/mips3/mpih-mul1.S b/mpi/mips3/mpih-mul1.S new file mode 100644 index 000000000..70409e8db --- /dev/null +++ b/mpi/mips3/mpih-mul1.S @@ -0,0 +1,89 @@ +/* mips3 mpih-mul1.S -- Multiply a limb vector with a limb and store + * the result in a second limb vector. + * + * Copyright (C) 1992, 1994, 1995, 1998, 2000 + * 2001 Free Software Foundation, Inc. + * + * This file is part of GnuPG. + * + * GnuPG is free software; you can redistribute it and/or modify + * it under the terms of 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. + * + * GnuPG 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 a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA + */ + + +/******************* + * mpi_limb_t + * mpihelp_mul_1( mpi_ptr_t res_ptr, (r4) + * mpi_ptr_t s1_ptr, (r5) + * mpi_size_t s1_size, (r6) + * mpi_limb_t s2_limb) (r7) + */ + + .text + .align 4 + .globl mpihelp_mul_1 + .ent mpihelp_mul_1 +mpihelp_mul_1: + .set noreorder + .set nomacro + +/* # warm up phase 0 */ + ld $8,0($5) + +/* # warm up phase 1 */ + daddiu $5,$5,8 + dmultu $8,$7 + + daddiu $6,$6,-1 + beq $6,$0,$LC0 + move $2,$0 # zero cy2 + + daddiu $6,$6,-1 + beq $6,$0,$LC1 + ld $8,0($5) # load new s1 limb as early as possible + +Loop: mflo $10 + mfhi $9 + daddiu $5,$5,8 + daddu $10,$10,$2 # add old carry limb to low product limb + dmultu $8,$7 + ld $8,0($5) # load new s1 limb as early as possible + daddiu $6,$6,-1 # decrement loop counter + sltu $2,$10,$2 # carry from previous addition -> $2 + sd $10,0($4) + daddiu $4,$4,8 + bne $6,$0,Loop + daddu $2,$9,$2 # add high product limb and carry from addition + +/* # cool down phase 1 */ +$LC1: mflo $10 + mfhi $9 + daddu $10,$10,$2 + sltu $2,$10,$2 + dmultu $8,$7 + sd $10,0($4) + daddiu $4,$4,8 + daddu $2,$9,$2 # add high product limb and carry from addition + +/* # cool down phase 0 */ +$LC0: mflo $10 + mfhi $9 + daddu $10,$10,$2 + sltu $2,$10,$2 + sd $10,0($4) + j $31 + daddu $2,$9,$2 # add high product limb and carry from addition + + .end mpihelp_mul_1 + diff --git a/mpi/mips3/mpih-mul2.S b/mpi/mips3/mpih-mul2.S new file mode 100644 index 000000000..1bddd0530 --- /dev/null +++ b/mpi/mips3/mpih-mul2.S @@ -0,0 +1,101 @@ +/* MIPS3 addmul_1 -- Multiply a limb vector with a single limb and + * add the product to a second limb vector. + * + * Copyright (C) 1992, 1994, 1995, 1998, 2000 + * 2001 Free Software Foundation, Inc. + * + * This file is part of GnuPG. + * + * GnuPG is free software; you can redistribute it and/or modify + * it under the terms of 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. + * + * GnuPG 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 a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA + */ + + +/******************* + * mpi_limb_t + * mpihelp_addmul_1( mpi_ptr_t res_ptr, (r4) + * mpi_ptr_t s1_ptr, (r5) + * mpi_size_t s1_size, (r6) + * mpi_limb_t s2_limb) (r7) + */ + + .text + .align 4 + .globl mpihelp_addmul_1 + .ent mpihelp_addmul_1 +mpihelp_addmul_1: + .set noreorder + .set nomacro + +/* # warm up phase 0 */ + ld $8,0($5) + +/* # warm up phase 1 */ + daddiu $5,$5,8 + dmultu $8,$7 + + daddiu $6,$6,-1 + beq $6,$0,$LC0 + move $2,$0 # zero cy2 + + daddiu $6,$6,-1 + beq $6,$0,$LC1 + ld $8,0($5) # load new s1 limb as early as possible + +Loop: ld $10,0($4) + mflo $3 + mfhi $9 + daddiu $5,$5,8 + daddu $3,$3,$2 # add old carry limb to low product limb + dmultu $8,$7 + ld $8,0($5) # load new s1 limb as early as possible + daddiu $6,$6,-1 # decrement loop counter + sltu $2,$3,$2 # carry from