diff options
author | Adhemerval Zanella <azanella@linux.vnet.ibm.com> | 2014-11-19 16:27:56 -0500 |
---|---|---|
committer | Adhemerval Zanella <azanella@linux.vnet.ibm.com> | 2014-12-02 13:34:02 -0500 |
commit | 0f0a1c82f5d8e7b8d24f9eddc2b7728abd3d5bc4 (patch) | |
tree | 4c0490845d3af2ee55f351c32741fac1af18d8ef /sysdeps/powerpc/powerpc64/power7 | |
parent | bb2542e0ae6f82197e804e7e9b78fed461553fc0 (diff) | |
download | glibc-0f0a1c82f5d8e7b8d24f9eddc2b7728abd3d5bc4.tar.gz glibc-0f0a1c82f5d8e7b8d24f9eddc2b7728abd3d5bc4.tar.xz glibc-0f0a1c82f5d8e7b8d24f9eddc2b7728abd3d5bc4.zip |
powerpc: Add powerpc64 strpbrk optimization
This patch makes the POWER7 optimized strpbrk generic by using default doubleword stores to zero the hash, instead of VSX instructions. Performance on POWER7/POWER8 does not change.
Diffstat (limited to 'sysdeps/powerpc/powerpc64/power7')
-rw-r--r-- | sysdeps/powerpc/powerpc64/power7/strpbrk.S | 148 |
1 files changed, 0 insertions, 148 deletions
diff --git a/sysdeps/powerpc/powerpc64/power7/strpbrk.S b/sysdeps/powerpc/powerpc64/power7/strpbrk.S deleted file mode 100644 index d6204a7754..0000000000 --- a/sysdeps/powerpc/powerpc64/power7/strpbrk.S +++ /dev/null @@ -1,148 +0,0 @@ -/* Optimized strpbrk implementation for PowerPC64/POWER7. - Copyright (C) 2014 Free Software Foundation, Inc. - This file is part of the GNU C Library. - - The GNU C 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 C 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 C Library; if not, see - <http://www.gnu.org/licenses/>. */ - -#include <sysdep.h> - -/* char [r3] *strpbrk(const char [r4] *s, const char [r5] *accept) */ - - .machine power7 -EALIGN (strpbrk, 4, 0) - CALL_MCOUNT 3 - - lbz r10,0(r4) - cmpdi cr7,r10,0 /* accept[0] == '\0' ? */ - beq cr7,L(nullfound) - - /* The idea to speed up the algorithm is to create a lookup table - for fast check if input character should be considered. For ASCII - or ISO-8859-X character sets it has 256 positions. */ - - /* First the table should be cleared and to avoid unaligned accesses - when using the VSX stores the table address is aligned to 16 - bytes. */ - xxlxor v0,v0,v0 - - /* PPC64 ELF ABI stack is aligned to 16 bytes */ - addi r9,r1,-256 - - li r5,16 - li r6,32 - li r8,48 - addi r12,r9,64 - /* Clear the table with 0 values */ - stxvw4x v0,r0,r9 - addi r11,r9,128 - addi r7,r9,192 - stxvw4x v0,r9,r5 - li r0,1 - stxvw4x v0,r9,r6 - stxvw4x v0,r9,r8 - stxvw4x v0,r0,r12 - stxvw4x v0,r12,r5 - stxvw4x v0,r12,r6 - stxvw4x v0,r12,r8 - stxvw4x v0,r0,r11 - stxvw4x v0,r11,r5 - stxvw4x v0,r11,r6 - stxvw4x v0,r11,r8 - stxvw4x v0,r0,r7 - stxvw4x v0,r7,r5 - stxvw4x v0,r7,r6 - stxvw4x v0,r7,r8 - - /* Initialize the table as: - for (i=0; accept[i]; i++ - table[accept[i]]] = 1 */ - .p2align 4,,15 -L(init_table): - stbx r0,r9,r10 - lbzu r10,1(r4) - cmpdi r0,r10,0 - bne cr0,L(init_table) -L(finish_table): - /* set table[0] = 1 */ - li r4,1 - stb r4,0(r9) - b L(mainloop) - - /* Unrool the loop 4 times and check using the table as: - i = 0; - while (1) - { - if (table[input[i++]] == 1) - return (s[i -1] ? s + i - 1: NULL); - if (table[input[i++]] == 1) - return (s[i -1] ? s + i - 1: NULL); - if (table[input[i++]] == 1) - return (s[i -1] ? s + i - 1: NULL); - if (table[input[i++]] == 1) - return (s[i -1] ? s + i - 1: NULL); - } */ - .p2align 4 -L(unroll): - lbz r0,1(r3) - lbzx r8,r9,r0 - cmpwi cr6,r8,1 - beq cr6,L(checkend2) - lbz r10,2(r3) - lbzx r4,r9,r10 - cmpwi cr7,r4,1 - beq cr7,L(checkend3) - lbz r12,3(r3) - addi r3,r3,4 - lbzx r11,r9,r12 - cmpwi cr0,r11,1 - beq cr0,L(checkend) -L(mainloop): - lbz r12,0(r3) - addi r11,r3,1 - addi r5,r3,2 - addi r7,r3,3 - lbzx r6,r9,r12 - cmpwi cr1,r6,1 - bne cr1,L(unroll) - cmpdi cr0,r12,0 - beq cr0,L(nullfound) -L(end): - blr - - .p2align 4 -L(checkend): - cmpdi cr1,r12,0 - mr r3,r7 - bne cr1,L(end) -L(nullfound): - /* return NULL */ - li 3,0 - blr - - .p2align 4 -L(checkend2): - cmpdi cr7,r0,0 - mr r3,r11 - beq cr7,L(nullfound) - blr - - .p2align 4 -L(checkend3): - cmpdi cr6,r10,0 - mr r3,r5 - beq cr6,L(nullfound) - blr -END (strpbrk) -libc_hidden_builtin_def (strpbrk) |