AuroraMiddleware/glibc
1
0
Fork 0
mirror of https://sourceware.org/git/glibc.git synced 2024-11-22 13:00:06 +00:00
Code Issues Packages Projects Releases Wiki Activity
3d7090f14b
glibc/sysdeps/aarch64/memcmp.S

211 lines
4.3 KiB
ArmAsm
Raw Normal View History

AArch64: Implement optimized memcmp.
2013-01-16 12:58:40 +00:00
/* memcmp - compare memory
Update copyright dates with scripts/update-copyrights
2023-01-06 21:08:04 +00:00
Copyright (C) 2013-2023 Free Software Foundation, Inc.
AArch64: Implement optimized memcmp.
2013-01-16 12:58:40 +00:00
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
Prefer https to http for gnu.org and fsf.org URLs Also, change sources.redhat.com to sourceware.org. This patch was automatically generated by running the following shell script, which uses GNU sed, and which avoids modifying files imported from upstream: sed -ri ' s,(http|ftp)(://(.*\.)?(gnu|fsf|sourceware)\.org($|[^.]|\.[^a-z])),https\2,g s,(http|ftp)(://(.*\.)?)sources\.redhat\.com($|[^.]|\.[^a-z]),https\2sourceware.org\4,g ' \ $(find $(git ls-files) -prune -type f \ ! -name '*.po' \ ! -name 'ChangeLog*' \ ! -path COPYING ! -path COPYING.LIB \ ! -path manual/fdl-1.3.texi ! -path manual/lgpl-2.1.texi \ ! -path manual/texinfo.tex ! -path scripts/config.guess \ ! -path scripts/config.sub ! -path scripts/install-sh \ ! -path scripts/mkinstalldirs ! -path scripts/move-if-change \ ! -path INSTALL ! -path locale/programs/charmap-kw.h \ ! -path po/libc.pot ! -path sysdeps/gnu/errlist.c \ ! '(' -name configure \ -execdir test -f configure.ac -o -f configure.in ';' ')' \ ! '(' -name preconfigure \ -execdir test -f preconfigure.ac ';' ')' \ -print) and then by running 'make dist-prepare' to regenerate files built from the altered files, and then executing the following to cleanup: chmod a+x sysdeps/unix/sysv/linux/riscv/configure # Omit irrelevant whitespace and comment-only changes, # perhaps from a slightly-different Autoconf version. git checkout -f \ sysdeps/csky/configure \ sysdeps/hppa/configure \ sysdeps/riscv/configure \ sysdeps/unix/sysv/linux/csky/configure # Omit changes that caused a pre-commit check to fail like this: # remote: *** error: sysdeps/powerpc/powerpc64/ppc-mcount.S: trailing lines git checkout -f \ sysdeps/powerpc/powerpc64/ppc-mcount.S \ sysdeps/unix/sysv/linux/s390/s390-64/syscall.S # Omit change that caused a pre-commit check to fail like this: # remote: *** error: sysdeps/sparc/sparc64/multiarch/memcpy-ultra3.S: last line does not end in newline git checkout -f sysdeps/sparc/sparc64/multiarch/memcpy-ultra3.S
2019-09-07 05:40:42 +00:00
<https://www.gnu.org/licenses/>. */
AArch64: Implement optimized memcmp.
2013-01-16 12:58:40 +00:00
#include <sysdep.h>
/* Assumptions:
*
AArch64: Optimize memcmp Rewrite memcmp to improve performance. On small and medium inputs performance is 10-20% better. Large inputs use a SIMD loop processing 64 bytes per iteration, which is 30-50% faster depending on the size. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-12-02 18:30:55 +00:00
* ARMv8-a, AArch64, Advanced SIMD, unaligned accesses.
AArch64: Implement optimized memcmp.
