diff --git a/libc/arch-x86_64/string/cache.h b/libc/arch-x86_64/string/cache.h
new file mode 100644
index 0000000..38acc6e
--- /dev/null
+++ b/libc/arch-x86_64/string/cache.h
@@ -0,0 +1,36 @@
+/*
+Copyright (c) 2014, Intel Corporation
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+    * Redistributions of source code must retain the above copyright notice,
+    * this list of conditions and the following disclaimer.
+
+    * Redistributions in binary form must reproduce the above copyright notice,
+    * this list of conditions and the following disclaimer in the documentation
+    * and/or other materials provided with the distribution.
+
+    * Neither the name of Intel Corporation nor the names of its contributors
+    * may be used to endorse or promote products derived from this software
+    * without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+/* Values are optimized for Silvermont */
+#define SHARED_CACHE_SIZE	(1024*1024)			/* Silvermont L2 Cache */
+#define DATA_CACHE_SIZE		(24*1024)			/* Silvermont L1 Data Cache */
+
+#define SHARED_CACHE_SIZE_HALF	(SHARED_CACHE_SIZE / 2)
+#define DATA_CACHE_SIZE_HALF	(DATA_CACHE_SIZE / 2)
diff --git a/libc/arch-x86_64/string/sse2-bcopy-slm.S b/libc/arch-x86_64/string/sse2-bcopy-slm.S
new file mode 100644
index 0000000..effab0e
--- /dev/null
+++ b/libc/arch-x86_64/string/sse2-bcopy-slm.S
@@ -0,0 +1,33 @@
+/*
+Copyright (c) 2014, Intel Corporation
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+    * Redistributions of source code must retain the above copyright notice,
+    * this list of conditions and the following disclaimer.
+
+    * Redistributions in binary form must reproduce the above copyright notice,
+    * this list of conditions and the following disclaimer in the documentation
+    * and/or other materials provided with the distribution.
+
+    * Neither the name of Intel Corporation nor the names of its contributors
+    * may be used to endorse or promote products derived from this software
+    * without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#define USE_AS_BCOPY
+#define MEMMOVE		bcopy
+#include "sse2-memmove-slm.S"
diff --git a/libc/arch-x86_64/string/sse2-bzero-slm.S b/libc/arch-x86_64/string/sse2-bzero-slm.S
new file mode 100644
index 0000000..446ea5b
--- /dev/null
+++ b/libc/arch-x86_64/string/sse2-bzero-slm.S
@@ -0,0 +1,33 @@
+/*
+Copyright (c) 2014, Intel Corporation
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+    * Redistributions of source code must retain the above copyright notice,
+    * this list of conditions and the following disclaimer.
+
+    * Redistributions in binary form must reproduce the above copyright notice,
+    * this list of conditions and the following disclaimer in the documentation
+    * and/or other materials provided with the distribution.
+
+    * Neither the name of Intel Corporation nor the names of its contributors
+    * may be used to endorse or promote products derived from this software
+    * without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#define USE_AS_BZERO_P
+#define MEMSET		bzero
+#include "sse2-memset-slm.S"
diff --git a/libc/arch-x86_64/string/sse2-memcpy-slm.S b/libc/arch-x86_64/string/sse2-memcpy-slm.S
new file mode 100644
index 0000000..4c30fb6
--- /dev/null
+++ b/libc/arch-x86_64/string/sse2-memcpy-slm.S
@@ -0,0 +1,299 @@
+/*
+Copyright (c) 2014, Intel Corporation
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+    * Redistributions of source code must retain the above copyright notice,
+    * this list of conditions and the following disclaimer.
+
+    * Redistributions in binary form must reproduce the above copyright notice,
+    * this list of conditions and the following disclaimer in the documentation
+    * and/or other materials provided with the distribution.
+
+    * Neither the name of Intel Corporation nor the names of its contributors
+    * may be used to endorse or promote products derived from this software
+    * without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "cache.h"
+
+#ifndef MEMCPY
+# define MEMCPY		memcpy
+#endif
+
+#ifndef L
+# define L(label)	.L##label
+#endif
+
+#ifndef cfi_startproc
+# define cfi_startproc	.cfi_startproc
+#endif
+
+#ifndef cfi_endproc
+# define cfi_endproc	.cfi_endproc
+#endif
+
+#ifndef cfi_rel_offset
+# define cfi_rel_offset(reg, off)	.cfi_rel_offset reg, off
+#endif
+
+#ifndef cfi_restore
+# define cfi_restore(reg)	.cfi_restore reg
+#endif
+
+#ifndef cfi_adjust_cfa_offset
+# define cfi_adjust_cfa_offset(off)	.cfi_adjust_cfa_offset off
+#endif
+
+#ifndef ENTRY
+# define ENTRY(name)		\
+	.type name,  @function;		\
+	.globl name;		\
+	.p2align 4;		\
+name:		\
+	cfi_startproc
+#endif
+
+#ifndef END
+# define END(name)		\
+	cfi_endproc;		\
+	.size name, .-name
+#endif
+
+#define CFI_PUSH(REG)		\
+	cfi_adjust_cfa_offset (4);		\
+	cfi_rel_offset (REG, 0)
+
+#define CFI_POP(REG)		\
+	cfi_adjust_cfa_offset (-4);		\
+	cfi_restore (REG)
+
+#define PUSH(REG)	push REG;
+#define POP(REG)	pop REG;
+
+#define ENTRANCE	PUSH (%rbx);
+#define RETURN_END	POP (%rbx); ret
+#define RETURN		RETURN_END;
+
+	.section .text.sse2,"ax",@progbits
+ENTRY (MEMCPY)
+	ENTRANCE
+	cmp	%rsi, %rdi
+	je	L(return)
+
+	cmp	$16, %rdx
+	jbe	L(len_0_16_bytes)
+
+	cmp	$SHARED_CACHE_SIZE_HALF, %rdx
+	jae	L(large_page)
+
+	movdqu	(%rsi), %xmm0
+	movdqu	-16(%rsi, %rdx), %xmm1
+	cmp	$32, %rdx
+	movdqu	%xmm0, (%rdi)
+	movdqu	%xmm1, -16(%rdi, %rdx)
+	jbe	L(return)
+
+	movdqu	16(%rsi), %xmm0
+	movdqu	-32(%rsi, %rdx), %xmm1
+	cmp	$64, %rdx
+	movdqu	%xmm0, 16(%rdi)
+	movdqu	%xmm1, -32(%rdi, %rdx)
+	jbe	L(return)
+
+	movdqu	32(%rsi), %xmm0
+	movdqu	48(%rsi), %xmm1
+	movdqu	-48(%rsi, %rdx), %xmm2
+	movdqu	-64(%rsi, %rdx), %xmm3
+	cmp	$128, %rdx
+	movdqu	%xmm0, 32(%rdi)
+	movdqu	%xmm1, 48(%rdi)
+	movdqu	%xmm2, -48(%rdi, %rdx)
+	movdqu	%xmm3, -64(%rdi, %rdx)
+	jbe	L(return)
+
+/* Now the main loop: we align the address of the destination.  */
+	lea	64(%rdi), %r8
+	and	$-64, %r8
+
+	add	%rdi, %rdx
+	and	$-64, %rdx
+
+	sub	%rdi, %rsi
+
+/* We should stop two iterations before the termination
+	(in order not to misprefetch).  */
+	sub	$64, %rdx
+	cmp	%r8, %rdx
+	je	L(main_loop_just_one_iteration)
+
+	sub	$64, %rdx
+	cmp	%r8, %rdx
+	je	L(main_loop_last_two_iterations)
+
+
+	.p2align 4
+L(main_loop_cache):
+
+	prefetcht0 128(%r8, %rsi)
+
+	movdqu	(%r8, %rsi), %xmm0
+	movdqu	16(%r8, %rsi), %xmm1
+	movdqu	32(%r8, %rsi), %xmm2
+	movdqu	48(%r8, %rsi), %xmm3
+	movdqa	%xmm0, (%r8)
+	movdqa	%xmm1, 16(%r8)
+	movdqa	%xmm2, 32(%r8)
+	movdqa	%xmm3, 48(%r8)
+	lea	64(%r8), %r8
+	cmp	%r8, %rdx
+	jne	L(main_loop_cache)
+
+L(main_loop_last_two_iterations):
+	movdqu	(%r8, %rsi), %xmm0
+	movdqu	16(%r8, %rsi), %xmm1
+	movdqu	32(%r8, %rsi), %xmm2
+	movdqu	48(%r8, %rsi), %xmm3
+	movdqu	64(%r8, %rsi), %xmm4
+	movdqu	80(%r8, %rsi), %xmm5
+	movdqu	96(%r8, %rsi), %xmm6
+	movdqu	112(%r8, %rsi), %xmm7
+	movdqa	%xmm0, (%r8)
+	movdqa	%xmm1, 16(%r8)
+	movdqa	%xmm2, 32(%r8)
+	movdqa	%xmm3, 48(%r8)
+	movdqa	%xmm4, 64(%r8)
+	movdqa	%xmm5, 80(%r8)
+	movdqa	%xmm6, 96(%r8)
+	movdqa	%xmm7, 112(%r8)
+	jmp	L(return)
+
+L(main_loop_just_one_iteration):
+	movdqu	(%r8, %rsi), %xmm0
+	movdqu	16(%r8, %rsi), %xmm1
+	movdqu	32(%r8, %rsi), %xmm2
+	movdqu	48(%r8, %rsi), %xmm3
+	movdqa	%xmm0, (%r8)
+	movdqa	%xmm1, 16(%r8)
+	movdqa	%xmm2, 32(%r8)
+	movdqa	%xmm3, 48(%r8)
+	jmp	L(return)
+
+L(large_page):
+	movdqu	(%rsi), %xmm0
+	movdqu	16(%rsi), %xmm1
+	movdqu	32(%rsi), %xmm2
+	movdqu	48(%rsi), %xmm3
+	movdqu	-64(%rsi, %rdx), %xmm4
+	movdqu	-48(%rsi, %rdx), %xmm5
+	movdqu	-32(%rsi, %rdx), %xmm6
+	movdqu	-16(%rsi, %rdx), %xmm7
+	movdqu	%xmm0, (%rdi)
+	movdqu	%xmm1, 16(%rdi)
+	movdqu	%xmm2, 32(%rdi)
+	movdqu	%xmm3, 48(%rdi)
+	movdqu	%xmm4, -64(%rdi, %rdx)
+	movdqu	%xmm5, -48(%rdi, %rdx)
+	movdqu	%xmm6, -32(%rdi, %rdx)
+	movdqu	%xmm7, -16(%rdi, %rdx)
+
+	movdqu	64(%rsi), %xmm0
+	movdqu	80(%rsi), %xmm1
+	movdqu	96(%rsi), %xmm2
+	movdqu	112(%rsi), %xmm3
+	movdqu	-128(%rsi, %rdx), %xmm4
+	movdqu	-112(%rsi, %rdx), %xmm5
+	movdqu	-96(%rsi, %rdx), %xmm6
+	movdqu	-80(%rsi, %rdx), %xmm7
+	movdqu	%xmm0, 64(%rdi)
+	movdqu	%xmm1, 80(%rdi)
+	movdqu	%xmm2, 96(%rdi)
+	movdqu	%xmm3, 112(%rdi)
+	movdqu	%xmm4, -128(%rdi, %rdx)
+	movdqu	%xmm5, -112(%rdi, %rdx)
+	movdqu	%xmm6, -96(%rdi, %rdx)
+	movdqu	%xmm7, -80(%rdi, %rdx)
+
+/* Now the main loop with non temporal stores. We align
+	the address of the destination.  */
+	lea	128(%rdi), %r8
+	and	$-128, %r8
+
+	add	%rdi, %rdx
+	and	$-128, %rdx
+
+	sub	%rdi, %rsi
+
+	.p2align 4
+L(main_loop_large_page):
+	movdqu	(%r8, %rsi), %xmm0
+	movdqu	16(%r8, %rsi), %xmm1
+	movdqu	32(%r8, %rsi), %xmm2
+	movdqu	48(%r8, %rsi), %xmm3
+	movdqu	64(%r8, %rsi), %xmm4
+	movdqu	80(%r8, %rsi), %xmm5
+	movdqu	96(%r8, %rsi), %xmm6
+	movdqu	112(%r8, %rsi), %xmm7
+	movntdq	%xmm0, (%r8)
+	movntdq	%xmm1, 16(%r8)
+	movntdq	%xmm2, 32(%r8)
+	movntdq	%xmm3, 48(%r8)
+	movntdq	%xmm4, 64(%r8)
+	movntdq	%xmm5, 80(%r8)
+	movntdq	%xmm6, 96(%r8)
+	movntdq	%xmm7, 112(%r8)
+	lea	128(%r8), %r8
+	cmp	%r8, %rdx
+	jne	L(main_loop_large_page)
+	sfence
+	jmp	L(return)
+
+L(len_0_16_bytes):
+	testb	$24, %dl
+	jne	L(len_9_16_bytes)
+	testb	$4, %dl
+	.p2align 4,,5
+	jne	L(len_5_8_bytes)
+	test	%rdx, %rdx
+	.p2align 4,,2
+	je	L(return)
+	movzbl	(%rsi), %ebx
+	testb	$2, %dl
+	movb	%bl, (%rdi)
+	je	L(return)
+	movzwl	-2(%rsi,%rdx), %ebx
+	movw	%bx, -2(%rdi,%rdx)
+	jmp	L(return)
+
+L(len_9_16_bytes):
+	movq	(%rsi), %xmm0
+	movq	-8(%rsi, %rdx), %xmm1
+	movq	%xmm0, (%rdi)
+	movq	%xmm1, -8(%rdi, %rdx)
+	jmp	L(return)
+
+L(len_5_8_bytes):
+	movl	(%rsi), %ebx
+	movl	%ebx, (%rdi)
+	movl	-4(%rsi,%rdx), %ebx
+	movl	%ebx, -4(%rdi,%rdx)
+	jmp	L(return)
+
+L(return):
+	mov 	%rdi, %rax
+	RETURN
+
+END (MEMCPY)
diff --git a/libc/arch-x86_64/string/sse2-memmove-slm.S b/libc/arch-x86_64/string/sse2-memmove-slm.S
new file mode 100644
index 0000000..ee8440e
--- /dev/null
+++ b/libc/arch-x86_64/string/sse2-memmove-slm.S
@@ -0,0 +1,635 @@
+/*
+Copyright (c) 2014, Intel Corporation
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+    * Redistributions of source code must retain the above copyright notice,
+    * this list of conditions and the following disclaimer.
