2021-03-05 14:24:52 +00:00
|
|
|
/* memchr/wmemchr optimized with 256-bit EVEX instructions.
|
2022-01-01 18:54:23 +00:00
|
|
|
Copyright (C) 2021-2022 Free Software Foundation, Inc.
|
2021-03-05 14:24:52 +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
|
|
|
|
<https://www.gnu.org/licenses/>. */
|
|
|
|
|
|
|
|
#if IS_IN (libc)
|
|
|
|
|
|
|
|
# include <sysdep.h>
|
|
|
|
|
|
|
|
# ifndef MEMCHR
|
|
|
|
# define MEMCHR __memchr_evex
|
|
|
|
# endif
|
|
|
|
|
|
|
|
# ifdef USE_AS_WMEMCHR
|
|
|
|
# define VPBROADCAST vpbroadcastd
|
2021-05-03 07:03:19 +00:00
|
|
|
# define VPMINU vpminud
|
|
|
|
# define VPCMP vpcmpd
|
|
|
|
# define VPCMPEQ vpcmpeqd
|
|
|
|
# define CHAR_SIZE 4
|
2021-03-05 14:24:52 +00:00
|
|
|
# else
|
|
|
|
# define VPBROADCAST vpbroadcastb
|
2021-05-03 07:03:19 +00:00
|
|
|
# define VPMINU vpminub
|
|
|
|
# define VPCMP vpcmpb
|
|
|
|
# define VPCMPEQ vpcmpeqb
|
|
|
|
# define CHAR_SIZE 1
|
2021-03-05 14:24:52 +00:00
|
|
|
# endif
|
|
|
|
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
/* In the 4x loop the RTM and non-RTM versions have data pointer
|
|
|
|
off by VEC_SIZE * 4 with RTM version being VEC_SIZE * 4 greater.
|
|
|
|
This is represented by BASE_OFFSET. As well because the RTM
|
|
|
|
version uses vpcmp which stores a bit per element compared where
|
|
|
|
the non-RTM version uses vpcmpeq which stores a bit per byte
|
|
|
|
compared RET_SCALE of CHAR_SIZE is only relevant for the RTM
|
|
|
|
version. */
|
|
|
|
# ifdef USE_IN_RTM
|
|
|
|
# define VZEROUPPER
|
|
|
|
# define BASE_OFFSET (VEC_SIZE * 4)
|
|
|
|
# define RET_SCALE CHAR_SIZE
|
|
|
|
# else
|
|
|
|
# define VZEROUPPER vzeroupper
|
|
|
|
# define BASE_OFFSET 0
|
|
|
|
# define RET_SCALE 1
|
|
|
|
# endif
|
|
|
|
|
|
|
|
/* In the return from 4x loop memchr and rawmemchr versions have
|
|
|
|
data pointers off by VEC_SIZE * 4 with memchr version being
|
|
|
|
VEC_SIZE * 4 greater. */
|
2021-05-03 07:03:19 +00:00
|
|
|
# ifdef USE_AS_RAWMEMCHR
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
# define RET_OFFSET (BASE_OFFSET - (VEC_SIZE * 4))
|
2021-05-03 07:03:19 +00:00
|
|
|
# define RAW_PTR_REG rcx
|
|
|
|
# define ALGN_PTR_REG rdi
|
|
|
|
# else
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
# define RET_OFFSET BASE_OFFSET
|
2021-05-03 07:03:19 +00:00
|
|
|
# define RAW_PTR_REG rdi
|
|
|
|
# define ALGN_PTR_REG rcx
|
|
|
|
# endif
|
|
|
|
|
|
|
|
# define XMMZERO xmm23
|
|
|
|
# define YMMZERO ymm23
|
2021-03-05 14:24:52 +00:00
|
|
|
# define XMMMATCH xmm16
|
|
|
|
# define YMMMATCH ymm16
|
|
|
|
# define YMM1 ymm17
|
|
|
|
# define YMM2 ymm18
|
|
|
|
# define YMM3 ymm19
|
|
|
|
# define YMM4 ymm20
|
|
|
|
# define YMM5 ymm21
|
|
|
|
# define YMM6 ymm22
|
|
|
|
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
# ifndef SECTION
|
|
|
|
# define SECTION(p) p##.evex
|
|
|
|
# endif
|
|
|
|
|
2021-03-05 14:24:52 +00:00
|
|
|
# define VEC_SIZE 32
|
2021-05-03 07:03:19 +00:00
|
|
|
# define CHAR_PER_VEC (VEC_SIZE / CHAR_SIZE)
|
