AuroraMiddleware/glibc
1
0
Fork 0
mirror of https://sourceware.org/git/glibc.git synced 2024-11-14 09:01:07 +00:00
Code Issues Packages Projects Releases Wiki Activity

x86: Optimize strchr-avx2.S

Browse Source 
No bug. This commit optimizes strchr-avx2.S. The optimizations are all
small things such as save an ALU in the alignment process, saving a
few instructions in the loop return, saving some bytes in the main
loop, and increasing the ILP in the return cases. test-strchr,
test-strchrnul, test-wcschr, and test-wcschrnul are all passing.

Signed-off-by: Noah Goldstein <goldstein.w.n@gmail.com>
This commit is contained in:
Noah Goldstein 2021-04-23 15:56:24 -04:00 committed by H.J. Lu
parent 24f261f27f
commit ccabe7971f
1 changed files with 172 additions and 120 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

292
sysdeps/x86_64/multiarch/strchr-avx2.S
View File

@ -49,132 +49,144 @@
.section SECTION(.text),"ax",@progbits .section SECTION(.text),"ax",@progbits
ENTRY (STRCHR) ENTRY (STRCHR)
movl %edi, %ecx
# ifndef USE_AS_STRCHRNUL
xorl %edx, %edx
# endif
/* Broadcast CHAR to YMM0. */ /* Broadcast CHAR to YMM0. */
vmovd %esi, %xmm0 vmovd %esi, %xmm0
movl %edi, %eax
andl $(PAGE_SIZE - 1), %eax
VPBROADCAST %xmm0, %ymm0
vpxor %xmm9, %xmm9, %xmm9 vpxor %xmm9, %xmm9, %xmm9
VPBROADCAST %xmm0, %ymm0
/* Check if we cross page boundary with one vector load. */ /* Check if we cross page boundary with one vector load. */
andl $(PAGE_SIZE - 1), %ecx cmpl $(PAGE_SIZE - VEC_SIZE), %eax
cmpl $(PAGE_SIZE - VEC_SIZE), %ecx ja L(cross_page_boundary)
ja L(cross_page_boundary)
/* Check the first VEC_SIZE bytes. Search for both CHAR and the /* Check the first VEC_SIZE bytes. Search for both CHAR and the
null byte. */ null byte. */
vmovdqu (%rdi), %ymm8 vmovdqu (%rdi), %ymm8
VPCMPEQ %ymm8, %ymm0, %ymm1 VPCMPEQ %ymm8, %ymm0, %ymm1
VPCMPEQ %ymm8, %ymm9, %ymm2 VPCMPEQ %ymm8, %ymm9, %ymm2
vpor %ymm1, %ymm2, %ymm1 vpor %ymm1, %ymm2, %ymm1
vpmovmskb %ymm1, %eax vpmovmskb %ymm1, %eax
testl %eax, %eax testl %eax, %eax
jz L(more_vecs) jz L(aligned_more)
tzcntl %eax, %eax tzcntl %eax, %eax
# ifndef USE_AS_STRCHRNUL
/* Found CHAR or the null byte. */ /* Found CHAR or the null byte. */
cmp (%rdi, %rax), %CHAR_REG
jne L(zero)
# endif
addq %rdi, %rax addq %rdi, %rax
# ifndef USE_AS_STRCHRNUL
cmp (%rax), %CHAR_REG
cmovne %rdx, %rax
# endif
L(return_vzeroupper):
ZERO_UPPER_VEC_REGISTERS_RETURN
.p2align 4
L(more_vecs):
/* Align data for aligned loads in the loop. */
andq $-VEC_SIZE, %rdi
L(aligned_more):