+
+    * Redistributions in binary form must reproduce the above copyright notice,
+    * this list of conditions and the following disclaimer in the documentation
+    * and/or other materials provided with the distribution.
+
+    * Neither the name of Intel Corporation nor the names of its contributors
+    * may be used to endorse or promote products derived from this software
+    * without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "cache.h"
+
+#ifndef MEMMOVE
+# define MEMMOVE		memmove
+#endif
+
+#ifndef L
+# define L(label)	.L##label
+#endif
+
+#ifndef cfi_startproc
+# define cfi_startproc	.cfi_startproc
+#endif
+
+#ifndef cfi_endproc
+# define cfi_endproc	.cfi_endproc
+#endif
+
+#ifndef cfi_rel_offset
+# define cfi_rel_offset(reg, off)	.cfi_rel_offset reg, off
+#endif
+
+#ifndef cfi_restore
+# define cfi_restore(reg)	.cfi_restore reg
+#endif
+
+#ifndef cfi_adjust_cfa_offset
+# define cfi_adjust_cfa_offset(off)	.cfi_adjust_cfa_offset off
+#endif
+
+#ifndef ENTRY
+# define ENTRY(name)		\
+	.type name,  @function;		\
+	.globl name;		\
+	.p2align 4;		\
+name:		\
+	cfi_startproc
+#endif
+
+#ifndef END
+# define END(name)		\
+	cfi_endproc;		\
+	.size name, .-name
+#endif
+
+#define CFI_PUSH(REG)		\
+	cfi_adjust_cfa_offset (4);		\
+	cfi_rel_offset (REG, 0)
+
+#define CFI_POP(REG)		\
+	cfi_adjust_cfa_offset (-4);		\
+	cfi_restore (REG)
+
+#define PUSH(REG)	push REG;
+#define POP(REG)	pop REG;
+
+#define ENTRANCE	PUSH (%rbx);
+#define RETURN_END	POP (%rbx); ret
+#define RETURN		RETURN_END;
+
+	.section .text.sse2,"ax",@progbits
+ENTRY (MEMMOVE)
+	ENTRANCE
+#ifdef USE_AS_BCOPY
+	xchg	%rsi, %rdi
+#endif
+	mov	%rdi, %rax
+
+/* Check whether we should copy backward or forward.  */
+	cmp	%rsi, %rdi
+	je	L(mm_return)
+	ja	L(mm_len_0_or_more_backward)
+
+/* Now do checks for lengths. We do [0..16], [0..32], [0..64], [0..128]
+	separately.  */
+	cmp	$16, %rdx
+	jbe	L(mm_len_0_16_bytes_forward)
+
+	cmp	$32, %rdx
+	jg	L(mm_len_32_or_more_forward)
+
+/* Copy [0..32] and return.  */
+	movdqu	(%rsi), %xmm0
+	movdqu	-16(%rsi, %rdx), %xmm1
+	movdqu	%xmm0, (%rdi)
+	movdqu	%xmm1, -16(%rdi, %rdx)
+	jmp	L(mm_return)
+
+L(mm_len_32_or_more_forward):
+	cmp	$64, %rdx
+	jg	L(mm_len_64_or_more_forward)
+
+/* Copy [0..64] and return.  */
+	movdqu	(%rsi), %xmm0
+	movdqu	16(%rsi), %xmm1
+	movdqu	-16(%rsi, %rdx), %xmm2
+	movdqu	-32(%rsi, %rdx), %xmm3
+	movdqu	%xmm0, (%rdi)
+	movdqu	%xmm1, 16(%rdi)
+	movdqu	%xmm2, -16(%rdi, %rdx)
+	movdqu	%xmm3, -32(%rdi, %rdx)
+	jmp	L(mm_return)
+
+L(mm_len_64_or_more_forward):
+	cmp	$128, %rdx
+	jg	L(mm_len_128_or_more_forward)
+
+/* Copy [0..128] and return.  */
+	movdqu	(%rsi), %xmm0
+	movdqu	16(%rsi), %xmm1
+	movdqu	32(%rsi), %xmm2
+	movdqu	48(%rsi), %xmm3
+	movdqu	-64(%rsi, %rdx), %xmm4
+	movdqu	-48(%rsi, %rdx), %xmm5
+	movdqu	-32(%rsi, %rdx), %xmm6
+	movdqu	-16(%rsi, %rdx), %xmm7
+	movdqu	%xmm0, (%rdi)
+	movdqu	%xmm1, 16(%rdi)
+	movdqu	%xmm2, 32(%rdi)
+	movdqu	%xmm3, 48(%rdi)
+	movdqu	%xmm4, -64(%rdi, %rdx)
+	movdqu	%xmm5, -48(%rdi, %rdx)
+	movdqu	%xmm6, -32(%rdi, %rdx)
+	movdqu	%xmm7, -16(%rdi, %rdx)
+	jmp	L(mm_return)
+
+L(mm_len_128_or_more_forward):
+
+	cmp	$SHARED_CACHE_SIZE_HALF, %rdx
+	jae	L(mm_large_page_forward)
+
+	mov	%rsi, %r8  // copy src to r8
+	mov	%rdi, %r9  // copy dst to r9
+
+/* Aligning the address of destination.  */
+/*  save first unaligned 64 bytes */
+	movdqu	(%rsi), %xmm0
+	movdqu	16(%rsi), %xmm1
+	movdqu	32(%rsi), %xmm2
+	movdqu	48(%rsi), %xmm3
+
+	lea	64(%r9), %rdi
+	and	$-64, %rdi  /* rdi now aligned to next 64 byte boundary */
+
+	sub	%r9, %rsi /* rsi = src - dst = diff */
+
+	movdqu	(%rdi, %rsi), %xmm4
+	movdqu	16(%rdi, %rsi), %xmm5
+	movdqu	32(%rdi, %rsi), %xmm6
+	movdqu	48(%rdi, %rsi), %xmm7
+
+	movdqu	%xmm0, (%r9)
+	movdqu	%xmm1, 16(%r9)
+	movdqu	%xmm2, 32(%r9)
+	movdqu	%xmm3, 48(%r9)
+	movdqa	%xmm4, (%rdi)
+	movdqa	%xmm5, 16(%rdi)
+	movdqa	%xmm6, 32(%rdi)
+	movdqa	%xmm7, 48(%rdi)
+	add	$64, %rdi
+
+	lea	(%r9, %rdx), %rbx
+	and	$-64, %rbx
+
+	cmp	%rdi, %rbx
+	jbe	L(mm_copy_remaining_forward)
+
+	.p2align 4
+L(mm_main_loop_forward):
+
+	prefetcht0 128(%rdi, %rsi)
+
+	movdqu	(%rdi, %rsi), %xmm0
+	movdqu	16(%rdi, %rsi), %xmm1
+	movdqu	32(%rdi, %rsi), %xmm2
+	movdqu	48(%rdi, %rsi), %xmm3
+	movdqa	%xmm0, (%rdi)
+	movdqa	%xmm1, 16(%rdi)
+	movdqa	%xmm2, 32(%rdi)
+	movdqa	%xmm3, 48(%rdi)
+	lea	64(%rdi), %rdi
+	cmp	%rdi, %rbx
+	ja	L(mm_main_loop_forward)
+
+L(mm_copy_remaining_forward):
+	add	%r9, %rdx
+	sub	%rdi, %rdx
+/* We copied all up till %rdi position in the dst.
+	In %rdx now is how many bytes are left to copy.
+	Now we need to advance %r8. */
+	lea	(%rdi, %rsi), %r8
+
+L(mm_remaining_0_64_bytes_forward):
+	cmp	$32, %rdx
+	ja	L(mm_remaining_33_64_bytes_forward)
+	cmp	$16, %rdx
+	ja	L(mm_remaining_17_32_bytes_forward)
+	test	%rdx, %rdx
+	.p2align 4,,2
+	je	L(mm_return)
+
+	cmpb	$8, %dl
+	ja	L(mm_remaining_9_16_bytes_forward)
+	cmpb	$4, %dl
+	.p2align 4,,5
+	ja	L(mm_remaining_5_8_bytes_forward)
+	cmpb	$2, %dl
+	.p2align 4,,1
+	ja	L(mm_remaining_3_4_bytes_forward)
+	movzbl	-1(%r8,%rdx), %esi
+	movzbl	(%r8), %ebx
+	movb	%sil, -1(%rdi,%rdx)
+	movb	%bl, (%rdi)
+	jmp	L(mm_return)
+
+L(mm_remaining_33_64_bytes_forward):
+	movdqu	(%r8), %xmm0
+	movdqu	16(%r8), %xmm1
+	movdqu	-32(%r8, %rdx), %xmm2
+	movdqu	-16(%r8, %rdx), %xmm3
+	movdqu	%xmm0, (%rdi)
+	movdqu	%xmm1, 16(%rdi)
+	movdqu	%xmm2, -32(%rdi, %rdx)
+	movdqu	%xmm3, -16(%rdi, %rdx)
+	jmp	L(mm_return)
+
+L(mm_remaining_17_32_bytes_forward):
+	movdqu	(%r8), %xmm0
+	movdqu	-16(%r8, %rdx), %xmm1
+	movdqu	%xmm0, (%rdi)
+	movdqu	%xmm1, -16(%rdi, %rdx)
+	jmp	L(mm_return)
+
+L(mm_remaining_3_4_bytes_forward):
+	movzwl	-2(%r8,%rdx), %esi
+	movzwl	(%r8), %ebx
+	movw	%si, -2(%rdi,%rdx)
+	movw	%bx, (%rdi)
+	jmp	L(mm_return)
+
+L(mm_remaining_5_8_bytes_forward):
+	movl	(%r8), %esi
+	movl	-4(%r8,%rdx), %ebx
+	movl	%esi, (%rdi)
+	movl	%ebx, -4(%rdi,%rdx)
+	jmp	L(mm_return)
+
+L(mm_remaining_9_16_bytes_forward):
+	mov	(%r8), %rsi
+	mov	-8(%r8, %rdx), %rbx
+	mov	%rsi, (%rdi)
+	mov	%rbx, -8(%rdi, %rdx)
+	jmp	L(mm_return)
+
+L(mm_len_0_16_bytes_forward):
+	testb	$24, %dl
+	jne	L(mm_len_9_16_bytes_forward)
+	testb	$4, %dl
+	.p2align 4,,5
+	jne	L(mm_len_5_8_bytes_forward)
+	test	%rdx, %rdx
+	.p2align 4,,2
+	je	L(mm_return)
+	testb	$2, %dl
+	.p2align 4,,1
+	jne	L(mm_len_2_4_bytes_forward)
+	movzbl	-1(%rsi,%rdx), %ebx
+	movzbl	(%rsi), %esi
+	movb	%bl, -1(%rdi,%rdx)
+	movb	%sil, (%rdi)
+	jmp	L(mm_return)
+
+L(mm_len_2_4_bytes_forward):
+	movzwl	-2(%rsi,%rdx), %ebx
+	movzwl	(%rsi), %esi
+	movw	%bx, -2(%rdi,%rdx)
+	movw	%si, (%rdi)
+	jmp	L(mm_return)
+
+L(mm_len_5_8_bytes_forward):
+	movl	(%rsi), %ebx
+	movl	-4(%rsi,%rdx), %esi
+	movl	%ebx, (%rdi)
+	movl	%esi, -4(%rdi,%rdx)
+	jmp	L(mm_return)
+
+L(mm_len_9_16_bytes_forward):
+	mov	(%rsi), %rbx
+	mov	-8(%rsi, %rdx), %rsi
+	mov	%rbx, (%rdi)
+	mov	%rsi, -8(%rdi, %rdx)
+	jmp	L(mm_return)
+
+/* The code for copying backwards.  */
+L(mm_len_0_or_more_backward):
+
+/* Now do checks for lengths. We do [0..16], [0..32], [0..64], [0..128]
+	separately.  */
+	cmp	$16, %rdx
+	jbe	L(mm_len_0_16_bytes_backward)
+
+	cmp	$32, %rdx
+	jg	L(mm_len_32_or_more_backward)
+
+/* Copy [0..32] and return.  */
+	movdqu	(%rsi), %xmm0
+	movdqu	-16(%rsi, %rdx), %xmm1
+	movdqu	%xmm0, (%rdi)
+	movdqu	%xmm1, -16(%rdi, %rdx)
+	jmp	L(mm_return)
+
+L(mm_len_32_or_more_backward):
+	cmp	$64, %rdx
+	jg	L(mm_len_64_or_more_backward)
+
+/* Copy [0..64] and return.  */
+	movdqu	(%rsi), %xmm0
+	movdqu	16(%rsi), %xmm1
+	movdqu	-16(%rsi, %rdx), %xmm2
+	movdqu	-32(%rsi, %rdx), %xmm3
+	movdqu	%xmm0, (%rdi)
+	movdqu	%xmm1, 16(%rdi)
+	movdqu	%xmm2, -16(%rdi, %rdx)
+	movdqu	%xmm3, -32(%rdi, %rdx)
+	jmp	L(mm_return)
+
+L(mm_len_64_or_more_backward):
+	cmp	$128, %rdx
+	jg	L(mm_len_128_or_more_backward)
+
+/* Copy [0..128] and return.  */
+	movdqu	(%rsi), %xmm0
+	movdqu	16(%rsi), %xmm1
+	movdqu	32(%rsi), %xmm2
+	movdqu	48(%rsi), %xmm3
+	movdqu	-64(%rsi, %rdx), %xmm4
+	movdqu	-48(%rsi, %rdx), %xmm5
+	movdqu	-32(%rsi, %rdx), %xmm6
+	movdqu	-16(%rsi, %rdx), %xmm7
+	movdqu	%xmm0, (%rdi)
+	movdqu	%xmm1, 16(%rdi)
+	movdqu	%xmm2, 32(%rdi)
+	movdqu	%xmm3, 48(%rdi)
+	movdqu	%xmm4, -64(%rdi, %rdx)
+	movdqu	%xmm5, -48(%rdi, %rdx)
+	movdqu	%xmm6, -32(%rdi, %rdx)
+	movdqu	%xmm7, -16(%rdi, %rdx)
+	jmp	L(mm_return)
+
+L(mm_len_128_or_more_backward):
+
+	cmp	$SHARED_CACHE_SIZE_HALF, %rdx
+	jae	L(mm_large_page_backward)
+
+/* Aligning the address of destination. We need to save
+	16 bits from the source in order not to overwrite them.  */
+	movdqu	-16(%rsi, %rdx), %xmm0
+	movdqu	-32(%rsi, %rdx), %xmm1
+	movdqu	-48(%rsi, %rdx), %xmm2
+	movdqu	-64(%rsi, %rdx), %xmm3
+
+	lea	(%rdi, %rdx), %r9
+	and	$-64, %r9 /* r9 = aligned dst */
+
+	mov	%rsi, %r8
+	sub	%rdi, %r8 /* r8 = src - dst, diff */
+
+	movdqu	-16(%r9, %r8), %xmm4
+	movdqu	-32(%r9, %r8), %xmm5
+	movdqu	-48(%r9, %r8), %xmm6
+	movdqu	-64(%r9, %r8), %xmm7
+
+	movdqu	%xmm0, -16(%rdi, %rdx)
+	movdqu	%xmm1, -32(%rdi, %rdx)
+	movdqu	%xmm2, -48(%rdi, %rdx)
+	movdqu	%xmm3, -64(%rdi, %rdx)
+	movdqa	%xmm4, -16(%r9)