|
|
|
# define PAGE_SIZE 4096
|
2021-03-05 14:24:52 +00:00
|
|
|
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
.section SECTION(.text),"ax",@progbits
|
2021-03-05 14:24:52 +00:00
|
|
|
ENTRY (MEMCHR)
|
|
|
|
# ifndef USE_AS_RAWMEMCHR
|
|
|
|
/* Check for zero length. */
|
|
|
|
test %RDX_LP, %RDX_LP
|
|
|
|
jz L(zero)
|
2021-05-03 07:03:19 +00:00
|
|
|
|
2021-03-05 14:24:52 +00:00
|
|
|
# ifdef __ILP32__
|
|
|
|
/* Clear the upper 32 bits. */
|
|
|
|
movl %edx, %edx
|
|
|
|
# endif
|
|
|
|
# endif
|
|
|
|
/* Broadcast CHAR to YMMMATCH. */
|
|
|
|
VPBROADCAST %esi, %YMMMATCH
|
|
|
|
/* Check if we may cross page boundary with one vector load. */
|
2021-05-03 07:03:19 +00:00
|
|
|
movl %edi, %eax
|
|
|
|
andl $(PAGE_SIZE - 1), %eax
|
|
|
|
cmpl $(PAGE_SIZE - VEC_SIZE), %eax
|
|
|
|
ja L(cross_page_boundary)
|
2021-03-05 14:24:52 +00:00
|
|
|
|
|
|
|
/* Check the first VEC_SIZE bytes. */
|
2021-05-03 07:03:19 +00:00
|
|
|
VPCMP $0, (%rdi), %YMMMATCH, %k0
|
|
|
|
kmovd %k0, %eax
|
2021-03-05 14:24:52 +00:00
|
|
|
# ifndef USE_AS_RAWMEMCHR
|
2021-05-03 07:03:19 +00:00
|
|
|
/* If length < CHAR_PER_VEC handle special. */
|
|
|
|
cmpq $CHAR_PER_VEC, %rdx
|
|
|
|
jbe L(first_vec_x0)
|
|
|
|
# endif
|
|
|
|
testl %eax, %eax
|
|
|
|
jz L(aligned_more)
|
|
|
|
tzcntl %eax, %eax
|
|
|
|
# ifdef USE_AS_WMEMCHR
|
|
|
|
/* NB: Multiply bytes by CHAR_SIZE to get the wchar_t count. */
|
|
|
|
leaq (%rdi, %rax, CHAR_SIZE), %rax
|
2021-03-05 14:24:52 +00:00
|
|
|
# else
|
2021-05-03 07:03:19 +00:00
|
|
|
addq %rdi, %rax
|
2021-03-05 14:24:52 +00:00
|
|
|
# endif
|
2021-05-03 07:03:19 +00:00
|
|
|
ret
|
2021-03-05 14:24:52 +00:00
|
|
|
|
|
|
|
# ifndef USE_AS_RAWMEMCHR
|
2021-05-03 07:03:19 +00:00
|
|
|
L(zero):
|
|
|
|
xorl %eax, %eax
|
|
|
|
ret
|
2021-03-05 14:24:52 +00:00
|
|
|
|
2021-05-03 07:03:19 +00:00
|
|
|
.p2align 5
|
|
|
|
L(first_vec_x0):
|
|
|
|
/* Check if first match was before length. */
|
|
|
|
tzcntl %eax, %eax
|
|
|
|
xorl %ecx, %ecx
|
|
|
|
cmpl %eax, %edx
|
|
|
|
leaq (%rdi, %rax, CHAR_SIZE), %rax
|
|
|
|
cmovle %rcx, %rax
|
|
|
|
ret
|
|
|
|
# else
|
|
|
|
/* NB: first_vec_x0 is 17 bytes which will leave
|
|
|
|
cross_page_boundary (which is relatively cold) close enough
|
|
|
|
to ideal alignment. So only realign L(cross_page_boundary) if
|
|
|
|
rawmemchr. */
|
2021-03-05 14:24:52 +00:00
|
|
|
.p2align 4
|
2021-05-03 07:03:19 +00:00
|
|
|
# endif
|
|
|
|
L(cross_page_boundary):
|
|
|
|
/* Save pointer before aligning as its original value is
|
|
|
|
necessary for computer return address if byte is found or
|
|
|
|
adjusting length if it is not and this is memchr. */
|
|
|
|
movq %rdi, %rcx
|
|
|
|
/* Align data to VEC_SIZE. ALGN_PTR_REG is rcx for memchr and rdi
|
|
|
|
for rawmemchr. */
|
|
|
|
andq $-VEC_SIZE, %ALGN_PTR_REG
|
|
|
|
VPCMP $0, (%ALGN_PTR_REG), %YMMMATCH, %k0
|
|
|
|
kmovd %k0, %r8d
|
2021-03-05 14:24:52 +00:00
|
|
|
# ifdef USE_AS_WMEMCHR
|
2021-05-03 07:03:19 +00:00
|
|
|