2013-01-16 12:58:40 +00:00
*/
AArch64: Optimize memcmp Rewrite memcmp to improve performance. On small and medium inputs performance is 10-20% better. Large inputs use a SIMD loop processing 64 bytes per iteration, which is 30-50% faster depending on the size. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-12-02 18:30:55 +00:00
#define src1 x0
#define src2 x1
#define limit x2
#define result w0
#define data1 x3
#define data1w w3
#define data2 x4
#define data2w w4
#define data3 x5
#define data3w w5
#define data4 x6
#define data4w w6
#define tmp x6
#define src1end x7
#define src2end x8
ENTRY (memcmp)
aarch64: use PTR_ARG and SIZE_ARG instead of DELOUSE DELOUSE was added to asm code to make them compatible with non-LP64 ABIs, but it is an unfortunate name and the code was not compatible with ABIs where pointer and size_t are different. Glibc currently only supports the LP64 ABI so these macros are not really needed or tested, but for now the name is changed to be more meaningful instead of removing them completely. Some DELOUSE macros were dropped: clone, strlen and strnlen used it unnecessarily. The out of tree ILP32 patches are currently not maintained and will likely need a rework to rebase them on top of the time64 changes.
2020-12-17 10:03:05 +00:00
PTR_ARG (0)
PTR_ARG (1)
SIZE_ARG (2)
AArch64: Implement optimized memcmp.
2013-01-16 12:58:40 +00:00
AArch64: Optimize memcmp Rewrite memcmp to improve performance. On small and medium inputs performance is 10-20% better. Large inputs use a SIMD loop processing 64 bytes per iteration, which is 30-50% faster depending on the size. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-12-02 18:30:55 +00:00
cmp limit, 16
aarch64: Optimized implementation of memcmp The loop body is expanded from a 16-byte comparison to a 64-byte comparison, and the usage of ldp is replaced by the Post-index mode to the Base plus offset mode. Hence, compare can faster 18% around > 128 bytes in all. Checked on aarch64-linux-gnu. Reviewed-by: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
2019-12-19 12:31:59 +00:00
b.lo L(less16)
AArch64: Optimize memcmp Rewrite memcmp to improve performance. On small and medium inputs performance is 10-20% better. Large inputs use a SIMD loop processing 64 bytes per iteration, which is 30-50% faster depending on the size. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-12-02 18:30:55 +00:00
ldp data1, data3, [src1]
ldp data2, data4, [src2]
aarch64: Optimized implementation of memcmp The loop body is expanded from a 16-byte comparison to a 64-byte comparison, and the usage of ldp is replaced by the Post-index mode to the Base plus offset mode. Hence, compare can faster 18% around > 128 bytes in all. Checked on aarch64-linux-gnu. Reviewed-by: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
2019-12-19 12:31:59 +00:00
ccmp data1, data2, 0, ne
AArch64: Optimize memcmp Rewrite memcmp to improve performance. On small and medium inputs performance is 10-20% better. Large inputs use a SIMD loop processing 64 bytes per iteration, which is 30-50% faster depending on the size. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-12-02 18:30:55 +00:00
ccmp data3, data4, 0, eq
b.ne L(return2)
[AArch64] Optimized memcmp. This is an optimized memcmp for AArch64. This is a complete rewrite using a different algorithm. The previous version split into cases where both inputs were aligned, the inputs were mutually aligned and unaligned using a byte loop. The new version combines all these cases, while small inputs of less than 8 bytes are handled separately. This allows the main code to be sped up using unaligned loads since there are now at least 8 bytes to be compared. After the first 8 bytes, align the first input. This ensures each iteration does at most one unaligned access and mutually aligned inputs behave as aligned. After the main loop, process the last 8 bytes using unaligned accesses. This improves performance of (mutually) aligned cases by 25% and unaligned by >500% (yes >6 times faster) on large inputs. * sysdeps/aarch64/memcmp.S (memcmp): Rewrite of optimized memcmp.
2017-08-10 16:00:38 +00:00
AArch64: Optimize memcmp Rewrite memcmp to improve performance. On small and medium inputs performance is 10-20% better. Large inputs use a SIMD loop processing 64 bytes per iteration, which is 30-50% faster depending on the size. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-12-02 18:30:55 +00:00
add src1end, src1, limit
add src2end, src2, limit
cmp limit, 32
aarch64: Optimized memcmp for medium to large sizes This improved memcmp provides a fast path for compares up to 16 bytes and then compares 16 bytes at a time, thus optimizing loads from both sources. The glibc memcmp microbenchmark retains performance (with an error of ~1ns) for smaller compare sizes and reduces up to 31% of execution time for compares up to 4K on the APM Mustang. On Qualcomm Falkor this improves to almost 48%, i.e. it is almost 2x improvement for sizes of 2K and above. * sysdeps/aarch64/memcmp.S: Widen comparison to 16 bytes at a time.