+	movdqa	%xmm5, -32(%r9)
+	movdqa	%xmm6, -48(%r9)
+	movdqa	%xmm7, -64(%r9)
+	lea	-64(%r9), %r9
+
+	lea	64(%rdi), %rbx
+	and	$-64, %rbx
+
+/* Compute in %rdx how many bytes are left to copy after
+	the main loop stops.  */
+	mov 	%rbx, %rdx
+	sub 	%rdi, %rdx
+
+	cmp	%r9, %rbx
+	jb	L(mm_main_loop_backward)
+	jmp	L(mm_len_0_or_more_backward)
+
+	.p2align 4
+L(mm_main_loop_backward):
+
+	prefetcht0 -128(%r9, %r8)
+
+	movdqu	-64(%r9, %r8), %xmm0
+	movdqu	-48(%r9, %r8), %xmm1
+	movdqu	-32(%r9, %r8), %xmm2
+	movdqu	-16(%r9, %r8), %xmm3
+	movdqa	%xmm0, -64(%r9)
+	movdqa	%xmm1, -48(%r9)
+	movdqa	%xmm2, -32(%r9)
+	movdqa	%xmm3, -16(%r9)
+	lea	-64(%r9), %r9
+	cmp	%r9, %rbx
+	jb	L(mm_main_loop_backward)
+	jmp	L(mm_len_0_or_more_backward)
+
+/* Copy [0..16] and return.  */
+L(mm_len_0_16_bytes_backward):
+	testb	$24, %dl
+	jnz	L(mm_len_9_16_bytes_backward)
+	testb	$4, %dl
+	.p2align 4,,5
+	jnz	L(mm_len_5_8_bytes_backward)
+	test	%rdx, %rdx
+	.p2align 4,,2
+	je	L(mm_return)
+	testb	$2, %dl
+	.p2align 4,,1
+	jne	L(mm_len_3_4_bytes_backward)
+	movzbl	-1(%rsi,%rdx), %ebx
+	movzbl	(%rsi), %ecx
+	movb	%bl, -1(%rdi,%rdx)
+	movb	%cl, (%rdi)
+	jmp	L(mm_return)
+
+L(mm_len_3_4_bytes_backward):
+	movzwl	-2(%rsi,%rdx), %ebx
+	movzwl	(%rsi), %ecx
+	movw	%bx, -2(%rdi,%rdx)
+	movw	%cx, (%rdi)
+	jmp	L(mm_return)
+
+L(mm_len_9_16_bytes_backward):
+	movl	-4(%rsi,%rdx), %ebx
+	movl	-8(%rsi,%rdx), %ecx
+	movl	%ebx, -4(%rdi,%rdx)
+	movl	%ecx, -8(%rdi,%rdx)
+	sub	$8, %rdx
+	jmp	L(mm_len_0_16_bytes_backward)
+
+L(mm_len_5_8_bytes_backward):
+	movl	(%rsi), %ebx
+	movl	-4(%rsi,%rdx), %ecx
+	movl	%ebx, (%rdi)
+	movl	%ecx, -4(%rdi,%rdx)
+
+L(mm_return):
+	RETURN
+
+/* Big length copy forward part.  */
+
+L(mm_large_page_forward):
+/* Aligning the address of destination. We need to save
+	16 bits from the source in order not to overwrite them.  */
+
+	mov	%rsi, %r8
+	mov	%rdi, %r9
+
+	movdqu	(%rsi), %xmm0
+	movdqu	16(%rsi), %xmm1
+	movdqu	32(%rsi), %xmm2
+	movdqu	48(%rsi), %xmm3
+
+	lea	64(%r9), %rdi
+	and	$-64, %rdi      /* rdi = aligned dst */
+
+	sub	%r9, %rsi        /* rsi = diff */
+
+	movdqu	(%rdi, %rsi), %xmm4
+	movdqu	16(%rdi, %rsi), %xmm5
+	movdqu	32(%rdi, %rsi), %xmm6
+	movdqu	48(%rdi, %rsi), %xmm7
+
+	movdqu	%xmm0, (%r9)
+	movdqu	%xmm1, 16(%r9)
+	movdqu	%xmm2, 32(%r9)
+	movdqu	%xmm3, 48(%r9)
+	movntdq	%xmm4, (%rdi)
+	movntdq	%xmm5, 16(%rdi)
+	movntdq	%xmm6, 32(%rdi)
+	movntdq	%xmm7, 48(%rdi)
+	add	$64, %rdi
+
+	lea	(%r9, %rdx), %rbx
+	and	$-128, %rbx
+
+	cmp	%rdi, %rbx
+	jbe	L(mm_copy_remaining_forward)
+
+	.p2align 4
+L(mm_large_page_loop_forward):
+	movdqu	(%rdi, %rsi), %xmm0
+	movdqu	16(%rdi, %rsi), %xmm1
+	movdqu	32(%rdi, %rsi), %xmm2
+	movdqu	48(%rdi, %rsi), %xmm3
+	movdqu	64(%rdi, %rsi), %xmm4
+	movdqu	80(%rdi, %rsi), %xmm5
+	movdqu	96(%rdi, %rsi), %xmm6
+	movdqu	112(%rdi, %rsi), %xmm7
+	movntdq	%xmm0, (%rdi)
+	movntdq	%xmm1, 16(%rdi)
+	movntdq	%xmm2, 32(%rdi)
+	movntdq	%xmm3, 48(%rdi)
+	movntdq	%xmm4, 64(%rdi)
+	movntdq	%xmm5, 80(%rdi)
+	movntdq	%xmm6, 96(%rdi)
+	movntdq	%xmm7, 112(%rdi)
+	lea 	128(%rdi), %rdi
+	cmp	%rdi, %rbx
+	ja	L(mm_large_page_loop_forward)
+	sfence
+
+	add 	%r9, %rdx
+	sub 	%rdi, %rdx
+/* We copied all up till %rdi position in the dst.
+	In %rdx now is how many bytes are left to copy.
+	Now we need to advance %r8. */
+	lea 	(%rdi, %rsi), %r8
+
+	cmp	$64, %rdx
+	jb	L(mm_remaining_0_64_bytes_forward)
+
+	movdqu	(%r8), %xmm0
+	movdqu	16(%r8), %xmm1
+	movdqu	32(%r8), %xmm2
+	movdqu	48(%r8), %xmm3
+	movdqu	-64(%r8, %rdx), %xmm4
+	movdqu	-48(%r8, %rdx), %xmm5
+	movdqu	-32(%r8, %rdx), %xmm6
+	movdqu	-16(%r8, %rdx), %xmm7
+	movdqu	%xmm0, (%rdi)
+	movdqu	%xmm1, 16(%rdi)
+	movdqu	%xmm2, 32(%rdi)
+	movdqu	%xmm3, 48(%rdi)
+	movdqu	%xmm4, -64(%rdi, %rdx)
+	movdqu	%xmm5, -48(%rdi, %rdx)
+	movdqu	%xmm6, -32(%rdi, %rdx)
+	movdqu	%xmm7, -16(%rdi, %rdx)
+	jmp	L(mm_return)
+
+
+/* Big length copy backward part.  */
+L(mm_large_page_backward):
+/* Aligning the address of destination. We need to save
+	16 bits from the source in order not to overwrite them.  */
+
+	movdqu	-16(%rsi, %rdx), %xmm0
+	movdqu	-32(%rsi, %rdx), %xmm1
+	movdqu	-48(%rsi, %rdx), %xmm2
+	movdqu	-64(%rsi, %rdx), %xmm3
+
+	lea	(%rdi, %rdx), %r9
+	and	$-64, %r9
+
+	mov 	%rsi, %r8
+	sub 	%rdi, %r8
+
+	movdqu	-16(%r9, %r8), %xmm4
+	movdqu	-32(%r9, %r8), %xmm5
+	movdqu	-48(%r9, %r8), %xmm6
+	movdqu	-64(%r9, %r8), %xmm7
+
+	movdqu	%xmm0, -16(%rdi, %rdx)
+	movdqu	%xmm1, -32(%rdi, %rdx)
+	movdqu	%xmm2, -48(%rdi, %rdx)
+	movdqu	%xmm3, -64(%rdi, %rdx)
+	movntdq	%xmm4, -16(%r9)
+	movntdq	%xmm5, -32(%r9)
+	movntdq	%xmm6, -48(%r9)
+	movntdq	%xmm7, -64(%r9)
+	lea 	-64(%r9), %r9
+
+	lea 	128(%rdi), %rbx
+	and 	$-64, %rbx
+
+/* Compute in %rdx how many bytes are left to copy after
+	the main loop stops.  */
+	mov 	%rbx, %rdx
+	sub 	%rdi, %rdx
+
+	cmp	%r9, %rbx
+	jae	L(mm_len_0_or_more_backward)
+
+	.p2align 4
+L(mm_large_page_loop_backward):
+	movdqu	-64(%r9, %r8), %xmm0
+	movdqu	-48(%r9, %r8), %xmm1
+	movdqu	-32(%r9, %r8), %xmm2
+	movdqu	-16(%r9, %r8), %xmm3
+	movntdq	%xmm0, -64(%r9)
+	movntdq	%xmm1, -48(%r9)
+	movntdq	%xmm2, -32(%r9)
+	movntdq	%xmm3, -16(%r9)
+	lea 	-64(%r9), %r9
+	cmp	%r9, %rbx
+	jb	L(mm_large_page_loop_backward)
+	jmp	L(mm_len_0_or_more_backward)
+
+END (MEMMOVE)
diff --git a/libc/arch-x86_64/string/sse2-memset-slm.S b/libc/arch-x86_64/string/sse2-memset-slm.S
new file mode 100644
index 0000000..bfcafae
--- /dev/null
+++ b/libc/arch-x86_64/string/sse2-memset-slm.S
@@ -0,0 +1,173 @@
+/*
+Copyright (c) 2014, Intel Corporation
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+    * Redistributions of source code must retain the above copyright notice,
+    * this list of conditions and the following disclaimer.
+
+    * Redistributions in binary form must reproduce the above copyright notice,
+    * this list of conditions and the following disclaimer in the documentation
+    * and/or other materials provided with the distribution.
+
+    * Neither the name of Intel Corporation nor the names of its contributors
+    * may be used to endorse or promote products derived from this software
+    * without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "cache.h"
+
+#ifndef MEMSET
+# define MEMSET		memset
+#endif
+
+#ifndef L
+# define L(label)	.L##label
+#endif
+
+#ifndef ALIGN
+# define ALIGN(n)	.p2align n
+#endif
+
+#ifndef cfi_startproc
+# define cfi_startproc			.cfi_startproc
+#endif
+
+#ifndef cfi_endproc
+# define cfi_endproc			.cfi_endproc
+#endif
+
+#ifndef ENTRY
+# define ENTRY(name)			\
+	.type name,  @function;	\
+	.globl name;			\
+name:					\
+	cfi_startproc
+#endif
+
+#ifndef END
+# define END(name)			\
+	cfi_endproc;			\
+	.size name, .-name
+#endif
+
+	.section .text.sse2,"ax",@progbits
+ENTRY (MEMSET)
+	movq	%rdi, %rax
+#ifdef USE_AS_BZERO_P
+	mov	%rsi, %rdx
+	xor	%rcx, %rcx
+#else
+	and	$0xff, %rsi
+	mov	$0x0101010101010101, %rcx
+	imul	%rsi, %rcx
+#endif
+	cmpq	$16, %rdx
+	jae	L(16bytesormore)
+	testb	$8, %dl
+	jnz	L(8_15bytes)
+	testb	$4, %dl
+	jnz	L(4_7bytes)
+	testb	$2, %dl
+	jnz	L(2_3bytes)
+	testb	$1, %dl
+	jz	L(return)
+	movb	%cl, (%rdi)
+L(return):
+	ret
+
+L(8_15bytes):
+	movq	%rcx, (%rdi)
+	movq	%rcx, -8(%rdi, %rdx)
+	ret
+
+L(4_7bytes):
+	movl	%ecx, (%rdi)
+	movl	%ecx, -4(%rdi, %rdx)
+	ret
+
+L(2_3bytes):
+	movw	%cx, (%rdi)
+	movw	%cx, -2(%rdi, %rdx)
+	ret
+
+	ALIGN (4)
+L(16bytesormore):
+#ifdef USE_AS_BZERO_P
+	pxor	%xmm0, %xmm0
+#else
+	movd	%rcx, %xmm0
+	pshufd	$0, %xmm0, %xmm0
+#endif
+	movdqu	%xmm0, (%rdi)
+	movdqu	%xmm0, -16(%rdi, %rdx)
+	cmpq	$32, %rdx
+	jbe	L(32bytesless)
+	movdqu	%xmm0, 16(%rdi)
+	movdqu	%xmm0, -32(%rdi, %rdx)
+	cmpq	$64, %rdx
+	jbe	L(64bytesless)
+	movdqu	%xmm0, 32(%rdi)
+	movdqu	%xmm0, 48(%rdi)
+	movdqu	%xmm0, -64(%rdi, %rdx)
+	movdqu	%xmm0, -48(%rdi, %rdx)
+	cmpq	$128, %rdx
+	ja	L(128bytesmore)
+L(32bytesless):
+L(64bytesless):
+	ret
+
+	ALIGN (4)
+L(128bytesmore):
+	leaq	64(%rdi), %rcx
+	andq	$-64, %rcx
+	movq	%rdx, %r8
+	addq	%rdi, %rdx
+	andq	$-64, %rdx
+	cmpq	%rcx, %rdx
+	je	L(return)
+
+#ifdef SHARED_CACHE_SIZE
+	cmp	$SHARED_CACHE_SIZE, %r8
+#else
+	cmp	__x86_64_shared_cache_size(%rip), %r8
+#endif
+	ja	L(128bytesmore_nt)
+
+	ALIGN (4)
+L(128bytesmore_normal):
+	movdqa	%xmm0, (%rcx)
+	movaps	%xmm0, 0x10(%rcx)
+	movaps	%xmm0, 0x20(%rcx)
+	movaps	%xmm0, 0x30(%rcx)
+	addq	$64, %rcx
+	cmpq	%rcx, %rdx
+	jne	L(128bytesmore_normal)
+	ret
+
+	ALIGN (4)
+L(128bytesmore_nt):
+	movntdq	%xmm0, (%rcx)
+	movntdq	%xmm0, 0x10(%rcx)
+	movntdq	%xmm0, 0x20(%rcx)
+	movntdq	%xmm0, 0x30(%rcx)
+	leaq	64(%rcx), %rcx
+	cmpq	%rcx, %rdx
+	jne	L(128bytesmore_nt)
+	sfence
+	ret
+
+END (MEMSET)
diff --git a/libc/arch-x86_64/string/sse2-stpcpy-slm.S b/libc/arch-x86_64/string/sse2-stpcpy-slm.S
new file mode 100644
index 0000000..0ad2d44
--- /dev/null
+++ b/libc/arch-x86_64/string/sse2-stpcpy-slm.S
@@ -0,0 +1,33 @@
+/*
+Copyright (c) 2014, Intel Corporation
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+    * Redistributions of source code must retain the above copyright notice,
+    * this list of conditions and the following disclaimer.