/* NB: Divide shift count by 4 since each bit in K0 represent 4
|
2021-03-05 14:24:52 +00:00
|
|
|
bytes. */
|
2021-05-03 07:03:19 +00:00
|
|
|
sarl $2, %eax
|
|
|
|
# endif
|
|
|
|
# ifndef USE_AS_RAWMEMCHR
|
|
|
|
movl $(PAGE_SIZE / CHAR_SIZE), %esi
|
|
|
|
subl %eax, %esi
|
2021-03-05 14:24:52 +00:00
|
|
|
# endif
|
|
|
|
# ifdef USE_AS_WMEMCHR
|
2021-05-03 07:03:19 +00:00
|
|
|
andl $(CHAR_PER_VEC - 1), %eax
|
2021-03-05 14:24:52 +00:00
|
|
|
# endif
|
2021-05-03 07:03:19 +00:00
|
|
|
/* Remove the leading bytes. */
|
|
|
|
sarxl %eax, %r8d, %eax
|
2021-03-05 14:24:52 +00:00
|
|
|
# ifndef USE_AS_RAWMEMCHR
|
|
|
|
/* Check the end of data. */
|
2021-05-03 07:03:19 +00:00
|
|
|
cmpq %rsi, %rdx
|
|
|
|
jbe L(first_vec_x0)
|
|
|
|
# endif
|
|
|
|
testl %eax, %eax
|
|
|
|
jz L(cross_page_continue)
|
|
|
|
tzcntl %eax, %eax
|
|
|
|
# ifdef USE_AS_WMEMCHR
|
|
|
|
/* NB: Multiply bytes by CHAR_SIZE to get the wchar_t count. */
|
|
|
|
leaq (%RAW_PTR_REG, %rax, CHAR_SIZE), %rax
|
|
|
|
# else
|
|
|
|
addq %RAW_PTR_REG, %rax
|
2021-03-05 14:24:52 +00:00
|
|
|
# endif
|
|
|
|
ret
|
|
|
|
|
|
|
|
.p2align 4
|
2021-05-03 07:03:19 +00:00
|
|
|
L(first_vec_x1):
|
|
|
|
tzcntl %eax, %eax
|
|
|
|
leaq VEC_SIZE(%rdi, %rax, CHAR_SIZE), %rax
|
|
|
|
ret
|
2021-03-05 14:24:52 +00:00
|
|
|
|
2021-05-03 07:03:19 +00:00
|
|
|
.p2align 4
|
|
|
|
L(first_vec_x2):
|
|
|
|
tzcntl %eax, %eax
|
|
|
|
leaq (VEC_SIZE * 2)(%rdi, %rax, CHAR_SIZE), %rax
|
|
|
|
ret
|
2021-03-05 14:24:52 +00:00
|
|
|
|
2021-05-03 07:03:19 +00:00
|
|
|
.p2align 4
|
|
|
|
L(first_vec_x3):
|
|
|
|
tzcntl %eax, %eax
|
|
|
|
leaq (VEC_SIZE * 3)(%rdi, %rax, CHAR_SIZE), %rax
|
|
|
|
ret
|
2021-03-05 14:24:52 +00:00
|
|
|
|
2021-05-03 07:03:19 +00:00
|
|
|
.p2align 4
|
|
|
|
L(first_vec_x4):
|
|
|
|
tzcntl %eax, %eax
|
|
|
|
leaq (VEC_SIZE * 4)(%rdi, %rax, CHAR_SIZE), %rax
|
|
|
|
ret
|
2021-03-05 14:24:52 +00:00
|
|
|
|
2021-05-03 07:03:19 +00:00
|
|
|
.p2align 5
|
|
|
|
L(aligned_more):
|
2021-03-05 14:24:52 +00:00
|
|
|
/* Check the first 4 * VEC_SIZE. Only one VEC_SIZE at a time
|
|
|
|
since data is only aligned to VEC_SIZE. */
|
|
|
|
|
2021-05-03 07:03:19 +00:00
|
|
|
# ifndef USE_AS_RAWMEMCHR
|
|
|
|
/* Align data to VEC_SIZE. */
|
|
|
|
L(cross_page_continue):
|
|
|
|
xorl %ecx, %ecx
|
|
|
|
subl %edi, %ecx
|
|
|
|
andq $-VEC_SIZE, %rdi
|
|
|
|
/* esi is for adjusting length to see if near the end. */
|
|
|
|
leal (VEC_SIZE * 5)(%rdi, %rcx), %esi
|
|
|
|
# ifdef USE_AS_WMEMCHR
|
|
|
|
/* NB: Divide bytes by 4 to get the wchar_t count. */
|
|
|
|
sarl $2, %esi
|
|
|
|
# endif
|
|
|
|
# else
|
|
|
|
andq $-VEC_SIZE, %rdi
|
|
|
|
L(cross_page_continue):
|
|
|
|
# endif
|
|
|
|
/* Load first VEC regardless. */
|
|
|
|
VPCMP $0, (VEC_SIZE)(%rdi), %YMMMATCH, %k0
|
|
|
|
kmovd %k0, %eax
|
|
|
|
# ifndef USE_AS_RAWMEMCHR
|
|
|
|
/* Adjust length. If near end handle specially. */
|
|
|
|
subq %rsi, %rdx
|
|
|
|
jbe L(last_4x_vec_or_less)
|
|
|
|
# endif
|
2021-03-05 14:24:52 +00:00
|
|
|
testl %eax, %eax