2018-03-06 13:52:39 +00:00
b.ls L(last_bytes)
AArch64: Optimize memcmp Rewrite memcmp to improve performance. On small and medium inputs performance is 10-20% better. Large inputs use a SIMD loop processing 64 bytes per iteration, which is 30-50% faster depending on the size. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-12-02 18:30:55 +00:00
cmp limit, 160
b.hs L(loop_align)
sub limit, limit, 32
[AArch64] Optimized memcmp. This is an optimized memcmp for AArch64. This is a complete rewrite using a different algorithm. The previous version split into cases where both inputs were aligned, the inputs were mutually aligned and unaligned using a byte loop. The new version combines all these cases, while small inputs of less than 8 bytes are handled separately. This allows the main code to be sped up using unaligned loads since there are now at least 8 bytes to be compared. After the first 8 bytes, align the first input. This ensures each iteration does at most one unaligned access and mutually aligned inputs behave as aligned. After the main loop, process the last 8 bytes using unaligned accesses. This improves performance of (mutually) aligned cases by 25% and unaligned by >500% (yes >6 times faster) on large inputs. * sysdeps/aarch64/memcmp.S (memcmp): Rewrite of optimized memcmp.
2017-08-10 16:00:38 +00:00
.p2align 4
AArch64: Optimize memcmp Rewrite memcmp to improve performance. On small and medium inputs performance is 10-20% better. Large inputs use a SIMD loop processing 64 bytes per iteration, which is 30-50% faster depending on the size. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-12-02 18:30:55 +00:00
L(loop32):
ldp data1, data3, [src1, 16]
ldp data2, data4, [src2, 16]
aarch64: Optimized memcmp for medium to large sizes This improved memcmp provides a fast path for compares up to 16 bytes and then compares 16 bytes at a time, thus optimizing loads from both sources. The glibc memcmp microbenchmark retains performance (with an error of ~1ns) for smaller compare sizes and reduces up to 31% of execution time for compares up to 4K on the APM Mustang. On Qualcomm Falkor this improves to almost 48%, i.e. it is almost 2x improvement for sizes of 2K and above. * sysdeps/aarch64/memcmp.S: Widen comparison to 16 bytes at a time.
2018-03-06 13:52:39 +00:00
cmp data1, data2
AArch64: Optimize memcmp Rewrite memcmp to improve performance. On small and medium inputs performance is 10-20% better. Large inputs use a SIMD loop processing 64 bytes per iteration, which is 30-50% faster depending on the size. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-12-02 18:30:55 +00:00
ccmp data3, data4, 0, eq
b.ne L(return2)
cmp limit, 16
b.ls L(last_bytes)
aarch64: Optimized implementation of memcmp The loop body is expanded from a 16-byte comparison to a 64-byte comparison, and the usage of ldp is replaced by the Post-index mode to the Base plus offset mode. Hence, compare can faster 18% around > 128 bytes in all. Checked on aarch64-linux-gnu. Reviewed-by: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
2019-12-19 12:31:59 +00:00
AArch64: Optimize memcmp Rewrite memcmp to improve performance. On small and medium inputs performance is 10-20% better. Large inputs use a SIMD loop processing 64 bytes per iteration, which is 30-50% faster depending on the size. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-12-02 18:30:55 +00:00
ldp data1, data3, [src1, 32]
ldp data2, data4, [src2, 32]
[AArch64] Optimized memcmp. This is an optimized memcmp for AArch64. This is a complete rewrite using a different algorithm. The previous version split into cases where both inputs were aligned, the inputs were mutually aligned and unaligned using a byte loop. The new version combines all these cases, while small inputs of less than 8 bytes are handled separately. This allows the main code to be sped up using unaligned loads since there are now at least 8 bytes to be compared. After the first 8 bytes, align the first input. This ensures each iteration does at most one unaligned access and mutually aligned inputs behave as aligned. After the main loop, process the last 8 bytes using unaligned accesses. This improves performance of (mutually) aligned cases by 25% and unaligned by >500% (yes >6 times faster) on large inputs. * sysdeps/aarch64/memcmp.S (memcmp): Rewrite of optimized memcmp.