+
+    * Redistributions in binary form must reproduce the above copyright notice,
+    * this list of conditions and the following disclaimer in the documentation
+    * and/or other materials provided with the distribution.
+
+    * Neither the name of Intel Corporation nor the names of its contributors
+    * may be used to endorse or promote products derived from this software
+    * without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#define USE_AS_STPCPY
+#define STRCPY		stpcpy
+#include "sse2-strcpy-slm.S"
diff --git a/libc/arch-x86_64/string/sse2-stpncpy-slm.S b/libc/arch-x86_64/string/sse2-stpncpy-slm.S
new file mode 100644
index 0000000..3066685
--- /dev/null
+++ b/libc/arch-x86_64/string/sse2-stpncpy-slm.S
@@ -0,0 +1,34 @@
+/*
+Copyright (c) 2014, Intel Corporation
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+    * Redistributions of source code must retain the above copyright notice,
+    * this list of conditions and the following disclaimer.
+
+    * Redistributions in binary form must reproduce the above copyright notice,
+    * this list of conditions and the following disclaimer in the documentation
+    * and/or other materials provided with the distribution.
+
+    * Neither the name of Intel Corporation nor the names of its contributors
+    * may be used to endorse or promote products derived from this software
+    * without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#define USE_AS_STRNCPY
+#define USE_AS_STPCPY
+#define STRCPY		stpncpy
+#include "sse2-strcpy-slm.S"
diff --git a/libc/arch-x86_64/string/sse2-strcat-slm.S b/libc/arch-x86_64/string/sse2-strcat-slm.S
new file mode 100644
index 0000000..dd8207f
--- /dev/null
+++ b/libc/arch-x86_64/string/sse2-strcat-slm.S
@@ -0,0 +1,87 @@
+/*
+Copyright (c) 2014, Intel Corporation
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+    * Redistributions of source code must retain the above copyright notice,
+    * this list of conditions and the following disclaimer.
+
+    * Redistributions in binary form must reproduce the above copyright notice,
+    * this list of conditions and the following disclaimer in the documentation
+    * and/or other materials provided with the distribution.
+
+    * Neither the name of Intel Corporation nor the names of its contributors
+    * may be used to endorse or promote products derived from this software
+    * without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#ifndef STRCAT
+# define STRCAT		strcat
+#endif
+
+#ifndef L
+# define L(label)		.L##label
+#endif
+
+#ifndef cfi_startproc
+# define cfi_startproc		 .cfi_startproc
+#endif
+
+#ifndef cfi_endproc
+# define cfi_endproc		.cfi_endproc
+#endif
+
+#ifndef ENTRY
+# define ENTRY(name)		\
+	.type name,  @function;		\
+	.globl name;		\
+	.p2align 4;		\
+name:		\
+	cfi_startproc
+#endif
+
+#ifndef END
+# define END(name)		\
+	cfi_endproc;		\
+	.size name, .-name
+#endif
+
+#define USE_AS_STRCAT
+
+.text
+ENTRY (STRCAT)
+	mov	%rdi, %r9
+#ifdef USE_AS_STRNCAT
+	mov	%rdx, %r8
+#endif
+
+#define RETURN jmp L(Strcpy)
+#include "sse2-strlen-slm.S"
+
+#undef RETURN
+#define RETURN ret
+
+L(Strcpy):
+	lea	(%r9, %rax), %rdi
+	mov	%rsi, %rcx
+	mov	%r9, %rax	/* save result */
+
+#ifdef USE_AS_STRNCAT
+	test	%r8, %r8
+	jz	L(ExitZero)
+# define USE_AS_STRNCPY
+#endif
+#include "sse2-strcpy-slm.S"
diff --git a/libc/arch-x86_64/string/sse2-strcpy-slm.S b/libc/arch-x86_64/string/sse2-strcpy-slm.S
new file mode 100644
index 0000000..3e146bf
--- /dev/null
+++ b/libc/arch-x86_64/string/sse2-strcpy-slm.S
@@ -0,0 +1,1921 @@
+/*
+Copyright (c) 2014, Intel Corporation
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+    * Redistributions of source code must retain the above copyright notice,
+    * this list of conditions and the following disclaimer.
+
+    * Redistributions in binary form must reproduce the above copyright notice,
+    * this list of conditions and the following disclaimer in the documentation
+    * and/or other materials provided with the distribution.
+
+    * Neither the name of Intel Corporation nor the names of its contributors
+    * may be used to endorse or promote products derived from this software
+    * without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#ifndef USE_AS_STRCAT
+
+# ifndef STRCPY
+#  define STRCPY	strcpy
+# endif
+
+# ifndef L
+#  define L(label)	.L##label
+# endif
+
+# ifndef cfi_startproc
+#  define cfi_startproc	.cfi_startproc
+# endif
+
+# ifndef cfi_endproc
+#  define cfi_endproc	.cfi_endproc
+# endif
+
+# ifndef ENTRY
+#  define ENTRY(name)	\
+	.type name, @function;	\
+	.globl name;	\
+	.p2align 4;	\
+name:	\
+	cfi_startproc
+# endif
+
+# ifndef END
+#  define END(name)	\
+	cfi_endproc;	\
+	.size name, .-name
+# endif
+
+#endif
+
+#define JMPTBL(I, B)	I - B
+#define BRANCH_TO_JMPTBL_ENTRY(TABLE, INDEX, SCALE)	\
+	lea	TABLE(%rip), %r11;	\
+	movslq	(%r11, INDEX, SCALE), %rcx;	\
+	lea	(%r11, %rcx), %rcx;	\
+	jmp	*%rcx
+
+#ifndef USE_AS_STRCAT
+
+# define RETURN ret
+
+.text
+ENTRY (STRCPY)
+# ifdef USE_AS_STRNCPY
+	mov	%rdx, %r8
+	test	%r8, %r8
+	jz	L(ExitZero)
+# endif
+	mov	%rsi, %rcx
+# ifndef USE_AS_STPCPY
+	mov	%rdi, %rax      /* save result */
+# endif
+
+#endif
+	and	$63, %rcx
+	cmp	$32, %rcx
+	jbe	L(SourceStringAlignmentLess32)
+
+	and	$-16, %rsi
+	and	$15, %rcx
+	pxor	%xmm0, %xmm0
+	pxor	%xmm1, %xmm1
+
+	pcmpeqb	(%rsi), %xmm1
+	pmovmskb %xmm1, %rdx
+	shr	%cl, %rdx
+#ifdef USE_AS_STRNCPY
+# if defined USE_AS_STPCPY || defined USE_AS_STRCAT
+	mov	$16, %r10
+	sub	%rcx, %r10
+	cmp	%r10, %r8
+# else
+	mov	$17, %r10
+	sub	%rcx, %r10
+	cmp	%r10, %r8
+# endif
+	jbe	L(CopyFrom1To16BytesTailCase2OrCase3)
+#endif
+	test	%rdx, %rdx
+	jnz	L(CopyFrom1To16BytesTail)
+
+	pcmpeqb	16(%rsi), %xmm0
+	pmovmskb %xmm0, %rdx
+#ifdef USE_AS_STRNCPY
+	add	$16, %r10
+	cmp	%r10, %r8
+	jbe	L(CopyFrom1To32BytesCase2OrCase3)
+#endif
+	test	%rdx, %rdx
+	jnz	L(CopyFrom1To32Bytes)
+
+	movdqu	(%rsi, %rcx), %xmm1   /* copy 16 bytes */
+	movdqu	%xmm1, (%rdi)
+
+/* If source adress alignment != destination adress alignment */
+	.p2align 4
+L(Unalign16Both):
+	sub	%rcx, %rdi
+#ifdef USE_AS_STRNCPY
+	add	%rcx, %r8
+#endif
+	mov	$16, %rcx
+	movdqa	(%rsi, %rcx), %xmm1
+	movaps	16(%rsi, %rcx), %xmm2
+	movdqu	%xmm1, (%rdi, %rcx)
+	pcmpeqb	%xmm2, %xmm0
+	pmovmskb %xmm0, %rdx
+	add	$16, %rcx
+#ifdef USE_AS_STRNCPY
+	sub	$48, %r8
+	jbe	L(CopyFrom1To16BytesCase2OrCase3)
+#endif
+	test	%rdx, %rdx
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	jnz	L(CopyFrom1To16BytesUnalignedXmm2)
+#else
+	jnz	L(CopyFrom1To16Bytes)
+#endif
+
+	movaps	16(%rsi, %rcx), %xmm3
+	movdqu	%xmm2, (%rdi, %rcx)
+	pcmpeqb	%xmm3, %xmm0
+	pmovmskb %xmm0, %rdx
+	add	$16, %rcx
+#ifdef USE_AS_STRNCPY
+	sub	$16, %r8
+	jbe	L(CopyFrom1To16BytesCase2OrCase3)
+#endif
+	test	%rdx, %rdx
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	jnz	L(CopyFrom1To16BytesUnalignedXmm3)
+#else
+	jnz	L(CopyFrom1To16Bytes)
+#endif
+
+	movaps	16(%rsi, %rcx), %xmm4
+	movdqu	%xmm3, (%rdi, %rcx)
+	pcmpeqb	%xmm4, %xmm0
+	pmovmskb %xmm0, %rdx
+	add	$16, %rcx
+#ifdef USE_AS_STRNCPY
+	sub	$16, %r8
+	jbe	L(CopyFrom1To16BytesCase2OrCase3)
+#endif
+	test	%rdx, %rdx
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	jnz	L(CopyFrom1To16BytesUnalignedXmm4)
+#else
+	jnz	L(CopyFrom1To16Bytes)
+#endif
+
+	movaps	16(%rsi, %rcx), %xmm1
+	movdqu	%xmm4, (%rdi, %rcx)
+	pcmpeqb	%xmm1, %xmm0
+	pmovmskb %xmm0, %rdx
+	add	$16, %rcx
+#ifdef USE_AS_STRNCPY
+	sub	$16, %r8
+	jbe	L(CopyFrom1To16BytesCase2OrCase3)
+#endif
+	test	%rdx, %rdx
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	jnz	L(CopyFrom1To16BytesUnalignedXmm1)
+#else
+	jnz	L(CopyFrom1To16Bytes)
+#endif
+
+	movaps	16(%rsi, %rcx), %xmm2
+	movdqu	%xmm1, (%rdi, %rcx)
+	pcmpeqb	%xmm2, %xmm0
+	pmovmskb %xmm0, %rdx
+	add	$16, %rcx
+#ifdef USE_AS_STRNCPY
+	sub	$16, %r8
+	jbe	L(CopyFrom1To16BytesCase2OrCase3)
+#endif
+	test	%rdx, %rdx
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	jnz	L(CopyFrom1To16BytesUnalignedXmm2)
+#else
+	jnz	L(CopyFrom1To16Bytes)
+#endif
+
+	movaps	16(%rsi, %rcx), %xmm3
+	movdqu	%xmm2, (%rdi, %rcx)
+	pcmpeqb	%xmm3, %xmm0
+	pmovmskb %xmm0, %rdx
+	add	$16, %rcx
+#ifdef USE_AS_STRNCPY
+	sub	$16, %r8
+	jbe	L(CopyFrom1To16BytesCase2OrCase3)
+#endif
+	test	%rdx, %rdx
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	jnz	L(CopyFrom1To16BytesUnalignedXmm3)
+#else
+	jnz	L(CopyFrom1To16Bytes)
+#endif
+
+	movdqu	%xmm3, (%rdi, %rcx)
+	mov	%rsi, %rdx
+	lea	16(%rsi, %rcx), %rsi
+	and	$-0x40, %rsi
+	sub	%rsi, %rdx
+	sub	%rdx, %rdi
+#ifdef USE_AS_STRNCPY
+	lea	128(%r8, %rdx), %r8
+#endif
+L(Unaligned64Loop):
+	movaps	(%rsi), %xmm2
+	movaps	%xmm2, %xmm4
+	movaps	16(%rsi), %xmm5
+	movaps	32(%rsi), %xmm3
+	movaps	%xmm3, %xmm6
+	movaps	48(%rsi), %xmm7
+	pminub	%xmm5, %xmm2
+	pminub	%xmm7, %xmm3
+	pminub	%xmm2, %xmm3
+	pcmpeqb	%xmm0, %xmm3
+	pmovmskb %xmm3, %rdx
+#ifdef USE_AS_STRNCPY
+	sub	$64, %r8
+	jbe	L(UnalignedLeaveCase2OrCase3)
+#endif
+	test	%rdx, %rdx
+	jnz	L(Unaligned64Leave)
+
+L(Unaligned64Loop_start):
+	add	$64, %rdi
+	add	$64, %rsi
+	movdqu	%xmm4, -64(%rdi)
+	movaps	(%rsi), %xmm2
+	movdqa	%xmm2, %xmm4
+	movdqu	%xmm5, -48(%rdi)
+	movaps	16(%rsi), %xmm5
+	pminub	%xmm5, %xmm2
+	movaps	32(%rsi), %xmm3
+	movdqu	%xmm6, -32(%rdi)
+	movaps	%xmm3, %xmm6
+	movdqu	%xmm7, -16(%rdi)
+	movaps	48(%rsi), %xmm7
+	pminub	%xmm7, %xmm3
+	pminub	%xmm2, %xmm3
+	pcmpeqb	%xmm0, %xmm3
+	pmovmskb %xmm3, %rdx
+#ifdef USE_AS_STRNCPY
+	sub	$64, %r8
+	jbe	L(UnalignedLeaveCase2OrCase3)
+#endif
+	test	%rdx, %rdx
+	jz	L(Unaligned64Loop_start)
+
+L(Unaligned64Leave):
+	pxor	%xmm1, %xmm1
+
+	pcmpeqb	%xmm4, %xmm0
+	pcmpeqb	%xmm5, %xmm1
+	pmovmskb %xmm0, %rdx
+	pmovmskb %xmm1, %rcx
+	test	%rdx, %rdx
+	jnz	L(CopyFrom1To16BytesUnaligned_0)
+	test	%rcx, %rcx
+	jnz	L(CopyFrom1To16BytesUnaligned_16)
+
+	pcmpeqb	%xmm6, %xmm0
+	pcmpeqb	%xmm7, %xmm1
+	pmovmskb %xmm0, %rdx
+	pmovmskb %xmm1, %rcx
+	test	%rdx, %rdx
+	jnz	L(CopyFrom1To16BytesUnaligned_32)
+
+	bsf	%rcx, %rdx
+	movdqu	%xmm4, (%rdi)
+	movdqu	%xmm5, 16(%rdi)
+	movdqu	%xmm6, 32(%rdi)
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+# ifdef USE_AS_STPCPY
+	lea	48(%rdi, %rdx), %rax
+# endif
+	movdqu	%xmm7, 48(%rdi)
+	add	$15, %r8
+	sub	%rdx, %r8
+	lea	49(%rdi, %rdx), %rdi
+	jmp	L(StrncpyFillTailWithZero)
+#else
+	add	$48, %rsi
+	add	$48, %rdi