|
|
|
|
jnz L(first_vec_x1)
|
|
|
|
|
2021-05-03 07:03:19 +00:00
|
|
|
VPCMP $0, (VEC_SIZE * 2)(%rdi), %YMMMATCH, %k0
|
|
|
|
kmovd %k0, %eax
|
2021-03-05 14:24:52 +00:00
|
|
|
testl %eax, %eax
|
|
|
|
jnz L(first_vec_x2)
|
|
|
|
|
2021-05-03 07:03:19 +00:00
|
|
|
VPCMP $0, (VEC_SIZE * 3)(%rdi), %YMMMATCH, %k0
|
|
|
|
kmovd %k0, %eax
|
2021-03-05 14:24:52 +00:00
|
|
|
testl %eax, %eax
|
|
|
|
jnz L(first_vec_x3)
|
|
|
|
|
2021-05-03 07:03:19 +00:00
|
|
|
VPCMP $0, (VEC_SIZE * 4)(%rdi), %YMMMATCH, %k0
|
|
|
|
kmovd %k0, %eax
|
|
|
|
testl %eax, %eax
|
|
|
|
jnz L(first_vec_x4)
|
|
|
|
|
2021-03-05 14:24:52 +00:00
|
|
|
|
|
|
|
# ifndef USE_AS_RAWMEMCHR
|
2021-05-03 07:03:19 +00:00
|
|
|
/* Check if at last CHAR_PER_VEC * 4 length. */
|
|
|
|
subq $(CHAR_PER_VEC * 4), %rdx
|
|
|
|
jbe L(last_4x_vec_or_less_cmpeq)
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
/* +VEC_SIZE if USE_IN_RTM otherwise +VEC_SIZE * 5. */
|
|
|
|
addq $(VEC_SIZE + (VEC_SIZE * 4 - BASE_OFFSET)), %rdi
|
2021-03-05 14:24:52 +00:00
|
|
|
|
2021-05-03 07:03:19 +00:00
|
|
|
/* Align data to VEC_SIZE * 4 for the loop and readjust length.
|
|
|
|
*/
|
|
|
|
# ifdef USE_AS_WMEMCHR
|
|
|
|
movl %edi, %ecx
|
2021-03-05 14:24:52 +00:00
|
|
|
andq $-(4 * VEC_SIZE), %rdi
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
subl %edi, %ecx
|
2021-05-03 07:03:19 +00:00
|
|
|
/* NB: Divide bytes by 4 to get the wchar_t count. */
|
|
|
|
sarl $2, %ecx
|
2021-03-05 14:24:52 +00:00
|
|
|
addq %rcx, %rdx
|
2021-05-03 07:03:19 +00:00
|
|
|
# else
|
|
|
|
addq %rdi, %rdx
|
|
|
|
andq $-(4 * VEC_SIZE), %rdi
|
|
|
|
subq %rdi, %rdx
|
|
|
|
# endif
|
|
|
|
# else
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
addq $(VEC_SIZE + (VEC_SIZE * 4 - BASE_OFFSET)), %rdi
|
2021-05-03 07:03:19 +00:00
|
|
|
andq $-(4 * VEC_SIZE), %rdi
|
2021-03-05 14:24:52 +00:00
|
|
|
# endif
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
# ifdef USE_IN_RTM
|
2021-05-03 07:03:19 +00:00
|
|
|
vpxorq %XMMZERO, %XMMZERO, %XMMZERO
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
# else
|
|
|
|
/* copy ymmmatch to ymm0 so we can use vpcmpeq which is not
|
|
|
|
encodable with EVEX registers (ymm16-ymm31). */
|
|
|
|
vmovdqa64 %YMMMATCH, %ymm0
|
|
|
|
# endif
|
2021-05-03 07:03:19 +00:00
|
|
|
|
|
|
|
/* Compare 4 * VEC at a time forward. */
|
2021-03-05 14:24:52 +00:00
|
|
|
.p2align 4
|
|
|
|
L(loop_4x_vec):
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
/* Two versions of the loop. One that does not require
|
|
|
|
vzeroupper by not using ymm0-ymm15 and another does that require
|
|
|
|
vzeroupper because it uses ymm0-ymm15. The reason why ymm0-ymm15
|
|
|
|
is used at all is because there is no EVEX encoding vpcmpeq and
|
|
|
|
with vpcmpeq this loop can be performed more efficiently. The
|
|
|
|
non-vzeroupper version is safe for RTM while the vzeroupper
|
|
|
|
version should be prefered if RTM are not supported. */
|
|
|
|
# ifdef USE_IN_RTM
|
2021-05-03 07:03:19 +00:00
|
|