/* Check the next 4 * VEC_SIZE. Only one VEC_SIZE at a time
since data is only aligned to VEC_SIZE. */
vmovdqa VEC_SIZE(%rdi), %ymm8
addq $VEC_SIZE, %rdi
VPCMPEQ %ymm8, %ymm0, %ymm1
VPCMPEQ %ymm8, %ymm9, %ymm2
vpor %ymm1, %ymm2, %ymm1
vpmovmskb %ymm1, %eax
testl %eax, %eax
jnz L(first_vec_x0)
vmovdqa VEC_SIZE(%rdi), %ymm8
VPCMPEQ %ymm8, %ymm0, %ymm1
VPCMPEQ %ymm8, %ymm9, %ymm2
vpor %ymm1, %ymm2, %ymm1
vpmovmskb %ymm1, %eax
testl %eax, %eax
jnz L(first_vec_x1)
vmovdqa (VEC_SIZE * 2)(%rdi), %ymm8
VPCMPEQ %ymm8, %ymm0, %ymm1
VPCMPEQ %ymm8, %ymm9, %ymm2
vpor %ymm1, %ymm2, %ymm1
vpmovmskb %ymm1, %eax
testl %eax, %eax
jnz L(first_vec_x2)
vmovdqa (VEC_SIZE * 3)(%rdi), %ymm8
VPCMPEQ %ymm8, %ymm0, %ymm1
VPCMPEQ %ymm8, %ymm9, %ymm2
vpor %ymm1, %ymm2, %ymm1
vpmovmskb %ymm1, %eax
testl %eax, %eax
jz L(prep_loop_4x)
tzcntl %eax, %eax
leaq (VEC_SIZE * 3)(%rdi, %rax), %rax
# ifndef USE_AS_STRCHRNUL
cmp (%rax), %CHAR_REG
cmovne %rdx, %rax
# endif
VZEROUPPER_RETURN VZEROUPPER_RETURN
.p2align 4 /* .p2align 5 helps keep performance more consistent if ENTRY()
L(first_vec_x0): alignment % 32 was either 16 or 0. As well this makes the
alignment % 32 of the loop_4x_vec fixed which makes tuning it
easier. */
.p2align 5
L(first_vec_x4):
tzcntl %eax, %eax tzcntl %eax, %eax
/* Found CHAR or the null byte. */ addq $(VEC_SIZE * 3 + 1), %rdi
addq %rdi, %rax
# ifndef USE_AS_STRCHRNUL # ifndef USE_AS_STRCHRNUL
cmp (%rax), %CHAR_REG /* Found CHAR or the null byte. */
cmovne %rdx, %rax cmp (%rdi, %rax), %CHAR_REG
jne L(zero)
# endif # endif
addq %rdi, %rax
VZEROUPPER_RETURN VZEROUPPER_RETURN
# ifndef USE_AS_STRCHRNUL
L(zero):
xorl %eax, %eax
VZEROUPPER_RETURN
# endif
.p2align 4 .p2align 4
L(first_vec_x1): L(first_vec_x1):
tzcntl %eax, %eax tzcntl %eax, %eax
leaq VEC_SIZE(%rdi, %rax), %rax incq %rdi
# ifndef USE_AS_STRCHRNUL # ifndef USE_AS_STRCHRNUL
cmp (%rax), %CHAR_REG /* Found CHAR or the null byte. */
cmovne %rdx, %rax cmp (%rdi, %rax), %CHAR_REG
jne L(zero)
# endif # endif
addq %rdi, %rax
VZEROUPPER_RETURN VZEROUPPER_RETURN
.p2align 4 .p2align 4
L(first_vec_x2): L(first_vec_x2):
tzcntl %eax, %eax tzcntl %eax, %eax
/* Found CHAR or the null byte. */ addq $(VEC_SIZE + 1), %rdi
leaq (VEC_SIZE * 2)(%rdi, %rax), %rax
# ifndef USE_AS_STRCHRNUL # ifndef USE_AS_STRCHRNUL
cmp (%rax), %CHAR_REG /* Found CHAR or the null byte. */
cmovne %rdx, %rax cmp (%rdi, %rax), %CHAR_REG
jne L(zero)
# endif # endif
addq %rdi, %rax
VZEROUPPER_RETURN VZEROUPPER_RETURN
L(prep_loop_4x): .p2align 4
/* Align data to 4 * VEC_SIZE. */ L(first_vec_x3):