+	BRANCH_TO_JMPTBL_ENTRY (L(ExitTable), %rdx, 4)
+#endif
+
+/* If source adress alignment == destination adress alignment */
+
+L(SourceStringAlignmentLess32):
+	pxor	%xmm0, %xmm0
+	movdqu	(%rsi), %xmm1
+	movdqu	16(%rsi), %xmm2
+	pcmpeqb	%xmm1, %xmm0
+	pmovmskb %xmm0, %rdx
+
+#ifdef USE_AS_STRNCPY
+# if defined USE_AS_STPCPY || defined USE_AS_STRCAT
+	cmp	$16, %r8
+# else
+	cmp	$17, %r8
+# endif
+	jbe	L(CopyFrom1To16BytesTail1Case2OrCase3)
+#endif
+	test	%rdx, %rdx
+	jnz	L(CopyFrom1To16BytesTail1)
+
+	pcmpeqb	%xmm2, %xmm0
+	movdqu	%xmm1, (%rdi)
+	pmovmskb %xmm0, %rdx
+
+#ifdef USE_AS_STRNCPY
+# if defined USE_AS_STPCPY || defined USE_AS_STRCAT
+	cmp	$32, %r8
+# else
+	cmp	$33, %r8
+# endif
+	jbe	L(CopyFrom1To32Bytes1Case2OrCase3)
+#endif
+	test	%rdx, %rdx
+	jnz	L(CopyFrom1To32Bytes1)
+
+	and	$15, %rcx
+	and	$-16, %rsi
+
+	jmp	L(Unalign16Both)
+
+/*------End of main part with loops---------------------*/
+
+/* Case1 */
+
+#if (!defined USE_AS_STRNCPY) || (defined USE_AS_STRCAT)
+	.p2align 4
+L(CopyFrom1To16Bytes):
+	add	%rcx, %rdi
+	add	%rcx, %rsi
+	bsf	%rdx, %rdx
+	BRANCH_TO_JMPTBL_ENTRY (L(ExitTable), %rdx, 4)
+#endif
+	.p2align 4
+L(CopyFrom1To16BytesTail):
+	add	%rcx, %rsi
+	bsf	%rdx, %rdx
+	BRANCH_TO_JMPTBL_ENTRY (L(ExitTable), %rdx, 4)
+
+	.p2align 4
+L(CopyFrom1To32Bytes1):
+	add	$16, %rsi
+	add	$16, %rdi
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$16, %r8
+#endif
+L(CopyFrom1To16BytesTail1):
+	bsf	%rdx, %rdx
+	BRANCH_TO_JMPTBL_ENTRY (L(ExitTable), %rdx, 4)
+
+	.p2align 4
+L(CopyFrom1To32Bytes):
+	bsf	%rdx, %rdx
+	add	%rcx, %rsi
+	add	$16, %rdx
+	sub	%rcx, %rdx
+	BRANCH_TO_JMPTBL_ENTRY (L(ExitTable), %rdx, 4)
+
+	.p2align 4
+L(CopyFrom1To16BytesUnaligned_0):
+	bsf	%rdx, %rdx
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+# ifdef USE_AS_STPCPY
+	lea	(%rdi, %rdx), %rax
+# endif
+	movdqu	%xmm4, (%rdi)
+	add	$63, %r8
+	sub	%rdx, %r8
+	lea	1(%rdi, %rdx), %rdi
+	jmp	L(StrncpyFillTailWithZero)
+#else
+	BRANCH_TO_JMPTBL_ENTRY (L(ExitTable), %rdx, 4)
+#endif
+
+	.p2align 4
+L(CopyFrom1To16BytesUnaligned_16):
+	bsf	%rcx, %rdx
+	movdqu	%xmm4, (%rdi)
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+# ifdef USE_AS_STPCPY
+	lea	16(%rdi, %rdx), %rax
+# endif
+	movdqu	%xmm5, 16(%rdi)
+	add	$47, %r8
+	sub	%rdx, %r8
+	lea	17(%rdi, %rdx), %rdi
+	jmp	L(StrncpyFillTailWithZero)
+#else
+	add	$16, %rsi
+	add	$16, %rdi
+	BRANCH_TO_JMPTBL_ENTRY (L(ExitTable), %rdx, 4)
+#endif
+
+	.p2align 4
+L(CopyFrom1To16BytesUnaligned_32):
+	bsf	%rdx, %rdx
+	movdqu	%xmm4, (%rdi)
+	movdqu	%xmm5, 16(%rdi)
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+# ifdef USE_AS_STPCPY
+	lea	32(%rdi, %rdx), %rax
+# endif
+	movdqu	%xmm6, 32(%rdi)
+	add	$31, %r8
+	sub	%rdx, %r8
+	lea	33(%rdi, %rdx), %rdi
+	jmp	L(StrncpyFillTailWithZero)
+#else
+	add	$32, %rsi
+	add	$32, %rdi
+	BRANCH_TO_JMPTBL_ENTRY (L(ExitTable), %rdx, 4)
+#endif
+
+#ifdef USE_AS_STRNCPY
+# ifndef USE_AS_STRCAT 
+	.p2align 4
+L(CopyFrom1To16BytesUnalignedXmm6):
+	movdqu	%xmm6, (%rdi, %rcx)
+	jmp	L(CopyFrom1To16BytesXmmExit)
+
+	.p2align 4
+L(CopyFrom1To16BytesUnalignedXmm5):
+	movdqu	%xmm5, (%rdi, %rcx)
+	jmp	L(CopyFrom1To16BytesXmmExit)
+
+	.p2align 4
+L(CopyFrom1To16BytesUnalignedXmm4):
+	movdqu	%xmm4, (%rdi, %rcx)
+	jmp	L(CopyFrom1To16BytesXmmExit)
+
+	.p2align 4
+L(CopyFrom1To16BytesUnalignedXmm3):
+	movdqu	%xmm3, (%rdi, %rcx)
+	jmp	L(CopyFrom1To16BytesXmmExit)
+
+	.p2align 4
+L(CopyFrom1To16BytesUnalignedXmm1):
+	movdqu	%xmm1, (%rdi, %rcx)
+	jmp	L(CopyFrom1To16BytesXmmExit)
+# endif
+
+	.p2align 4
+L(CopyFrom1To16BytesExit):
+	BRANCH_TO_JMPTBL_ENTRY (L(ExitTable), %rdx, 4)
+
+/* Case2 */
+
+	.p2align 4
+L(CopyFrom1To16BytesCase2):
+	add	$16, %r8
+	add	%rcx, %rdi
+	add	%rcx, %rsi
+	bsf	%rdx, %rdx
+	cmp	%r8, %rdx
+	jb	L(CopyFrom1To16BytesExit)
+	BRANCH_TO_JMPTBL_ENTRY (L(ExitStrncpyTable), %r8, 4)
+
+	.p2align 4
+L(CopyFrom1To32BytesCase2):
+	add	%rcx, %rsi
+	bsf	%rdx, %rdx
+	add	$16, %rdx
+	sub	%rcx, %rdx
+	cmp	%r8, %rdx
+	jb	L(CopyFrom1To16BytesExit)
+	BRANCH_TO_JMPTBL_ENTRY (L(ExitStrncpyTable), %r8, 4)
+
+L(CopyFrom1To16BytesTailCase2):
+	add	%rcx, %rsi
+	bsf	%rdx, %rdx
+	cmp	%r8, %rdx
+	jb	L(CopyFrom1To16BytesExit)
+	BRANCH_TO_JMPTBL_ENTRY (L(ExitStrncpyTable), %r8, 4)
+
+L(CopyFrom1To16BytesTail1Case2):
+	bsf	%rdx, %rdx
+	cmp	%r8, %rdx
+	jb	L(CopyFrom1To16BytesExit)
+	BRANCH_TO_JMPTBL_ENTRY (L(ExitStrncpyTable), %r8, 4)
+
+/* Case2 or Case3,  Case3 */
+
+	.p2align 4
+L(CopyFrom1To16BytesCase2OrCase3):
+	test	%rdx, %rdx
+	jnz	L(CopyFrom1To16BytesCase2)
+L(CopyFrom1To16BytesCase3):
+	add	$16, %r8
+	add	%rcx, %rdi
+	add	%rcx, %rsi
+	BRANCH_TO_JMPTBL_ENTRY (L(ExitStrncpyTable), %r8, 4)
+
+	.p2align 4
+L(CopyFrom1To32BytesCase2OrCase3):
+	test	%rdx, %rdx
+	jnz	L(CopyFrom1To32BytesCase2)
+	add	%rcx, %rsi
+	BRANCH_TO_JMPTBL_ENTRY (L(ExitStrncpyTable), %r8, 4)
+
+	.p2align 4
+L(CopyFrom1To16BytesTailCase2OrCase3):
+	test	%rdx, %rdx
+	jnz	L(CopyFrom1To16BytesTailCase2)
+	add	%rcx, %rsi
+	BRANCH_TO_JMPTBL_ENTRY (L(ExitStrncpyTable), %r8, 4)
+
+	.p2align 4
+L(CopyFrom1To32Bytes1Case2OrCase3):
+	add	$16, %rdi
+	add	$16, %rsi
+	sub	$16, %r8
+L(CopyFrom1To16BytesTail1Case2OrCase3):
+	test	%rdx, %rdx
+	jnz	L(CopyFrom1To16BytesTail1Case2)
+	BRANCH_TO_JMPTBL_ENTRY (L(ExitStrncpyTable), %r8, 4)
+
+#endif
+
+/*------------End labels regarding with copying 1-16 bytes--and 1-32 bytes----*/
+
+	.p2align 4
+L(Exit1):
+	mov	%dh, (%rdi)
+#ifdef USE_AS_STPCPY
+	lea	(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$1, %r8
+	lea	1(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit2):
+	mov	(%rsi), %dx
+	mov	%dx, (%rdi)
+#ifdef USE_AS_STPCPY
+	lea	1(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$2, %r8
+	lea	2(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit3):
+	mov	(%rsi), %cx
+	mov	%cx, (%rdi)
+	mov	%dh, 2(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	2(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$3, %r8
+	lea	3(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit4):
+	mov	(%rsi), %edx
+	mov	%edx, (%rdi)
+#ifdef USE_AS_STPCPY
+	lea	3(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$4, %r8
+	lea	4(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit5):
+	mov	(%rsi), %ecx
+	mov	%dh, 4(%rdi)
+	mov	%ecx, (%rdi)
+#ifdef USE_AS_STPCPY
+	lea	4(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$5, %r8
+	lea	5(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit6):
+	mov	(%rsi), %ecx
+	mov	4(%rsi), %dx
+	mov	%ecx, (%rdi)
+	mov	%dx, 4(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	5(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$6, %r8
+	lea	6(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit7):
+	mov	(%rsi), %ecx
+	mov	3(%rsi), %edx
+	mov	%ecx, (%rdi)
+	mov	%edx, 3(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	6(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$7, %r8
+	lea	7(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit8):
+	mov	(%rsi), %rdx
+	mov	%rdx, (%rdi)
+#ifdef USE_AS_STPCPY
+	lea	7(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$8, %r8
+	lea	8(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit9):
+	mov	(%rsi), %rcx
+	mov	%dh, 8(%rdi)
+	mov	%rcx, (%rdi)
+#ifdef USE_AS_STPCPY
+	lea	8(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$9, %r8
+	lea	9(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit10):
+	mov	(%rsi), %rcx
+	mov	8(%rsi), %dx
+	mov	%rcx, (%rdi)
+	mov	%dx, 8(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	9(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$10, %r8
+	lea	10(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit11):
+	mov	(%rsi), %rcx
+	mov	7(%rsi), %edx
+	mov	%rcx, (%rdi)
+	mov	%edx, 7(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	10(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$11, %r8
+	lea	11(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit12):
+	mov	(%rsi), %rcx
+	mov	8(%rsi), %edx
+	mov	%rcx, (%rdi)
+	mov	%edx, 8(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	11(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$12, %r8
+	lea	12(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit13):
+	mov	(%rsi), %rcx
+	mov	5(%rsi), %rdx
+	mov	%rcx, (%rdi)
+	mov	%rdx, 5(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	12(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$13, %r8
+	lea	13(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit14):
+	mov	(%rsi), %rcx
+	mov	6(%rsi), %rdx
+	mov	%rcx, (%rdi)
+	mov	%rdx, 6(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	13(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$14, %r8
+	lea	14(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit15):
+	mov	(%rsi), %rcx
+	mov	7(%rsi), %rdx
+	mov	%rcx, (%rdi)
+	mov	%rdx, 7(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	14(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$15, %r8
+	lea	15(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit16):
+	movdqu	(%rsi), %xmm0
+	movdqu	%xmm0, (%rdi)
+#ifdef USE_AS_STPCPY
+	lea	15(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$16, %r8
+	lea	16(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit17):
+	movdqu	(%rsi), %xmm0
+	movdqu	%xmm0, (%rdi)
+	mov	%dh, 16(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	16(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$17, %r8
+	lea	17(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit18):
+	movdqu	(%rsi), %xmm0
+	mov	16(%rsi), %cx
+	movdqu	%xmm0, (%rdi)
+	mov	%cx, 16(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	17(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$18, %r8
+	lea	18(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit19):
+	movdqu	(%rsi), %xmm0
+	mov	15(%rsi), %ecx
+	movdqu	%xmm0, (%rdi)
+	mov	%ecx, 15(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	18(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$19, %r8
+	lea	19(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit20):
+	movdqu	(%rsi), %xmm0
+	mov	16(%rsi), %ecx
+	movdqu	%xmm0, (%rdi)
+	mov	%ecx, 16(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	19(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$20, %r8
+	lea	20(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit21):
+	movdqu	(%rsi), %xmm0
+	mov	16(%rsi), %ecx
+	movdqu	%xmm0, (%rdi)
+	mov	%ecx, 16(%rdi)
+	mov	%dh, 20(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	20(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$21, %r8