2017-08-10 16:00:38 +00:00
cmp data1, data2
AArch64: Optimize memcmp Rewrite memcmp to improve performance. On small and medium inputs performance is 10-20% better. Large inputs use a SIMD loop processing 64 bytes per iteration, which is 30-50% faster depending on the size. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-12-02 18:30:55 +00:00
ccmp data3, data4, 0, eq
b.ne L(return2)
add src1, src1, 32
add src2, src2, 32
L(last64):
subs limit, limit, 32
b.hi L(loop32)
[AArch64] Optimized memcmp. This is an optimized memcmp for AArch64. This is a complete rewrite using a different algorithm. The previous version split into cases where both inputs were aligned, the inputs were mutually aligned and unaligned using a byte loop. The new version combines all these cases, while small inputs of less than 8 bytes are handled separately. This allows the main code to be sped up using unaligned loads since there are now at least 8 bytes to be compared. After the first 8 bytes, align the first input. This ensures each iteration does at most one unaligned access and mutually aligned inputs behave as aligned. After the main loop, process the last 8 bytes using unaligned accesses. This improves performance of (mutually) aligned cases by 25% and unaligned by >500% (yes >6 times faster) on large inputs. * sysdeps/aarch64/memcmp.S (memcmp): Rewrite of optimized memcmp.
2017-08-10 16:00:38 +00:00
aarch64: Optimized memcmp for medium to large sizes This improved memcmp provides a fast path for compares up to 16 bytes and then compares 16 bytes at a time, thus optimizing loads from both sources. The glibc memcmp microbenchmark retains performance (with an error of ~1ns) for smaller compare sizes and reduces up to 31% of execution time for compares up to 4K on the APM Mustang. On Qualcomm Falkor this improves to almost 48%, i.e. it is almost 2x improvement for sizes of 2K and above. * sysdeps/aarch64/memcmp.S: Widen comparison to 16 bytes at a time.
2018-03-06 13:52:39 +00:00
/* Compare last 1-16 bytes using unaligned access. */
aarch64: Use the L() macro for labels in memcmp The L() macro makes the assembly a bit more readable. * sysdeps/aarch64/memcmp.S: Use L() macro for labels.
2018-02-02 04:45:20 +00:00
L(last_bytes):
AArch64: Optimize memcmp Rewrite memcmp to improve performance. On small and medium inputs performance is 10-20% better. Large inputs use a SIMD loop processing 64 bytes per iteration, which is 30-50% faster depending on the size. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-12-02 18:30:55 +00:00
ldp data1, data3, [src1end, -16]
ldp data2, data4, [src2end, -16]
L(return2):
cmp data1, data2
csel data1, data1, data3, ne
csel data2, data2, data4, ne
[AArch64] Optimized memcmp. This is an optimized memcmp for AArch64. This is a complete rewrite using a different algorithm. The previous version split into cases where both inputs were aligned, the inputs were mutually aligned and unaligned using a byte loop. The new version combines all these cases, while small inputs of less than 8 bytes are handled separately. This allows the main code to be sped up using unaligned loads since there are now at least 8 bytes to be compared. After the first 8 bytes, align the first input. This ensures each iteration does at most one unaligned access and mutually aligned inputs behave as aligned. After the main loop, process the last 8 bytes using unaligned accesses. This improves performance of (mutually) aligned cases by 25% and unaligned by >500% (yes >6 times faster) on large inputs. * sysdeps/aarch64/memcmp.S (memcmp): Rewrite of optimized memcmp.
2017-08-10 16:00:38 +00:00
/* Compare data bytes and set return value to 0, -1 or 1. */
aarch64: Use the L() macro for labels in memcmp The L() macro makes the assembly a bit more readable. * sysdeps/aarch64/memcmp.S: Use L() macro for labels.
2018-02-02 04:45:20 +00:00
L(return):
[AArch64] Optimized memcmp. This is an optimized memcmp for AArch64. This is a complete rewrite using a different algorithm. The previous version split into cases where both inputs were aligned, the inputs were mutually aligned and unaligned using a byte loop. The new version combines all these cases, while small inputs of less than 8 bytes are handled separately. This allows the main code to be sped up using unaligned loads since there are now at least 8 bytes to be compared. After the first 8 bytes, align the first input. This ensures each iteration does at most one unaligned access and mutually aligned inputs behave as aligned. After the main loop, process the last 8 bytes using unaligned accesses. This improves performance of (mutually) aligned cases by 25% and unaligned by >500% (yes >6 times faster) on large inputs. * sysdeps/aarch64/memcmp.S (memcmp): Rewrite of optimized memcmp.