|
/* It would be possible to save some instructions using 4x VPCMP
|
|
|
|
but bottleneck on port 5 makes it not woth it. */
|
|
|
|
VPCMP $4, (VEC_SIZE * 4)(%rdi), %YMMMATCH, %k1
|
|
|
|
/* xor will set bytes match esi to zero. */
|
|
|
|
vpxorq (VEC_SIZE * 5)(%rdi), %YMMMATCH, %YMM2
|
|
|
|
vpxorq (VEC_SIZE * 6)(%rdi), %YMMMATCH, %YMM3
|
|
|
|
VPCMP $0, (VEC_SIZE * 7)(%rdi), %YMMMATCH, %k3
|
|
|
|
/* Reduce VEC2 / VEC3 with min and VEC1 with zero mask. */
|
2021-05-08 02:03:21 +00:00
|
|
|
VPMINU %YMM2, %YMM3, %YMM3{%k1}{z}
|
2021-05-03 07:03:19 +00:00
|
|
|
VPCMP $0, %YMM3, %YMMZERO, %k2
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
# else
|
|
|
|
/* Since vptern can only take 3x vectors fastest to do 1 vec
|
|
|
|
seperately with EVEX vpcmp. */
|
|
|
|
# ifdef USE_AS_WMEMCHR
|
|
|
|
/* vptern can only accept masks for epi32/epi64 so can only save
|
|
|
|
instruction using not equals mask on vptern with wmemchr. */
|
|
|
|
VPCMP $4, (%rdi), %YMMMATCH, %k1
|
|
|
|
# else
|
|
|
|
VPCMP $0, (%rdi), %YMMMATCH, %k1
|
|
|
|
# endif
|
|
|
|
/* Compare 3x with vpcmpeq and or them all together with vptern.
|
|
|
|
*/
|
|
|
|
VPCMPEQ VEC_SIZE(%rdi), %ymm0, %ymm2
|
|
|
|
VPCMPEQ (VEC_SIZE * 2)(%rdi), %ymm0, %ymm3
|
|
|
|
VPCMPEQ (VEC_SIZE * 3)(%rdi), %ymm0, %ymm4
|
|
|
|
# ifdef USE_AS_WMEMCHR
|
|
|
|
/* This takes the not of or between ymm2, ymm3, ymm4 as well as
|
|
|
|
combines result from VEC0 with zero mask. */
|
|
|
|
vpternlogd $1, %ymm2, %ymm3, %ymm4{%k1}{z}
|
|
|
|
vpmovmskb %ymm4, %ecx
|
|
|
|
# else
|
|
|
|
/* 254 is mask for oring ymm2, ymm3, ymm4 into ymm4. */
|
|
|
|
vpternlogd $254, %ymm2, %ymm3, %ymm4
|
|
|
|
vpmovmskb %ymm4, %ecx
|
|
|
|
kmovd %k1, %eax
|
|
|
|
# endif
|
|
|
|
# endif
|
|
|
|
|
2021-03-05 14:24:52 +00:00
|
|
|
# ifdef USE_AS_RAWMEMCHR
|
2021-05-03 07:03:19 +00:00
|
|
|
subq $-(VEC_SIZE * 4), %rdi
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
# endif
|
|
|
|
# ifdef USE_IN_RTM
|
2021-05-03 07:03:19 +00:00
|
|
|
kortestd %k2, %k3
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
# else
|
|
|
|
# ifdef USE_AS_WMEMCHR
|
|
|
|
/* ecx contains not of matches. All 1s means no matches. incl will
|
|
|
|
overflow and set zeroflag if that is the case. */
|
|
|
|
incl %ecx
|
|
|
|
# else
|
|
|
|
/* If either VEC1 (eax) or VEC2-VEC4 (ecx) are not zero. Adding
|
|
|
|
to ecx is not an issue because if eax is non-zero it will be
|
|
|
|
used for returning the match. If it is zero the add does
|
|
|
|
nothing. */
|
|
|
|
addq %rax, %rcx
|
|
|
|
# endif
|
|
|
|
# endif
|
|
|
|
# ifdef USE_AS_RAWMEMCHR
|
2021-05-03 07:03:19 +00:00
|
|
|
jz L(loop_4x_vec)
|
2021-03-05 14:24:52 +00:00
|
|
|
# else
|
2021-05-03 07:03:19 +00:00
|
|
|
jnz L(loop_4x_vec_end)
|
|
|
|
|
|
|
|
subq $-(VEC_SIZE * 4), %rdi
|
|
|
|
|
|
|
|
subq $(CHAR_PER_VEC * 4), %rdx
|
2021-03-05 14:24:52 +00:00
|
|
|
ja L(loop_4x_vec)
|
|
|
|
|
2021-05-03 07:03:19 +00:00
|
|
|
/* Fall through into less than 4 remaining vectors of length case.