andq $-(VEC_SIZE * 4), %rdi tzcntl %eax, %eax
addq $(VEC_SIZE * 2 + 1), %rdi
# ifndef USE_AS_STRCHRNUL
/* Found CHAR or the null byte. */
cmp (%rdi, %rax), %CHAR_REG
jne L(zero)
# endif
addq %rdi, %rax
VZEROUPPER_RETURN
.p2align 4
L(aligned_more):
/* Align data to VEC_SIZE - 1. This is the same number of
instructions as using andq -VEC_SIZE but saves 4 bytes of code
on x4 check. */
orq $(VEC_SIZE - 1), %rdi
L(cross_page_continue):
/* Check the next 4 * VEC_SIZE. Only one VEC_SIZE at a time
since data is only aligned to VEC_SIZE. */
vmovdqa 1(%rdi), %ymm8
VPCMPEQ %ymm8, %ymm0, %ymm1
VPCMPEQ %ymm8, %ymm9, %ymm2
vpor %ymm1, %ymm2, %ymm1
vpmovmskb %ymm1, %eax
testl %eax, %eax
jnz L(first_vec_x1)
vmovdqa (VEC_SIZE + 1)(%rdi), %ymm8
VPCMPEQ %ymm8, %ymm0, %ymm1
VPCMPEQ %ymm8, %ymm9, %ymm2
vpor %ymm1, %ymm2, %ymm1
vpmovmskb %ymm1, %eax
testl %eax, %eax
jnz L(first_vec_x2)
vmovdqa (VEC_SIZE * 2 + 1)(%rdi), %ymm8
VPCMPEQ %ymm8, %ymm0, %ymm1
VPCMPEQ %ymm8, %ymm9, %ymm2
vpor %ymm1, %ymm2, %ymm1
vpmovmskb %ymm1, %eax
testl %eax, %eax
jnz L(first_vec_x3)
vmovdqa (VEC_SIZE * 3 + 1)(%rdi), %ymm8
VPCMPEQ %ymm8, %ymm0, %ymm1
VPCMPEQ %ymm8, %ymm9, %ymm2
vpor %ymm1, %ymm2, %ymm1
vpmovmskb %ymm1, %eax
testl %eax, %eax
jnz L(first_vec_x4)
/* Align data to VEC_SIZE * 4 - 1. */
addq $(VEC_SIZE * 4 + 1), %rdi
andq $-(VEC_SIZE * 4), %rdi
.p2align 4 .p2align 4
L(loop_4x_vec): L(loop_4x_vec):
/* Compare 4 * VEC at a time forward. */ /* Compare 4 * VEC at a time forward. */
vmovdqa (VEC_SIZE * 4)(%rdi), %ymm5 vmovdqa (%rdi), %ymm5
vmovdqa (VEC_SIZE * 5)(%rdi), %ymm6 vmovdqa (VEC_SIZE)(%rdi), %ymm6
vmovdqa (VEC_SIZE * 6)(%rdi), %ymm7 vmovdqa (VEC_SIZE * 2)(%rdi), %ymm7
vmovdqa (VEC_SIZE * 7)(%rdi), %ymm8 vmovdqa (VEC_SIZE * 3)(%rdi), %ymm8
/* Leaves only CHARS matching esi as 0. */ /* Leaves only CHARS matching esi as 0. */
vpxor %ymm5, %ymm0, %ymm1 vpxor %ymm5, %ymm0, %ymm1
@ -190,62 +202,102 @@ L(loop_4x_vec):
VPMINU %ymm1, %ymm2, %ymm5 VPMINU %ymm1, %ymm2, %ymm5
VPMINU %ymm3, %ymm4, %ymm6 VPMINU %ymm3, %ymm4, %ymm6
VPMINU %ymm5, %ymm6, %ymm5 VPMINU %ymm5, %ymm6, %ymm6
VPCMPEQ %ymm5, %ymm9, %ymm5 VPCMPEQ %ymm6, %ymm9, %ymm6
vpmovmskb %ymm5, %eax vpmovmskb %ymm6, %ecx
subq $-(VEC_SIZE * 4), %rdi
testl %ecx, %ecx
jz L(loop_4x_vec)
addq $(VEC_SIZE * 4), %rdi
testl %eax, %eax
jz L(loop_4x_vec)
VPCMPEQ %ymm1, %ymm9, %ymm1 VPCMPEQ %ymm1, %ymm9, %ymm1
vpmovmskb %ymm1, %eax vpmovmskb %ymm1, %eax
testl %eax, %eax testl %eax, %eax
jnz L(first_vec_x0) jnz L(last_vec_x0)
VPCMPEQ %ymm2, %ymm9, %ymm2
VPCMPEQ %ymm5, %ymm9, %ymm2