previous addition -> $2 + daddu $3,$10,$3 + sltu $10,$3,$10 + daddu $2,$2,$10 + sd $3,0($4) + daddiu $4,$4,8 + bne $6,$0,Loop + daddu $2,$9,$2 # add high product limb and carry from addition + +/* # cool down phase 1 */ +$LC1: ld $10,0($4) + mflo $3 + mfhi $9 + daddu $3,$3,$2 + sltu $2,$3,$2 + dmultu $8,$7 + daddu $3,$10,$3 + sltu $10,$3,$10 + daddu $2,$2,$10 + sd $3,0($4) + daddiu $4,$4,8 + daddu $2,$9,$2 # add high product limb and carry from addition + +/* # cool down phase 0 */ +$LC0: ld $10,0($4) + mflo $3 + mfhi $9 + daddu $3,$3,$2 + sltu $2,$3,$2 + daddu $3,$10,$3 + sltu $10,$3,$10 + daddu $2,$2,$10 + sd $3,0($4) + j $31 + daddu $2,$9,$2 # add high product limb and carry from addition + + .end mpihelp_addmul_1 + diff --git a/mpi/mips3/mpih-mul3.S b/mpi/mips3/mpih-mul3.S new file mode 100644 index 000000000..82fe47d53 --- /dev/null +++ b/mpi/mips3/mpih-mul3.S @@ -0,0 +1,101 @@ +/* MIPS3 submul_1 -- Multiply a limb vector with a single limb and + * subtract the product from a second limb vector. + * + * Copyright (C) 1992, 1994, 1995, 1998, 2000 + * 2001 Free Software Foundation, Inc. + * + * This file is part of GnuPG. + * + * GnuPG is free software; you can redistribute it and/or modify + * it under the terms of 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. + * + * GnuPG 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 a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA + */ + + +/******************* + * mpi_limb_t + * mpihelp_submul_1( mpi_ptr_t res_ptr, (r4) + * mpi_ptr_t s1_ptr, (r5) + * mpi_size_t s1_size, (r6) + * mpi_limb_t s2_limb) (r7) + */ + + .text + .align 4 + .globl mpihelp_submul_1 + .ent mpihelp_submul_1 +mpihelp_submul_1: + .set noreorder + .set nomacro + +/* # warm up phase 0 */ + ld $8,0($5) + +/* # warm up phase 1 */ + daddiu $5,$5,8 + dmultu $8,$7 + + daddiu $6,$6,-1 + beq $6,$0,$LC0 + move $2,$0 # zero cy2 + + daddiu $6,$6,-1 + beq $6,$0,$LC1 + ld $8,0($5) # load new s1 limb as early as possible + +Loop: ld $10,0($4) + mflo $3 + mfhi $9 + daddiu $5,$5,8 + daddu $3,$3,$2 # add old carry limb to low product limb + dmultu $8,$7 + ld $8,0($5) # load new s1 limb as early as possible + daddiu $6,$6,-1 # decrement loop counter + sltu $2,$3,$2 # carry from previous addition -> $2 + dsubu $3,$10,$3 + sgtu $10,$3,$10 + daddu $2,$2,$10 + sd $3,0($4) + daddiu $4,$4,8 + bne $6,$0,Loop + daddu $2,$9,$2 # add high product limb and carry from addition + +/* # cool down phase 1 */ +$LC1: ld $10,0($4) + mflo $3 + mfhi $9 + daddu $3,$3,$2 + sltu $2,$3,$2 + dmultu $8,$7 + dsubu $3,$10,$3 + sgtu $10,$3,$10 + daddu $2,$2,$10 + sd $3,0($4) + daddiu $4,$4,8 + daddu $2,$9,$2 # add high product limb and carry from addition + +/* # cool down phase 0 */ +$LC0: ld $10,0($4) + mflo $3 + mfhi $9 + daddu $3,$3,$2 + sltu $2,$3,$2 + dsubu $3,$10,$3 + sgtu $10,$3,$10 + daddu $2,$2,$10 + sd $3,0($4) + j $31 + daddu $2,$9,$2 # add high product limb and carry from addition + + .end mpihelp_submul_1 + diff --git a/mpi/mips3/mpih-rshift.S b/mpi/mips3/mpih-rshift.S new file mode 100644 index 000000000..74a6cc96d --- /dev/null +++ b/mpi/mips3/mpih-rshift.S @@ -0,0 +1,95 @@ +/* mips3 rshift + * + * Copyright (C) 1995, 1998, 2000 + * 2001 Free Software Foundation, Inc. + * + * This file is part of GnuPG. + * + * GnuPG is free software; you can redistribute it and/or modify + * it under the terms of 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. + * + * GnuPG 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 