2017-08-10 16:00:38 +00:00
#ifndef __AARCH64EB__
AArch64: Implement optimized memcmp.
2013-01-16 12:58:40 +00:00
rev data1, data1
rev data2, data2
#endif
aarch64: Optimized implementation of memcmp The loop body is expanded from a 16-byte comparison to a 64-byte comparison, and the usage of ldp is replaced by the Post-index mode to the Base plus offset mode. Hence, compare can faster 18% around > 128 bytes in all. Checked on aarch64-linux-gnu. Reviewed-by: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
2019-12-19 12:31:59 +00:00
cmp data1, data2
[AArch64] Optimized memcmp. This is an optimized memcmp for AArch64. This is a complete rewrite using a different algorithm. The previous version split into cases where both inputs were aligned, the inputs were mutually aligned and unaligned using a byte loop. The new version combines all these cases, while small inputs of less than 8 bytes are handled separately. This allows the main code to be sped up using unaligned loads since there are now at least 8 bytes to be compared. After the first 8 bytes, align the first input. This ensures each iteration does at most one unaligned access and mutually aligned inputs behave as aligned. After the main loop, process the last 8 bytes using unaligned accesses. This improves performance of (mutually) aligned cases by 25% and unaligned by >500% (yes >6 times faster) on large inputs. * sysdeps/aarch64/memcmp.S (memcmp): Rewrite of optimized memcmp.
2017-08-10 16:00:38 +00:00
cset result, ne
cneg result, result, lo
ret
.p2align 4
aarch64: Optimized implementation of memcmp The loop body is expanded from a 16-byte comparison to a 64-byte comparison, and the usage of ldp is replaced by the Post-index mode to the Base plus offset mode. Hence, compare can faster 18% around > 128 bytes in all. Checked on aarch64-linux-gnu. Reviewed-by: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
2019-12-19 12:31:59 +00:00
L(less16):
AArch64: Optimize memcmp Rewrite memcmp to improve performance. On small and medium inputs performance is 10-20% better. Large inputs use a SIMD loop processing 64 bytes per iteration, which is 30-50% faster depending on the size. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-12-02 18:30:55 +00:00
add src1end, src1, limit
add src2end, src2, limit
tbz limit, 3, L(less8)
aarch64: Optimized implementation of memcmp The loop body is expanded from a 16-byte comparison to a 64-byte comparison, and the usage of ldp is replaced by the Post-index mode to the Base plus offset mode. Hence, compare can faster 18% around > 128 bytes in all. Checked on aarch64-linux-gnu. Reviewed-by: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
2019-12-19 12:31:59 +00:00
ldr data1, [src1]
ldr data2, [src2]
AArch64: Optimize memcmp Rewrite memcmp to improve performance. On small and medium inputs performance is 10-20% better. Large inputs use a SIMD loop processing 64 bytes per iteration, which is 30-50% faster depending on the size. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-12-02 18:30:55 +00:00
ldr data3, [src1end, -8]
ldr data4, [src2end, -8]
b L(return2)
aarch64: Optimized implementation of memcmp The loop body is expanded from a 16-byte comparison to a 64-byte comparison, and the usage of ldp is replaced by the Post-index mode to the Base plus offset mode. Hence, compare can faster 18% around > 128 bytes in all. Checked on aarch64-linux-gnu. Reviewed-by: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
2019-12-19 12:31:59 +00:00
.p2align 4
aarch64: Use the L() macro for labels in memcmp The L() macro makes the assembly a bit more readable. * sysdeps/aarch64/memcmp.S: Use L() macro for labels.