vpmovmskb %ymm2, %eax vpmovmskb %ymm2, %eax
testl %eax, %eax testl %eax, %eax
jnz L(first_vec_x1) jnz L(last_vec_x1)
VPCMPEQ %ymm3, %ymm9, %ymm3 VPCMPEQ %ymm3, %ymm9, %ymm3
VPCMPEQ %ymm4, %ymm9, %ymm4 vpmovmskb %ymm3, %eax
vpmovmskb %ymm3, %ecx /* rcx has combined result from all 4 VEC. It will only be used
vpmovmskb %ymm4, %eax if the first 3 other VEC all did not contain a match. */
salq $32, %rax salq $32, %rcx
orq %rcx, %rax orq %rcx, %rax
tzcntq %rax, %rax tzcntq %rax, %rax
leaq (VEC_SIZE * 2)(%rdi, %rax), %rax subq $(VEC_SIZE * 2), %rdi
# ifndef USE_AS_STRCHRNUL # ifndef USE_AS_STRCHRNUL
cmp (%rax), %CHAR_REG /* Found CHAR or the null byte. */
cmovne %rdx, %rax cmp (%rdi, %rax), %CHAR_REG
jne L(zero_end)
# endif # endif
addq %rdi, %rax
VZEROUPPER_RETURN VZEROUPPER_RETURN
.p2align 4
L(last_vec_x0):
tzcntl %eax, %eax
addq $-(VEC_SIZE * 4), %rdi
# ifndef USE_AS_STRCHRNUL
/* Found CHAR or the null byte. */
cmp (%rdi, %rax), %CHAR_REG
jne L(zero_end)
# endif
addq %rdi, %rax
VZEROUPPER_RETURN
# ifndef USE_AS_STRCHRNUL
L(zero_end):
xorl %eax, %eax
VZEROUPPER_RETURN
# endif
.p2align 4
L(last_vec_x1):
tzcntl %eax, %eax
subq $(VEC_SIZE * 3), %rdi
# ifndef USE_AS_STRCHRNUL
/* Found CHAR or the null byte. */
cmp (%rdi, %rax), %CHAR_REG
jne L(zero_end)
# endif
addq %rdi, %rax
VZEROUPPER_RETURN
/* Cold case for crossing page with first load. */ /* Cold case for crossing page with first load. */
.p2align 4 .p2align 4
L(cross_page_boundary): L(cross_page_boundary):
andq $-VEC_SIZE, %rdi movq %rdi, %rdx
andl $(VEC_SIZE - 1), %ecx /* Align rdi to VEC_SIZE - 1. */
orq $(VEC_SIZE - 1), %rdi
vmovdqa (%rdi), %ymm8 vmovdqa -(VEC_SIZE - 1)(%rdi), %ymm8
VPCMPEQ %ymm8, %ymm0, %ymm1 VPCMPEQ %ymm8, %ymm0, %ymm1
VPCMPEQ %ymm8, %ymm9, %ymm2 VPCMPEQ %ymm8, %ymm9, %ymm2
vpor %ymm1, %ymm2, %ymm1 vpor %ymm1, %ymm2, %ymm1
vpmovmskb %ymm1, %eax vpmovmskb %ymm1, %eax
/* Remove the leading bits. */ /* Remove the leading bytes. sarxl only uses bits [5:0] of COUNT
sarxl %ecx, %eax, %eax so no need to manually mod edx. */
sarxl %edx, %eax, %eax
testl %eax, %eax testl %eax, %eax
jz L(aligned_more) jz L(cross_page_continue)
tzcntl %eax, %eax tzcntl %eax, %eax
addq %rcx, %rdi
addq %rdi, %rax
# ifndef USE_AS_STRCHRNUL # ifndef USE_AS_STRCHRNUL
cmp (%rax), %CHAR_REG xorl %ecx, %ecx
cmovne %rdx, %rax /* Found CHAR or the null byte. */
cmp (%rdx, %rax), %CHAR_REG
leaq (%rdx, %rax), %rax
cmovne %rcx, %rax
# else
addq %rdx, %rax
# endif # endif
VZEROUPPER_RETURN L(return_vzeroupper):
ZERO_UPPER_VEC_REGISTERS_RETURN
END (STRCHR) END (STRCHR)
# endif # endif