|
|
|
|
*/
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
VPCMP $0, BASE_OFFSET(%rdi), %YMMMATCH, %k0
|
|
|
|
addq $(BASE_OFFSET - VEC_SIZE), %rdi
|
2021-05-03 07:03:19 +00:00
|
|
|
kmovd %k0, %eax
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
VZEROUPPER
|
|
|
|
|
2021-03-05 14:24:52 +00:00
|
|
|
L(last_4x_vec_or_less):
|
2021-05-03 07:03:19 +00:00
|
|
|
/* Check if first VEC contained match. */
|
2021-03-05 14:24:52 +00:00
|
|
|
testl %eax, %eax
|
2021-05-03 07:03:19 +00:00
|
|
|
jnz L(first_vec_x1_check)
|
2021-03-05 14:24:52 +00:00
|
|
|
|
2021-05-03 07:03:19 +00:00
|
|
|
/* If remaining length > CHAR_PER_VEC * 2. */
|
|
|
|
addl $(CHAR_PER_VEC * 2), %edx
|
|
|
|
jg L(last_4x_vec)
|
2021-03-05 14:24:52 +00:00
|
|
|
|
2021-05-03 07:03:19 +00:00
|
|
|
L(last_2x_vec):
|
|
|
|
/* If remaining length < CHAR_PER_VEC. */
|
|
|
|
addl $CHAR_PER_VEC, %edx
|
|
|
|
jle L(zero_end)
|
2021-03-05 14:24:52 +00:00
|
|
|
|
2021-05-03 07:03:19 +00:00
|
|
|
/* Check VEC2 and compare any match with remaining length. */
|
|
|
|
VPCMP $0, (VEC_SIZE * 2)(%rdi), %YMMMATCH, %k0
|
|
|
|
kmovd %k0, %eax
|
|
|
|
tzcntl %eax, %eax
|
|
|
|
cmpl %eax, %edx
|
|
|
|
jbe L(set_zero_end)
|
|
|
|
leaq (VEC_SIZE * 2)(%rdi, %rax, CHAR_SIZE), %rax
|
|
|
|
L(zero_end):
|
|
|
|
ret
|
2021-03-05 14:24:52 +00:00
|
|
|
|
|
|
|
|
2021-05-03 07:03:19 +00:00
|
|
|
.p2align 4
|
|
|
|
L(first_vec_x1_check):
|
|
|
|
tzcntl %eax, %eax
|
|
|
|
/* Adjust length. */
|
|
|
|
subl $-(CHAR_PER_VEC * 4), %edx
|
|
|
|
/* Check if match within remaining length. */
|
|
|
|
cmpl %eax, %edx
|
|
|
|
jbe L(set_zero_end)
|
|
|
|
/* NB: Multiply bytes by CHAR_SIZE to get the wchar_t count. */
|
|
|
|
leaq VEC_SIZE(%rdi, %rax, CHAR_SIZE), %rax
|
|
|
|
ret
|
|
|
|
L(set_zero_end):
|
2021-03-05 14:24:52 +00:00
|
|
|
xorl %eax, %eax
|
|
|
|
ret
|
|
|
|
|
|
|
|
.p2align 4
|
2021-05-03 07:03:19 +00:00
|
|
|
L(loop_4x_vec_end):
|
|
|
|
# endif
|
|
|
|
/* rawmemchr will fall through into this if match was found in
|
|
|
|
loop. */
|
|
|
|
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
# if defined USE_IN_RTM || defined USE_AS_WMEMCHR
|
2021-05-03 07:03:19 +00:00
|
|
|
/* k1 has not of matches with VEC1. */
|
2021-03-05 14:24:52 +00:00
|
|
|
kmovd %k1, %eax
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
# ifdef USE_AS_WMEMCHR
|
2021-05-03 07:03:19 +00:00
|
|
|
subl $((1 << CHAR_PER_VEC) - 1), %eax
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
# else
|
2021-05-03 07:03:19 +00:00
|
|
|
incl %eax
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
# endif
|
|
|
|
# else
|
|
|
|
/* eax already has matches for VEC1. */
|
|
|
|
testl %eax, %eax
|
2021-05-03 07:03:19 +00:00
|
|
|
# endif
|
|
|
|
jnz L(last_vec_x1_return)
|
2021-03-05 14:24:52 +00:00
|
|
|
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
# ifdef USE_IN_RTM
|
2021-05-03 07:03:19 +00:00
|
|
|
VPCMP $0, %YMM2, %YMMZERO, %k0
|
|
|
|
kmovd %k0, %eax
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
# else
|
|
|
|
vpmovmskb %ymm2, %eax
|
|
|
|
# endif
|
2021-05-03 07:03:19 +00:00
|
|
|
testl %eax, %eax
|
|
|
|
jnz L(last_vec_x2_return)
|
2021-03-05 14:24:52 +00:00
|
|
|
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
# ifdef USE_IN_RTM
|
2021-05-03 07:03:19 +00:00
|
|
|
kmovd %k2, %eax
|
2021-03-05 14:24:52 +00:00
|
|
|
testl %eax, %eax
|
2021-05-03 07:03:19 +00:00
|
|
|
jnz L(last_vec_x3_return)
|
2021-03-05 14:24:52 +00:00
|
|
|
|
2021-05-03 07:03:19 +00:00
|
|
|
kmovd %k3, %eax
|