a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA + */ + +/******************* + * mpi_limb_t + * mpihelp_rshift( mpi_ptr_t wp, ($4) + * mpi_ptr_t up, ($5) + * mpi_size_t usize, ($6) + * unsigned cnt) ($7) + */ + + .text + .align 2 + .globl mpihelp_rshift + .ent mpihelp_rshift +mpihelp_rshift: + .set noreorder + .set nomacro + + ld $10,0($5) # load first limb + dsubu $13,$0,$7 + daddiu $6,$6,-1 + and $9,$6,4-1 # number of limbs in first loop + beq $9,$0,.L0 # if multiple of 4 limbs, skip first loop + dsll $2,$10,$13 # compute function result + + dsubu $6,$6,$9 + +.Loop0: ld $3,8($5) + daddiu $4,$4,8 + daddiu $5,$5,8 + daddiu $9,$9,-1 + dsrl $11,$10,$7 + dsll $12,$3,$13 + move $10,$3 + or $8,$11,$12 + bne $9,$0,.Loop0 + sd $8,-8($4) + +.L0: beq $6,$0,.Lend + nop + +.Loop: ld $3,8($5) + daddiu $4,$4,32 + daddiu $6,$6,-4 + dsrl $11,$10,$7 + dsll $12,$3,$13 + + ld $10,16($5) + dsrl $14,$3,$7 + or $8,$11,$12 + sd $8,-32($4) + dsll $9,$10,$13 + + ld $3,24($5) + dsrl $11,$10,$7 + or $8,$14,$9 + sd $8,-24($4) + dsll $12,$3,$13 + + ld $10,32($5) + dsrl $14,$3,$7 + or $8,$11,$12 + sd $8,-16($4) + dsll $9,$10,$13 + + daddiu $5,$5,32 + or $8,$14,$9 + bgtz $6,.Loop + sd $8,-8($4) + +.Lend: dsrl $8,$10,$7 + j $31 + sd $8,0($4) + .end mpihelp_rshift + diff --git a/mpi/mips3/mpih-sub1.S b/mpi/mips3/mpih-sub1.S new file mode 100644 index 000000000..72f506ea2 --- /dev/null +++ b/mpi/mips3/mpih-sub1.S @@ -0,0 +1,125 @@ +/* mips3 sub_n -- Subtract two limb vectors of the same length > 0 and + * store difference in a third limb vector. + * + * Copyright (C) 1995, 1998, 1999, 2000, + * 2001 Free Software Foundation, Inc. + * + * This file is part of GnuPG. + * + * GnuPG is free software; you can redistribute it and/or modify + * it under the terms of 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. + * + * GnuPG 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 a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA + */ + + +/******************* + * mpi_limb_t + * mpihelp_sub_n( mpi_ptr_t res_ptr, (r4) + * mpi_ptr_t s1_ptr, (r5) + * mpi_ptr_t s2_ptr, (r6) + * mpi_size_t size) (r7) + */ + + + .text + .align 2 + .globl mpihelp_sub_n + .ent mpihelp_sub_n +mpihelp_sub_n: + .set noreorder + .set nomacro + + ld $10,0($5) + ld $11,0($6) + + daddiu $7,$7,-1 + and $9,$7,4-1 # number of limbs in first loop + beq $9,$0,.L0 # if multiple of 4 limbs, skip first loop + move $2,$0 + + dsubu $7,$7,$9 + +.Loop0: daddiu $9,$9,-1 + ld $12,8($5) + daddu $11,$11,$2 + ld $13,8($6) + sltu $8,$11,$2 + dsubu $11,$10,$11 + sltu $2,$10,$11 + sd $11,0($4) + or $2,$2,$8 + + daddiu $5,$5,8 + daddiu $6,$6,8 + move $10,$12 + move $11,$13 + bne $9,$0,.Loop0 + daddiu $4,$4,8 + +.L0: beq $7,$0,.Lend + nop + +.Loop: daddiu $7,$7,-4 + + ld $12,8($5) + daddu $11,$11,$2 + ld $13,8($6) + sltu $8,$11,$2 + dsubu $11,$10,$11 + sltu $2,$10,$11 + sd $11,0($4) + or $2,$2,$8 + + ld $10,16($5) + daddu $13,$13,$2 + ld $11,16($6) + sltu $8,$13,$2 + dsubu $13,$12,$13 + sltu $2,$12,$13 + sd $13,8($4) + or $2,$2,$8 + + ld $12,24($5) + daddu $11,$11,$2 + ld $13,24($6) + sltu $8,$11,$2 + dsubu $11,$10,$11 + sltu $2,$10,$11 + sd $11,16($4) + or $2,$2,$8 + + ld $10,32($5) + daddu $13,$13,$2 + ld $11,32($6) + sltu $8,$13,$2 + dsubu $13,$12,$13 + sltu $2,$12,$13 + sd $13,24($4) + or $2,$2,$8 + + daddiu $5,$5,32 + daddiu $6,$6,32 + + bne $7,$0,.Loop + daddiu $4,$4,32 + +.Lend: daddu $11,$11,$2 + sltu $8,$11,$2 + dsubu $11,$10,$11 + sltu $2,$10,$11 + sd $11,0($4) + j $31 + or $2,$2,$8 + + .end mpihelp_sub_n + |