+	lea	21(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit22):
+	movdqu	(%rsi), %xmm0
+	mov	14(%rsi), %rcx
+	movdqu	%xmm0, (%rdi)
+	mov	%rcx, 14(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	21(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$22, %r8
+	lea	22(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit23):
+	movdqu	(%rsi), %xmm0
+	mov	15(%rsi), %rcx
+	movdqu	%xmm0, (%rdi)
+	mov	%rcx, 15(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	22(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$23, %r8
+	lea	23(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit24):
+	movdqu	(%rsi), %xmm0
+	mov	16(%rsi), %rcx
+	movdqu	%xmm0, (%rdi)
+	mov	%rcx, 16(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	23(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$24, %r8
+	lea	24(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit25):
+	movdqu	(%rsi), %xmm0
+	mov	16(%rsi), %rcx
+	movdqu	%xmm0, (%rdi)
+	mov	%rcx, 16(%rdi)
+	mov	%dh, 24(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	24(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$25, %r8
+	lea	25(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit26):
+	movdqu	(%rsi), %xmm0
+	mov	16(%rsi), %rdx
+	mov	24(%rsi), %cx
+	movdqu	%xmm0, (%rdi)
+	mov	%rdx, 16(%rdi)
+	mov	%cx, 24(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	25(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$26, %r8
+	lea	26(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit27):
+	movdqu	(%rsi), %xmm0
+	mov	16(%rsi), %rdx
+	mov	23(%rsi), %ecx
+	movdqu	%xmm0, (%rdi)
+	mov	%rdx, 16(%rdi)
+	mov	%ecx, 23(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	26(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$27, %r8
+	lea	27(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit28):
+	movdqu	(%rsi), %xmm0
+	mov	16(%rsi), %rdx
+	mov	24(%rsi), %ecx
+	movdqu	%xmm0, (%rdi)
+	mov	%rdx, 16(%rdi)
+	mov	%ecx, 24(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	27(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$28, %r8
+	lea	28(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit29):
+	movdqu	(%rsi), %xmm0
+	movdqu	13(%rsi), %xmm2
+	movdqu	%xmm0, (%rdi)
+	movdqu	%xmm2, 13(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	28(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$29, %r8
+	lea	29(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit30):
+	movdqu	(%rsi), %xmm0
+	movdqu	14(%rsi), %xmm2
+	movdqu	%xmm0, (%rdi)
+	movdqu	%xmm2, 14(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	29(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$30, %r8
+	lea	30(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit31):
+	movdqu	(%rsi), %xmm0
+	movdqu	15(%rsi), %xmm2
+	movdqu	%xmm0, (%rdi)
+	movdqu	%xmm2, 15(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	30(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$31, %r8
+	lea	31(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+	.p2align 4
+L(Exit32):
+	movdqu	(%rsi), %xmm0
+	movdqu	16(%rsi), %xmm2
+	movdqu	%xmm0, (%rdi)
+	movdqu	%xmm2, 16(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	31(%rdi), %rax
+#endif
+#if defined USE_AS_STRNCPY && !defined USE_AS_STRCAT
+	sub	$32, %r8
+	lea	32(%rdi), %rdi
+	jnz	L(StrncpyFillTailWithZero)
+#endif
+	RETURN
+
+#ifdef USE_AS_STRNCPY
+
+	.p2align 4
+L(StrncpyExit0):
+#ifdef USE_AS_STPCPY
+	mov	%rdi, %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, (%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit1):
+	mov	(%rsi), %dl
+	mov	%dl, (%rdi)
+#ifdef USE_AS_STPCPY
+	lea	1(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 1(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit2):
+	mov	(%rsi), %dx
+	mov	%dx, (%rdi)
+#ifdef USE_AS_STPCPY
+	lea	2(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 2(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit3):
+	mov	(%rsi), %cx
+	mov	2(%rsi), %dl
+	mov	%cx, (%rdi)
+	mov	%dl, 2(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	3(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 3(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit4):
+	mov	(%rsi), %edx
+	mov	%edx, (%rdi)
+#ifdef USE_AS_STPCPY
+	lea	4(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 4(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit5):
+	mov	(%rsi), %ecx
+	mov	4(%rsi), %dl
+	mov	%ecx, (%rdi)
+	mov	%dl, 4(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	5(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 5(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit6):
+	mov	(%rsi), %ecx
+	mov	4(%rsi), %dx
+	mov	%ecx, (%rdi)
+	mov	%dx, 4(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	6(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 6(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit7):
+	mov	(%rsi), %ecx
+	mov	3(%rsi), %edx
+	mov	%ecx, (%rdi)
+	mov	%edx, 3(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	7(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 7(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit8):
+	mov	(%rsi), %rdx
+	mov	%rdx, (%rdi)
+#ifdef USE_AS_STPCPY
+	lea	8(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 8(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit9):
+	mov	(%rsi), %rcx
+	mov	8(%rsi), %dl
+	mov	%rcx, (%rdi)
+	mov	%dl, 8(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	9(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 9(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit10):
+	mov	(%rsi), %rcx
+	mov	8(%rsi), %dx
+	mov	%rcx, (%rdi)
+	mov	%dx, 8(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	10(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 10(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit11):
+	mov	(%rsi), %rcx
+	mov	7(%rsi), %edx
+	mov	%rcx, (%rdi)
+	mov	%edx, 7(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	11(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 11(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit12):
+	mov	(%rsi), %rcx
+	mov	8(%rsi), %edx
+	mov	%rcx, (%rdi)
+	mov	%edx, 8(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	12(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 12(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit13):
+	mov	(%rsi), %rcx
+	mov	5(%rsi), %rdx
+	mov	%rcx, (%rdi)
+	mov	%rdx, 5(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	13(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 13(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit14):
+	mov	(%rsi), %rcx
+	mov	6(%rsi), %rdx
+	mov	%rcx, (%rdi)
+	mov	%rdx, 6(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	14(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 14(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit15):
+	mov	(%rsi), %rcx
+	mov	7(%rsi), %rdx
+	mov	%rcx, (%rdi)
+	mov	%rdx, 7(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	15(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 15(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit16):
+	movdqu	(%rsi), %xmm0
+	movdqu	%xmm0, (%rdi)
+#ifdef USE_AS_STPCPY
+	lea	16(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 16(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit17):
+	movdqu	(%rsi), %xmm0
+	mov	16(%rsi), %cl
+	movdqu	%xmm0, (%rdi)
+	mov	%cl, 16(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	17(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 17(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit18):
+	movdqu	(%rsi), %xmm0
+	mov	16(%rsi), %cx
+	movdqu	%xmm0, (%rdi)
+	mov	%cx, 16(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	18(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 18(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit19):
+	movdqu	(%rsi), %xmm0
+	mov	15(%rsi), %ecx
+	movdqu	%xmm0, (%rdi)
+	mov	%ecx, 15(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	19(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 19(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit20):
+	movdqu	(%rsi), %xmm0
+	mov	16(%rsi), %ecx
+	movdqu	%xmm0, (%rdi)
+	mov	%ecx, 16(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	20(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 20(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit21):
+	movdqu	(%rsi), %xmm0
+	mov	16(%rsi), %ecx
+	mov	20(%rsi), %dl
+	movdqu	%xmm0, (%rdi)
+	mov	%ecx, 16(%rdi)
+	mov	%dl, 20(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	21(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 21(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit22):
+	movdqu	(%rsi), %xmm0
+	mov	14(%rsi), %rcx
+	movdqu	%xmm0, (%rdi)
+	mov	%rcx, 14(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	22(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 22(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit23):
+	movdqu	(%rsi), %xmm0
+	mov	15(%rsi), %rcx
+	movdqu	%xmm0, (%rdi)
+	mov	%rcx, 15(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	23(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 23(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit24):
+	movdqu	(%rsi), %xmm0
+	mov	16(%rsi), %rcx
+	movdqu	%xmm0, (%rdi)
+	mov	%rcx, 16(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	24(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 24(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit25):
+	movdqu	(%rsi), %xmm0
+	mov	16(%rsi), %rdx
+	mov	24(%rsi), %cl
+	movdqu	%xmm0, (%rdi)
+	mov	%rdx, 16(%rdi)
+	mov	%cl, 24(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	25(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 25(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit26):
+	movdqu	(%rsi), %xmm0
+	mov	16(%rsi), %rdx
+	mov	24(%rsi), %cx
+	movdqu	%xmm0, (%rdi)
+	mov	%rdx, 16(%rdi)
+	mov	%cx, 24(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	26(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 26(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit27):
+	movdqu	(%rsi), %xmm0
+	mov	16(%rsi), %rdx
+	mov	23(%rsi), %ecx
+	movdqu	%xmm0, (%rdi)
+	mov	%rdx, 16(%rdi)
+	mov	%ecx, 23(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	27(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 27(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit28):
+	movdqu	(%rsi), %xmm0
+	mov	16(%rsi), %rdx
+	mov	24(%rsi), %ecx
+	movdqu	%xmm0, (%rdi)
+	mov	%rdx, 16(%rdi)