2021-03-05 14:24:52 +00:00
|
|
|
tzcntl %eax, %eax
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
leaq (VEC_SIZE * 3 + RET_OFFSET)(%rdi, %rax, CHAR_SIZE), %rax
|
2021-05-03 07:03:19 +00:00
|
|
|
# else
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
vpmovmskb %ymm3, %eax
|
|
|
|
/* Combine matches in VEC3 (eax) with matches in VEC4 (ecx). */
|
|
|
|
salq $VEC_SIZE, %rcx
|
|
|
|
orq %rcx, %rax
|
|
|
|
tzcntq %rax, %rax
|
|
|
|
leaq (VEC_SIZE * 2 + RET_OFFSET)(%rdi, %rax), %rax
|
|
|
|
VZEROUPPER
|
2021-03-05 14:24:52 +00:00
|
|
|
# endif
|
|
|
|
ret
|
|
|
|
|
|
|
|
.p2align 4
|
2021-05-03 07:03:19 +00:00
|
|
|
L(last_vec_x1_return):
|
2021-03-05 14:24:52 +00:00
|
|
|
tzcntl %eax, %eax
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
# if defined USE_AS_WMEMCHR || RET_OFFSET != 0
|
2021-05-03 07:03:19 +00:00
|
|
|
/* NB: Multiply bytes by CHAR_SIZE to get the wchar_t count. */
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
leaq RET_OFFSET(%rdi, %rax, CHAR_SIZE), %rax
|
2021-05-03 07:03:19 +00:00
|
|
|
# else
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
addq %rdi, %rax
|
2021-03-05 14:24:52 +00:00
|
|
|
# endif
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
VZEROUPPER
|
2021-03-05 14:24:52 +00:00
|
|
|
ret
|
|
|
|
|
|
|
|
.p2align 4
|
2021-05-03 07:03:19 +00:00
|
|
|
L(last_vec_x2_return):
|
2021-03-05 14:24:52 +00:00
|
|
|
tzcntl %eax, %eax
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
/* NB: Multiply bytes by RET_SCALE to get the wchar_t count
|
|
|
|
if relevant (RET_SCALE = CHAR_SIZE if USE_AS_WMEMCHAR and
|
|
|
|
USE_IN_RTM are both defined. Otherwise RET_SCALE = 1. */
|
|
|
|
leaq (VEC_SIZE + RET_OFFSET)(%rdi, %rax, RET_SCALE), %rax
|
|
|
|
VZEROUPPER
|
2021-03-05 14:24:52 +00:00
|
|
|
ret
|
|
|
|
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
# ifdef USE_IN_RTM
|
2021-03-05 14:24:52 +00:00
|
|
|
.p2align 4
|
2021-05-03 07:03:19 +00:00
|
|
|
L(last_vec_x3_return):
|
2021-03-05 14:24:52 +00:00
|
|
|
tzcntl %eax, %eax
|
2021-05-03 07:03:19 +00:00
|
|
|
/* NB: Multiply bytes by CHAR_SIZE to get the wchar_t count. */
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
leaq (VEC_SIZE * 2 + RET_OFFSET)(%rdi, %rax, CHAR_SIZE), %rax
|
2021-03-05 14:24:52 +00:00
|
|
|
ret
|
x86: Add EVEX optimized memchr family not safe for RTM
No bug.
This commit adds a new implementation for EVEX memchr that is not safe
for RTM because it uses vzeroupper. The benefit is that by using
ymm0-ymm15 it can use vpcmpeq and vpternlogd in the 4x loop which is
faster than the RTM safe version which cannot use vpcmpeq because
there is no EVEX encoding for the instruction. All parts of the
implementation aside from the 4x loop are the same for the two
versions and the optimization is only relevant for large sizes.
Tigerlake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 9.2 , 9.04 , no-RTM , 0.16
512 , 7 , 224 , 9.19 , 8.98 , no-RTM , 0.21
2048 , 0 , 256 , 10.74 , 10.54 , no-RTM , 0.2
2048 , 0 , 512 , 14.81 , 14.87 , RTM , 0.06
2048 , 0 , 1024 , 22.97 , 22.57 , no-RTM , 0.4
2048 , 0 , 2048 , 37.49 , 34.51 , no-RTM , 2.98 <--
Icelake:
size , algn , Pos , Cur T , New T , Win , Dif
512 , 6 , 192 , 7.6 , 7.3 , no-RTM , 0.3
512 , 7 , 224 , 7.63 , 7.27 , no-RTM , 0.36
2048 , 0 , 256 , 8.48 , 8.38 , no-RTM , 0.1
2048 , 0 , 512 , 11.57 , 11.42 , no-RTM , 0.15
2048 , 0 , 1024 , 17.92 , 17.38 , no-RTM , 0.54
2048 , 0 , 2048 , 30.37 , 27.34 , no-RTM , 3.03 <--
test-memchr, test-wmemchr, and test-rawmemchr are all passing.
Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
Reviewed-by: H.J. Lu <hjl.tools@gmail.com>
2021-05-04 23:02:40 +00:00
|
|
|
# endif
|
2021-05-03 07:03:19 +00:00
|
|
|
|
|
|
|
# ifndef USE_AS_RAWMEMCHR
|
|
|
|
L(last_4x_vec_or_less_cmpeq):
|
|
|
|
VPCMP $0, (VEC_SIZE * 5)(%rdi), %YMMMATCH, %k0
|
|
|
|
kmovd %k0, %eax
|
|
|
|
subq $-(VEC_SIZE * 4), %rdi
|
|
|
|
/* Check first VEC regardless. */
|
|
|
|
testl %eax, %eax
|
|
|
|
jnz L(first_vec_x1_check)
|
|
|
|
|
|
|
|
/* If remaining length <= CHAR_PER_VEC * 2. */
|
|
|
|
addl $(CHAR_PER_VEC * 2), %edx
|
|
|
|
jle L(last_2x_vec)
|
|
|
|
|
2021-03-05 14:24:52 +00:00
|
|
|
.p2align 4
|
2021-05-03 07:03:19 +00:00
|
|
|
L(last_4x_vec):
|
|
|
|
VPCMP $0, (VEC_SIZE * 2)(%rdi), %YMMMATCH, %k0
|
|
|
|
kmovd %k0, %eax
|
|
|
|
testl %eax, %eax
|
|
|
|
jnz L(last_vec_x2)
|
|
|
|
|
|
|
|
|
|
|
|
VPCMP $0, (VEC_SIZE * 3)(%rdi), %YMMMATCH, %k0
|
|
|
|
kmovd %k0, %eax
|
|
|
|
/* Create mask for possible matches within remaining length. */
|
|
|
|
# ifdef USE_AS_WMEMCHR
|
|
|
|
movl $((1 << (CHAR_PER_VEC * 2)) - 1), %ecx
|
|
|
|
bzhil %edx, %ecx, %ecx
|
|
|
|
# else
|
|
|
|
movq $-1, %rcx
|
|
|
|
bzhiq %rdx, %rcx, %rcx
|
|
|
|
# endif
|
|
|
|
/* Test matches in data against length match. */
|
|
|
|
andl %ecx, %eax
|
|
|
|
jnz L(last_vec_x3)
|
|
|
|
|
|
|
|
/* if remaining length <= CHAR_PER_VEC * 3 (Note this is after
|
|
|
|
remaining length was found to be > CHAR_PER_VEC * 2. */
|
|
|
|
subl $CHAR_PER_VEC, %edx
|
|
|
|
jbe L(zero_end2)
|
|
|
|
|
|
|
|
|
|
|
|
VPCMP $0, (VEC_SIZE * 4)(%rdi), %YMMMATCH, %k0
|
|
|
|
kmovd %k0, %eax
|
|
|
|
/* Shift remaining length mask for last VEC. */
|
|
|
|
# ifdef USE_AS_WMEMCHR
|
|
|
|
shrl $CHAR_PER_VEC, %ecx
|
|
|
|
# else
|
|
|
|
shrq $CHAR_PER_VEC, %rcx
|
|
|
|
# endif
|
|
|
|
andl %ecx, %eax
|
|
|
|
jz L(zero_end2)
|
2021-03-05 14:24:52 +00:00
|
|
|
tzcntl %eax, %eax
|
2021-05-03 07:03:19 +00:00
|
|
|
leaq (VEC_SIZE * 4)(%rdi, %rax, CHAR_SIZE), %rax
|
|
|
|
L(zero_end2):
|
2021-03-05 14:24:52 +00:00
|
|
|
ret
|
|
|
|
|
2021-05-03 07:03:19 +00:00
|
|
|
L(last_vec_x2):
|
2021-03-05 14:24:52 +00:00
|
|
|
tzcntl %eax, %eax
|
2021-05-03 07:03:19 +00:00
|
|
|
leaq (VEC_SIZE * 2)(%rdi, %rax, CHAR_SIZE), %rax
|
2021-03-05 14:24:52 +00:00
|
|
|
ret
|
|
|
|
|
|
|
|
.p2align 4
|
2021-05-03 07:03:19 +00:00
|
|
|
L(last_vec_x3):
|
2021-03-05 14:24:52 +00:00
|
|
|
tzcntl %eax, %eax
|
2021-05-03 07:03:19 +00:00
|
|
|
leaq (VEC_SIZE * 3)(%rdi, %rax, CHAR_SIZE), %rax
|
2021-03-05 14:24:52 +00:00
|
|
|
ret
|
2021-05-03 07:03:19 +00:00
|
|
|
# endif
|
2021-03-05 14:24:52 +00:00
|
|
|
|
|
|
|
END (MEMCHR)
|
|
|
|
#endif
|