2018-02-02 04:45:20 +00:00
L(less8):
AArch64: Optimize memcmp Rewrite memcmp to improve performance. On small and medium inputs performance is 10-20% better. Large inputs use a SIMD loop processing 64 bytes per iteration, which is 30-50% faster depending on the size. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-12-02 18:30:55 +00:00
tbz limit, 2, L(less4)
aarch64: Optimized implementation of memcmp The loop body is expanded from a 16-byte comparison to a 64-byte comparison, and the usage of ldp is replaced by the Post-index mode to the Base plus offset mode. Hence, compare can faster 18% around > 128 bytes in all. Checked on aarch64-linux-gnu. Reviewed-by: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
2019-12-19 12:31:59 +00:00
ldr data1w, [src1]
ldr data2w, [src2]
AArch64: Optimize memcmp Rewrite memcmp to improve performance. On small and medium inputs performance is 10-20% better. Large inputs use a SIMD loop processing 64 bytes per iteration, which is 30-50% faster depending on the size. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-12-02 18:30:55 +00:00
ldr data3w, [src1end, -4]
ldr data4w, [src2end, -4]
b L(return2)
aarch64: Optimized implementation of memcmp The loop body is expanded from a 16-byte comparison to a 64-byte comparison, and the usage of ldp is replaced by the Post-index mode to the Base plus offset mode. Hence, compare can faster 18% around > 128 bytes in all. Checked on aarch64-linux-gnu. Reviewed-by: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
2019-12-19 12:31:59 +00:00
aarch64: Use the L() macro for labels in memcmp The L() macro makes the assembly a bit more readable. * sysdeps/aarch64/memcmp.S: Use L() macro for labels.
2018-02-02 04:45:20 +00:00
L(less4):
AArch64: Optimize memcmp Rewrite memcmp to improve performance. On small and medium inputs performance is 10-20% better. Large inputs use a SIMD loop processing 64 bytes per iteration, which is 30-50% faster depending on the size. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-12-02 18:30:55 +00:00
tbz limit, 1, L(less2)
ldrh data1w, [src1]
ldrh data2w, [src2]
cmp data1w, data2w
b.ne L(return)
L(less2):
mov result, 0
tbz limit, 0, L(return_zero)
ldrb data1w, [src1end, -1]
ldrb data2w, [src2end, -1]
[AArch64] Optimized memcmp. This is an optimized memcmp for AArch64. This is a complete rewrite using a different algorithm. The previous version split into cases where both inputs were aligned, the inputs were mutually aligned and unaligned using a byte loop. The new version combines all these cases, while small inputs of less than 8 bytes are handled separately. This allows the main code to be sped up using unaligned loads since there are now at least 8 bytes to be compared. After the first 8 bytes, align the first input. This ensures each iteration does at most one unaligned access and mutually aligned inputs behave as aligned. After the main loop, process the last 8 bytes using unaligned accesses. This improves performance of (mutually) aligned cases by 25% and unaligned by >500% (yes >6 times faster) on large inputs. * sysdeps/aarch64/memcmp.S (memcmp): Rewrite of optimized memcmp.
2017-08-10 16:00:38 +00:00
sub result, data1w, data2w
AArch64: Optimize memcmp Rewrite memcmp to improve performance. On small and medium inputs performance is 10-20% better. Large inputs use a SIMD loop processing 64 bytes per iteration, which is 30-50% faster depending on the size. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-12-02 18:30:55 +00:00
L(return_zero):
[AArch64] Optimized memcmp. This is an optimized memcmp for AArch64. This is a complete rewrite using a different algorithm. The previous version split into cases where both inputs were aligned, the inputs were mutually aligned and unaligned using a byte loop. The new version combines all these cases, while small inputs of less than 8 bytes are handled separately. This allows the main code to be sped up using unaligned loads since there are now at least 8 bytes to be compared. After the first 8 bytes, align the first input. This ensures each iteration does at most one unaligned access and mutually aligned inputs behave as aligned. After the main loop, process the last 8 bytes using unaligned accesses. This improves performance of (mutually) aligned cases by 25% and unaligned by >500% (yes >6 times faster) on large inputs. * sysdeps/aarch64/memcmp.S (memcmp): Rewrite of optimized memcmp.
2017-08-10 16:00:38 +00:00
ret
AArch64: Optimize memcmp Rewrite memcmp to improve performance. On small and medium inputs performance is 10-20% better. Large inputs use a SIMD loop processing 64 bytes per iteration, which is 30-50% faster depending on the size. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-12-02 18:30:55 +00:00
L(loop_align):
ldp data1, data3, [src1, 16]
ldp data2, data4, [src2, 16]
cmp data1, data2
ccmp data3, data4, 0, eq
b.ne L(return2)
/* Align src2 and adjust src1, src2 and limit. */
and tmp, src2, 15
sub tmp, tmp, 16
sub src2, src2, tmp
add limit, limit, tmp
sub src1, src1, tmp
sub limit, limit, 64 + 16
.p2align 4
L(loop64):
ldr q0, [src1, 16]
ldr q1, [src2, 16]
subs limit, limit, 64
ldr q2, [src1, 32]
ldr q3, [src2, 32]
eor v0.16b, v0.16b, v1.16b
eor v1.16b, v2.16b, v3.16b
ldr q2, [src1, 48]
ldr q3, [src2, 48]
umaxp v0.16b, v0.16b, v1.16b
ldr q4, [src1, 64]!