+	mov	%ecx, 24(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	28(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 28(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit29):
+	movdqu	(%rsi), %xmm0
+	movdqu	13(%rsi), %xmm2
+	movdqu	%xmm0, (%rdi)
+	movdqu	%xmm2, 13(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	29(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 29(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit30):
+	movdqu	(%rsi), %xmm0
+	movdqu	14(%rsi), %xmm2
+	movdqu	%xmm0, (%rdi)
+	movdqu	%xmm2, 14(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	30(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 30(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit31):
+	movdqu	(%rsi), %xmm0
+	movdqu	15(%rsi), %xmm2
+	movdqu	%xmm0, (%rdi)
+	movdqu	%xmm2, 15(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	31(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 31(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit32):
+	movdqu	(%rsi), %xmm0
+	movdqu	16(%rsi), %xmm2
+	movdqu	%xmm0, (%rdi)
+	movdqu	%xmm2, 16(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	32(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 32(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(StrncpyExit33):
+	movdqu	(%rsi), %xmm0
+	movdqu	16(%rsi), %xmm2
+	mov	32(%rsi), %cl
+	movdqu	%xmm0, (%rdi)
+	movdqu	%xmm2, 16(%rdi)
+	mov	%cl, 32(%rdi)
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 33(%rdi)
+#endif
+	RETURN
+
+#ifndef USE_AS_STRCAT
+
+	.p2align 4
+L(Fill0):
+	RETURN
+
+	.p2align 4
+L(Fill1):
+	mov	%dl, (%rdi)
+	RETURN
+
+	.p2align 4
+L(Fill2):
+	mov	%dx, (%rdi)
+	RETURN
+
+	.p2align 4
+L(Fill3):
+	mov	%edx, -1(%rdi)
+	RETURN
+
+	.p2align 4
+L(Fill4):
+	mov	%edx, (%rdi)
+	RETURN
+
+	.p2align 4
+L(Fill5):
+	mov	%edx, (%rdi)
+	mov	%dl, 4(%rdi)
+	RETURN
+
+	.p2align 4
+L(Fill6):
+	mov	%edx, (%rdi)
+	mov	%dx, 4(%rdi)
+	RETURN
+
+	.p2align 4
+L(Fill7):
+	mov	%rdx, -1(%rdi)
+	RETURN
+
+	.p2align 4
+L(Fill8):
+	mov	%rdx, (%rdi)
+	RETURN
+
+	.p2align 4
+L(Fill9):
+	mov	%rdx, (%rdi)
+	mov	%dl, 8(%rdi)
+	RETURN
+
+	.p2align 4
+L(Fill10):
+	mov	%rdx, (%rdi)
+	mov	%dx, 8(%rdi)
+	RETURN
+
+	.p2align 4
+L(Fill11):
+	mov	%rdx, (%rdi)
+	mov	%edx, 7(%rdi)
+	RETURN
+
+	.p2align 4
+L(Fill12):
+	mov	%rdx, (%rdi)
+	mov	%edx, 8(%rdi)
+	RETURN
+
+	.p2align 4
+L(Fill13):
+	mov	%rdx, (%rdi)
+	mov	%rdx, 5(%rdi)
+	RETURN
+
+	.p2align 4
+L(Fill14):
+	mov	%rdx, (%rdi)
+	mov	%rdx, 6(%rdi)
+	RETURN
+
+	.p2align 4
+L(Fill15):
+	movdqu	%xmm0, -1(%rdi)
+	RETURN
+
+	.p2align 4
+L(Fill16):
+	movdqu	%xmm0, (%rdi)
+	RETURN
+
+	.p2align 4
+L(CopyFrom1To16BytesUnalignedXmm2):
+	movdqu	%xmm2, (%rdi, %rcx)
+
+	.p2align 4
+L(CopyFrom1To16BytesXmmExit):
+	bsf	%rdx, %rdx
+	add	$15, %r8
+	add	%rcx, %rdi
+#ifdef USE_AS_STPCPY
+	lea	(%rdi, %rdx), %rax
+#endif
+	sub	%rdx, %r8
+	lea	1(%rdi, %rdx), %rdi
+
+	.p2align 4
+L(StrncpyFillTailWithZero):
+	pxor	%xmm0, %xmm0
+	xor	%rdx, %rdx
+	sub	$16, %r8
+	jbe	L(StrncpyFillExit)
+
+	movdqu	%xmm0, (%rdi)
+	add	$16, %rdi
+
+	mov	%rdi, %rsi
+	and	$0xf, %rsi
+	sub	%rsi, %rdi
+	add	%rsi, %r8
+	sub	$64, %r8
+	jb	L(StrncpyFillLess64)
+
+L(StrncpyFillLoopMovdqa):
+	movdqa	%xmm0, (%rdi)
+	movdqa	%xmm0, 16(%rdi)
+	movdqa	%xmm0, 32(%rdi)
+	movdqa	%xmm0, 48(%rdi)
+	add	$64, %rdi
+	sub	$64, %r8
+	jae	L(StrncpyFillLoopMovdqa)
+
+L(StrncpyFillLess64):
+	add	$32, %r8
+	jl	L(StrncpyFillLess32)
+	movdqa	%xmm0, (%rdi)
+	movdqa	%xmm0, 16(%rdi)
+	add	$32, %rdi
+	sub	$16, %r8
+	jl	L(StrncpyFillExit)
+	movdqa	%xmm0, (%rdi)
+	add	$16, %rdi
+	BRANCH_TO_JMPTBL_ENTRY (L(FillTable), %r8, 4)
+
+L(StrncpyFillLess32):
+	add	$16, %r8
+	jl	L(StrncpyFillExit)
+	movdqa	%xmm0, (%rdi)
+	add	$16, %rdi
+	BRANCH_TO_JMPTBL_ENTRY (L(FillTable), %r8, 4)
+
+L(StrncpyFillExit):
+	add	$16, %r8
+	BRANCH_TO_JMPTBL_ENTRY (L(FillTable), %r8, 4)
+
+/* end of ifndef USE_AS_STRCAT */
+#endif
+
+	.p2align 4
+L(UnalignedLeaveCase2OrCase3):
+	test	%rdx, %rdx
+	jnz	L(Unaligned64LeaveCase2)
+L(Unaligned64LeaveCase3):
+	lea	64(%r8), %rcx
+	and	$-16, %rcx
+	add	$48, %r8
+	jl	L(CopyFrom1To16BytesCase3)
+	movdqu	%xmm4, (%rdi)
+	sub	$16, %r8
+	jb	L(CopyFrom1To16BytesCase3)
+	movdqu	%xmm5, 16(%rdi)
+	sub	$16, %r8
+	jb	L(CopyFrom1To16BytesCase3)
+	movdqu	%xmm6, 32(%rdi)
+	sub	$16, %r8
+	jb	L(CopyFrom1To16BytesCase3)
+	movdqu	%xmm7, 48(%rdi)
+#ifdef USE_AS_STPCPY
+	lea	64(%rdi), %rax
+#endif
+#ifdef USE_AS_STRCAT
+	xor	%ch, %ch
+	movb	%ch, 64(%rdi)
+#endif
+	RETURN
+
+	.p2align 4
+L(Unaligned64LeaveCase2):
+	xor	%rcx, %rcx
+	pcmpeqb	%xmm4, %xmm0
+	pmovmskb %xmm0, %rdx
+	add	$48, %r8
+	jle	L(CopyFrom1To16BytesCase2OrCase3)
+	test	%rdx, %rdx
+#ifndef USE_AS_STRCAT
+	jnz	L(CopyFrom1To16BytesUnalignedXmm4)
+#else
+	jnz	L(CopyFrom1To16Bytes)
+#endif
+	pcmpeqb	%xmm5, %xmm0
+	pmovmskb %xmm0, %rdx
+	movdqu	%xmm4, (%rdi)
+	add	$16, %rcx
+	sub	$16, %r8
+	jbe	L(CopyFrom1To16BytesCase2OrCase3)
+	test	%rdx, %rdx
+#ifndef USE_AS_STRCAT
+	jnz	L(CopyFrom1To16BytesUnalignedXmm5)
+#else
+	jnz	L(CopyFrom1To16Bytes)
+#endif
+
+	pcmpeqb	%xmm6, %xmm0
+	pmovmskb %xmm0, %rdx
+	movdqu	%xmm5, 16(%rdi)
+	add	$16, %rcx
+	sub	$16, %r8
+	jbe	L(CopyFrom1To16BytesCase2OrCase3)
+	test	%rdx, %rdx
+#ifndef USE_AS_STRCAT
+	jnz	L(CopyFrom1To16BytesUnalignedXmm6)
+#else
+	jnz	L(CopyFrom1To16Bytes)
+#endif
+
+	pcmpeqb	%xmm7, %xmm0
+	pmovmskb %xmm0, %rdx
+	movdqu	%xmm6, 32(%rdi)
+	lea	16(%rdi, %rcx), %rdi
+	lea	16(%rsi, %rcx), %rsi
+	bsf	%rdx, %rdx
+	cmp	%r8, %rdx
+	jb	L(CopyFrom1To16BytesExit)
+	BRANCH_TO_JMPTBL_ENTRY (L(ExitStrncpyTable), %r8, 4)
+
+	.p2align 4
+L(ExitZero):
+#ifndef USE_AS_STRCAT
+	mov	%rdi, %rax
+#endif
+	RETURN
+
+#endif
+
+#ifndef USE_AS_STRCAT
+END (STRCPY)
+#else
+END (STRCAT)
+#endif
+	.p2align 4
+	.section .rodata
+L(ExitTable):
+	.int	JMPTBL(L(Exit1), L(ExitTable))
+	.int	JMPTBL(L(Exit2), L(ExitTable))
+	.int	JMPTBL(L(Exit3), L(ExitTable))
+	.int	JMPTBL(L(Exit4), L(ExitTable))
+	.int	JMPTBL(L(Exit5), L(ExitTable))
+	.int	JMPTBL(L(Exit6), L(ExitTable))
+	.int	JMPTBL(L(Exit7), L(ExitTable))
+	.int	JMPTBL(L(Exit8), L(ExitTable))
+	.int	JMPTBL(L(Exit9), L(ExitTable))
+	.int	JMPTBL(L(Exit10), L(ExitTable))
+	.int	JMPTBL(L(Exit11), L(ExitTable))
+	.int	JMPTBL(L(Exit12), L(ExitTable))
+	.int	JMPTBL(L(Exit13), L(ExitTable))
+	.int	JMPTBL(L(Exit14), L(ExitTable))
+	.int	JMPTBL(L(Exit15), L(ExitTable))
+	.int	JMPTBL(L(Exit16), L(ExitTable))
+	.int	JMPTBL(L(Exit17), L(ExitTable))
+	.int	JMPTBL(L(Exit18), L(ExitTable))
+	.int	JMPTBL(L(Exit19), L(ExitTable))
+	.int	JMPTBL(L(Exit20), L(ExitTable))
+	.int	JMPTBL(L(Exit21), L(ExitTable))
+	.int	JMPTBL(L(Exit22), L(ExitTable))
+	.int	JMPTBL(L(Exit23), L(ExitTable))
+	.int	JMPTBL(L(Exit24), L(ExitTable))
+	.int	JMPTBL(L(Exit25), L(ExitTable))
+	.int	JMPTBL(L(Exit26), L(ExitTable))
+	.int	JMPTBL(L(Exit27), L(ExitTable))
+	.int	JMPTBL(L(Exit28), L(ExitTable))
+	.int	JMPTBL(L(Exit29), L(ExitTable))
+	.int	JMPTBL(L(Exit30), L(ExitTable))
+	.int	JMPTBL(L(Exit31), L(ExitTable))
+	.int	JMPTBL(L(Exit32), L(ExitTable))
+#ifdef USE_AS_STRNCPY
+L(ExitStrncpyTable):
+	.int	JMPTBL(L(StrncpyExit0), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit1), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit2), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit3), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit4), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit5), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit6), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit7), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit8), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit9), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit10), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit11), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit12), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit13), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit14), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit15), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit16), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit17), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit18), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit19), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit20), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit21), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit22), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit23), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit24), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit25), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit26), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit27), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit28), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit29), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit30), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit31), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit32), L(ExitStrncpyTable))
+	.int	JMPTBL(L(StrncpyExit33), L(ExitStrncpyTable))
+# ifndef USE_AS_STRCAT
+	.p2align 4
+L(FillTable):
+	.int	JMPTBL(L(Fill0), L(FillTable))
+	.int	JMPTBL(L(Fill1), L(FillTable))
+	.int	JMPTBL(L(Fill2), L(FillTable))
+	.int	JMPTBL(L(Fill3), L(FillTable))
+	.int	JMPTBL(L(Fill4), L(FillTable))
+	.int	JMPTBL(L(Fill5), L(FillTable))
+	.int	JMPTBL(L(Fill6), L(FillTable))
+	.int	JMPTBL(L(Fill7), L(FillTable))
+	.int	JMPTBL(L(Fill8), L(FillTable))
+	.int	JMPTBL(L(Fill9), L(FillTable))
+	.int	JMPTBL(L(Fill10), L(FillTable))
+	.int	JMPTBL(L(Fill11), L(FillTable))
+	.int	JMPTBL(L(Fill12), L(FillTable))
+	.int	JMPTBL(L(Fill13), L(FillTable))
+	.int	JMPTBL(L(Fill14), L(FillTable))
+	.int	JMPTBL(L(Fill15), L(FillTable))
+	.int	JMPTBL(L(Fill16), L(FillTable))
+# endif
+#endif
diff --git a/libc/arch-x86_64/string/sse2-strlen-slm.S b/libc/arch-x86_64/string/sse2-strlen-slm.S
new file mode 100644
index 0000000..3772fe7
--- /dev/null
+++ b/libc/arch-x86_64/string/sse2-strlen-slm.S
@@ -0,0 +1,294 @@
+/*
+Copyright (c) 2014, Intel Corporation
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+    * Redistributions of source code must retain the above copyright notice,
+    * this list of conditions and the following disclaimer.
+
+    * Redistributions in binary form must reproduce the above copyright notice,
+    * this list of conditions and the following disclaimer in the documentation
+    * and/or other materials provided with the distribution.