ldr q5, [src2, 64]!
eor v1.16b, v2.16b, v3.16b
eor v2.16b, v4.16b, v5.16b
umaxp v1.16b, v1.16b, v2.16b
umaxp v0.16b, v0.16b, v1.16b
umaxp v0.16b, v0.16b, v0.16b
fmov tmp, d0
ccmp tmp, 0, 0, hi
b.eq L(loop64)
/* If equal, process last 1-64 bytes using scalar loop. */
add limit, limit, 64 + 16
cbz tmp, L(last64)
/* Determine the 8-byte aligned offset of the first difference. */
#ifdef __AARCH64EB__
rev16 tmp, tmp
#endif
rev tmp, tmp
clz tmp, tmp
bic tmp, tmp, 7
sub tmp, tmp, 48
ldr data1, [src1, tmp]
ldr data2, [src2, tmp]
#ifndef __AARCH64EB__
rev data1, data1
rev data2, data2
#endif
mov result, 1
cmp data1, data2
cneg result, result, lo
aarch64: Optimized implementation of memcmp The loop body is expanded from a 16-byte comparison to a 64-byte comparison, and the usage of ldp is replaced by the Post-index mode to the Base plus offset mode. Hence, compare can faster 18% around > 128 bytes in all. Checked on aarch64-linux-gnu. Reviewed-by: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
2019-12-19 12:31:59 +00:00
ret
[AArch64] Optimized memcmp. This is an optimized memcmp for AArch64. This is a complete rewrite using a different algorithm. The previous version split into cases where both inputs were aligned, the inputs were mutually aligned and unaligned using a byte loop. The new version combines all these cases, while small inputs of less than 8 bytes are handled separately. This allows the main code to be sped up using unaligned loads since there are now at least 8 bytes to be compared. After the first 8 bytes, align the first input. This ensures each iteration does at most one unaligned access and mutually aligned inputs behave as aligned. After the main loop, process the last 8 bytes using unaligned accesses. This improves performance of (mutually) aligned cases by 25% and unaligned by >500% (yes >6 times faster) on large inputs. * sysdeps/aarch64/memcmp.S (memcmp): Rewrite of optimized memcmp.
2017-08-10 16:00:38 +00:00
AArch64: Implement optimized memcmp.
2013-01-16 12:58:40 +00:00
END (memcmp)
#undef bcmp
weak_alias (memcmp, bcmp)
String: Add support for __memcmpeq() ABI on all targets No bug. This commit adds support for __memcmpeq() as a new ABI for all targets. In this commit __memcmpeq() is implemented only as an alias to the corresponding targets memcmp() implementation. __memcmpeq() is added as a new symbol starting with GLIBC_2.35 and defined in string.h with comments explaining its behavior. Basic tests that it is callable and works where added in string/tester.c As discussed in the proposal "Add new ABI '__memcmpeq()' to libc" __memcmpeq() is essentially a reserved namespace for bcmp(). The means is shares the same specifications as memcmp() except the return value for non-equal byte sequences is any non-zero value. This is less strict than memcmp()'s return value specification and can be better optimized when a boolean return is all that is needed. __memcmpeq() is meant to only be called by compilers if they can prove that the return value of a memcmp() call is only used for its boolean value. All tests in string/tester.c passed. As well build succeeds on x86_64-linux-gnu target.
2021-10-21 20:54:57 +00:00
#undef __memcmpeq
strong_alias (memcmp, __memcmpeq)
AArch64: Implement optimized memcmp.
2013-01-16 12:58:40 +00:00
libc_hidden_builtin_def (memcmp)
String: Add hidden defs for __memcmpeq() to enable internal usage No bug. This commit adds hidden defs for all declarations of __memcmpeq. This enables usage of __memcmpeq without the PLT for usage internal to GLIBC.
2021-10-21 20:53:58 +00:00
libc_hidden_def (__memcmpeq)
Reference in New Issue Copy Permalink