+
+    * Neither the name of Intel Corporation nor the names of its contributors
+    * may be used to endorse or promote products derived from this software
+    * without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#ifndef USE_AS_STRCAT
+
+#ifndef STRLEN
+# define STRLEN		strlen
+#endif
+
+#ifndef L
+# define L(label)	.L##label
+#endif
+
+#ifndef cfi_startproc
+# define cfi_startproc			.cfi_startproc
+#endif
+
+#ifndef cfi_endproc
+# define cfi_endproc			.cfi_endproc
+#endif
+
+#ifndef ENTRY
+# define ENTRY(name)			\
+	.type name,  @function; 	\
+	.globl name;			\
+	.p2align 4;			\
+name:					\
+	cfi_startproc
+#endif
+
+#ifndef END
+# define END(name)			\
+	cfi_endproc;			\
+	.size name, .-name
+#endif
+#define RETURN ret
+	.section .text.sse2,"ax",@progbits
+ENTRY (STRLEN)
+/* end ifndef USE_AS_STRCAT */
+#endif
+	xor	%rax, %rax
+	mov	%edi, %ecx
+	and	$0x3f, %ecx
+	pxor	%xmm0, %xmm0
+	cmp	$0x30, %ecx
+	ja	L(next)
+	movdqu	(%rdi), %xmm1
+	pcmpeqb	%xmm1, %xmm0
+	pmovmskb %xmm0, %edx
+	test	%edx, %edx
+	jnz	L(exit_less16)
+	mov	%rdi, %rax
+	and	$-16, %rax
+	jmp	L(align16_start)
+L(next):
+	mov	%rdi, %rax
+	and	$-16, %rax
+	pcmpeqb	(%rax), %xmm0
+	mov	$-1, %r10d
+	sub	%rax, %rcx
+	shl	%cl, %r10d
+	pmovmskb %xmm0, %edx
+	and	%r10d, %edx
+	jnz	L(exit)
+L(align16_start):
+	pxor	%xmm0, %xmm0
+	pxor	%xmm1, %xmm1
+	pxor	%xmm2, %xmm2
+	pxor	%xmm3, %xmm3
+	pcmpeqb	16(%rax), %xmm0
+	pmovmskb %xmm0, %edx
+	test	%edx, %edx
+	jnz	L(exit16)
+
+	pcmpeqb	32(%rax), %xmm1
+	pmovmskb %xmm1, %edx
+	test	%edx, %edx
+	jnz	L(exit32)
+
+	pcmpeqb	48(%rax), %xmm2
+	pmovmskb %xmm2, %edx
+	test	%edx, %edx
+	jnz	L(exit48)
+
+	pcmpeqb	64(%rax), %xmm3
+	pmovmskb %xmm3, %edx
+	test	%edx, %edx
+	jnz	L(exit64)
+
+	pcmpeqb	80(%rax), %xmm0
+	add	$64, %rax
+	pmovmskb %xmm0, %edx
+	test	%edx, %edx
+	jnz	L(exit16)
+
+	pcmpeqb	32(%rax), %xmm1
+	pmovmskb %xmm1, %edx
+	test	%edx, %edx
+	jnz	L(exit32)
+
+	pcmpeqb	48(%rax), %xmm2
+	pmovmskb %xmm2, %edx
+	test	%edx, %edx
+	jnz	L(exit48)
+
+	pcmpeqb	64(%rax), %xmm3
+	pmovmskb %xmm3, %edx
+	test	%edx, %edx
+	jnz	L(exit64)
+
+	pcmpeqb	80(%rax), %xmm0
+	add	$64, %rax
+	pmovmskb %xmm0, %edx
+	test	%edx, %edx
+	jnz	L(exit16)
+
+	pcmpeqb	32(%rax), %xmm1
+	pmovmskb %xmm1, %edx
+	test	%edx, %edx
+	jnz	L(exit32)
+
+	pcmpeqb	48(%rax), %xmm2
+	pmovmskb %xmm2, %edx
+	test	%edx, %edx
+	jnz	L(exit48)
+
+	pcmpeqb	64(%rax), %xmm3
+	pmovmskb %xmm3, %edx
+	test	%edx, %edx
+	jnz	L(exit64)
+	
+	pcmpeqb	80(%rax), %xmm0
+	add	$64, %rax
+	pmovmskb %xmm0, %edx
+	test	%edx, %edx
+	jnz	L(exit16)
+
+	pcmpeqb	32(%rax), %xmm1
+	pmovmskb %xmm1, %edx
+	test	%edx, %edx
+	jnz	L(exit32)
+
+	pcmpeqb	48(%rax), %xmm2
+	pmovmskb %xmm2, %edx
+	test	%edx, %edx
+	jnz	L(exit48)
+
+	pcmpeqb	64(%rax), %xmm3
+	pmovmskb %xmm3, %edx
+	test	%edx, %edx
+	jnz	L(exit64)
+	
+
+	test	$0x3f, %rax
+	jz	L(align64_loop)
+
+	pcmpeqb	80(%rax), %xmm0
+	add	$80, %rax
+	pmovmskb %xmm0, %edx
+	test	%edx, %edx
+	jnz	L(exit)
+
+	test	$0x3f, %rax
+	jz	L(align64_loop)
+
+	pcmpeqb	16(%rax), %xmm1
+	add	$16, %rax
+	pmovmskb %xmm1, %edx
+	test	%edx, %edx
+	jnz	L(exit)
+
+	test	$0x3f, %rax
+	jz	L(align64_loop)
+
+	pcmpeqb	16(%rax), %xmm2
+	add	$16, %rax
+	pmovmskb %xmm2, %edx
+	test	%edx, %edx
+	jnz	L(exit)
+
+	test	$0x3f, %rax
+	jz	L(align64_loop)
+
+	pcmpeqb	16(%rax), %xmm3
+	add	$16, %rax
+	pmovmskb %xmm3, %edx
+	test	%edx, %edx
+	jnz	L(exit)
+
+	add	$16, %rax
+	.p2align 4
+	L(align64_loop):
+	movaps	(%rax),	%xmm4
+	pminub	16(%rax), 	%xmm4
+	movaps	32(%rax), 	%xmm5
+	pminub	48(%rax), 	%xmm5
+	add	$64, 	%rax
+	pminub	%xmm4,	%xmm5
+	pcmpeqb	%xmm0,	%xmm5
+	pmovmskb %xmm5,	%edx
+	test	%edx,	%edx
+	jz	L(align64_loop)
+
+
+	pcmpeqb	-64(%rax), %xmm0
+	sub	$80, 	%rax
+	pmovmskb %xmm0, %edx
+	test	%edx, %edx
+	jnz	L(exit16)
+
+	pcmpeqb	32(%rax), %xmm1
+	pmovmskb %xmm1, %edx
+	test	%edx, %edx
+	jnz	L(exit32)
+
+	pcmpeqb	48(%rax), %xmm2
+	pmovmskb %xmm2, %edx
+	test	%edx, %edx
+	jnz	L(exit48)
+
+	pcmpeqb	64(%rax), %xmm3
+	pmovmskb %xmm3, %edx
+	sub	%rdi, %rax
+	bsf	%rdx, %rdx
+	add	%rdx, %rax
+	add	$64, %rax
+	RETURN
+
+	.p2align 4
+L(exit):
+	sub	%rdi, %rax
+L(exit_less16):
+	bsf	%rdx, %rdx
+	add	%rdx, %rax
+	RETURN
+	.p2align 4
+L(exit16):
+	sub	%rdi, %rax
+	bsf	%rdx, %rdx
+	add	%rdx, %rax
+	add	$16, %rax
+	RETURN
+	.p2align 4
+L(exit32):
+	sub	%rdi, %rax
+	bsf	%rdx, %rdx
+	add	%rdx, %rax
+	add	$32, %rax
+	RETURN
+	.p2align 4
+L(exit48):
+	sub	%rdi, %rax
+	bsf	%rdx, %rdx
+	add	%rdx, %rax
+	add	$48, %rax
+	RETURN
+	.p2align 4
+L(exit64):
+	sub	%rdi, %rax
+	bsf	%rdx, %rdx
+	add	%rdx, %rax
+	add	$64, %rax
+#ifndef USE_AS_STRCAT
+	RETURN
+
+END (STRLEN)
+#endif
diff --git a/libc/arch-x86_64/string/sse2-strncat-slm.S b/libc/arch-x86_64/string/sse2-strncat-slm.S
new file mode 100644
index 0000000..6b4a430
--- /dev/null
+++ b/libc/arch-x86_64/string/sse2-strncat-slm.S
@@ -0,0 +1,33 @@
+/*
+Copyright (c) 2014, Intel Corporation
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+    * Redistributions of source code must retain the above copyright notice,
+    * this list of conditions and the following disclaimer.
+
+    * Redistributions in binary form must reproduce the above copyright notice,
+    * this list of conditions and the following disclaimer in the documentation
+    * and/or other materials provided with the distribution.
+
+    * Neither the name of Intel Corporation nor the names of its contributors
+    * may be used to endorse or promote products derived from this software
+    * without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#define USE_AS_STRNCAT
+#define STRCAT		strncat
+#include "sse2-strcat-slm.S"
diff --git a/libc/arch-x86_64/string/sse2-strncpy-slm.S b/libc/arch-x86_64/string/sse2-strncpy-slm.S
new file mode 100644
index 0000000..594e78f
--- /dev/null
+++ b/libc/arch-x86_64/string/sse2-strncpy-slm.S
@@ -0,0 +1,33 @@
+/*
+Copyright (c) 2014, Intel Corporation
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+    * Redistributions of source code must retain the above copyright notice,
+    * this list of conditions and the following disclaimer.
+
+    * Redistributions in binary form must reproduce the above copyright notice,
+    * this list of conditions and the following disclaimer in the documentation
+    * and/or other materials provided with the distribution.
+
+    * Neither the name of Intel Corporation nor the names of its contributors
+    * may be used to endorse or promote products derived from this software
+    * without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#define USE_AS_STRNCPY
+#define STRCPY		strncpy
+#include "sse2-strcpy-slm.S"
diff --git a/libc/arch-x86_64/string/sse4-memcmp-slm.S b/libc/arch-x86_64/string/sse4-memcmp-slm.S
new file mode 100644
index 0000000..8a8b180
--- /dev/null
+++ b/libc/arch-x86_64/string/sse4-memcmp-slm.S
@@ -0,0 +1,1799 @@
+/*
+Copyright (c) 2014, Intel Corporation
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+    * Redistributions of source code must retain the above copyright notice,
+    * this list of conditions and the following disclaimer.
+
+    * Redistributions in binary form must reproduce the above copyright notice,
+    * this list of conditions and the following disclaimer in the documentation
+    * and/or other materials provided with the distribution.
+
+    * Neither the name of Intel Corporation nor the names of its contributors
+    * may be used to endorse or promote products derived from this software
+    * without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "cache.h"
+
+#ifndef MEMCMP
+# define MEMCMP		memcmp
+#endif
+
+#ifndef L
+# define L(label)	.L##label
+#endif
+
+#ifndef ALIGN
+# define ALIGN(n)	.p2align n
+#endif
+
+#ifndef cfi_startproc
+# define cfi_startproc			.cfi_startproc
+#endif
+
+#ifndef cfi_endproc
+# define cfi_endproc			.cfi_endproc
+#endif
+
+#ifndef ENTRY
+# define ENTRY(name)			\
+	.type name,  @function; 	\
+	.globl name;			\
+	.p2align 4;			\
+name:					\
+	cfi_startproc
+#endif
+
+#ifndef END
+# define END(name)			\
+	cfi_endproc;			\
+	.size name, .-name
+#endif
+
+#ifndef ALIGN
+# define ALIGN(n)	.p2align n
+#endif
+
+#define JMPTBL(I, B)	(I - B)
+
+#define BRANCH_TO_JMPTBL_ENTRY(TABLE, INDEX, SCALE)		\
+  lea		TABLE(%rip), %r11;				\
+  movslq	(%r11, INDEX, SCALE), %rcx;			\
+  add		%r11, %rcx;					\
+  jmp		*%rcx;						\
+  ud2
+
+	.section .text.sse4.1,"ax",@progbits
+ENTRY (MEMCMP)
+#ifdef USE_AS_WMEMCMP
+	shl	$2, %rdx
+#endif
+	pxor	%xmm0, %xmm0
+	cmp	$79, %rdx
+	ja	L(79bytesormore)
+#ifndef USE_AS_WMEMCMP
+	cmp	$1, %rdx
+	je	L(firstbyte)
+#endif
+	add	%rdx, %rsi
+	add	%rdx, %rdi
+	BRANCH_TO_JMPTBL_ENTRY(L(table_64bytes), %rdx, 4)
+
+#ifndef USE_AS_WMEMCMP
+	ALIGN (4)
+L(firstbyte):
+	movzbl	(%rdi), %eax
+	movzbl	(%rsi), %ecx
+	sub	%ecx, %eax
+	ret
+#endif
+
+	ALIGN (4)
+L(79bytesormore):
+	movdqu	(%rsi), %xmm1
+	movdqu	(%rdi), %xmm2
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(16bytesin256)
+	mov	%rsi, %rcx
+	and	$-16, %rsi
+	add	$16, %rsi
+	sub	%rsi, %rcx
+
+	sub	%rcx, %rdi
+	add	%rcx, %rdx
+	test	$0xf, %rdi
+	jz	L(2aligned)
+
+	cmp	$128, %rdx
+	ja	L(128bytesormore)
+L(less128bytes):
+	sub	$64, %rdx
+
+	movdqu	(%rdi), %xmm2
+	pxor	(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(16bytesin256)
+
+	movdqu	16(%rdi), %xmm2
+	pxor	16(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(32bytesin256)
+
+	movdqu	32(%rdi), %xmm2
+	pxor	32(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(48bytesin256)
+
+	movdqu	48(%rdi), %xmm2
+	pxor	48(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(64bytesin256)
+	cmp	$32, %rdx
+	jb	L(less32bytesin64)
+
+	movdqu	64(%rdi), %xmm2
+	pxor	64(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(80bytesin256)
+
+	movdqu	80(%rdi), %xmm2
+	pxor	80(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(96bytesin256)
+	sub	$32, %rdx
+	add	$32, %rdi
+	add	$32, %rsi
+L(less32bytesin64):
+	add	$64, %rdi
+	add	$64, %rsi
+	add	%rdx, %rsi
+	add	%rdx, %rdi
+	BRANCH_TO_JMPTBL_ENTRY(L(table_64bytes), %rdx, 4)
+
+L(128bytesormore):
+	cmp	$512, %rdx
+	ja	L(512bytesormore)
+	cmp	$256, %rdx
+	ja	L(less512bytes)
+L(less256bytes):
+	sub	$128, %rdx
+
+	movdqu	(%rdi), %xmm2
+	pxor	(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(16bytesin256)
+
+	movdqu	16(%rdi), %xmm2
+	pxor	16(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(32bytesin256)
+
+	movdqu	32(%rdi), %xmm2
+	pxor	32(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(48bytesin256)
+
+	movdqu	48(%rdi), %xmm2
+	pxor	48(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(64bytesin256)
+
+	movdqu	64(%rdi), %xmm2
+	pxor	64(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(80bytesin256)
+
+	movdqu	80(%rdi), %xmm2
+	pxor	80(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(96bytesin256)
+
+	movdqu	96(%rdi), %xmm2
+	pxor	96(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(112bytesin256)
+
+	movdqu	112(%rdi), %xmm2
+	pxor	112(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(128bytesin256)
+
+	add	$128, %rsi
+	add	$128, %rdi
+
+	cmp	$64, %rdx
+	jae	L(less128bytes)
+
+	cmp	$32, %rdx
+	jb	L(less32bytesin128)
+
+	movdqu	(%rdi), %xmm2
+	pxor	(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(16bytesin256)
+
+	movdqu	16(%rdi), %xmm2
+	pxor	16(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(32bytesin256)
+	sub	$32, %rdx
+	add	$32, %rdi
+	add	$32, %rsi
+L(less32bytesin128):
+	add	%rdx, %rsi
+	add	%rdx, %rdi
+	BRANCH_TO_JMPTBL_ENTRY(L(table_64bytes), %rdx, 4)
+
+L(less512bytes):
+	sub	$256, %rdx
+	movdqu	(%rdi), %xmm2
+	pxor	(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(16bytesin256)
+
+	movdqu	16(%rdi), %xmm2
+	pxor	16(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(32bytesin256)
+
+	movdqu	32(%rdi), %xmm2
+	pxor	32(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(48bytesin256)
+
+	movdqu	48(%rdi), %xmm2
+	pxor	48(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(64bytesin256)
+
+	movdqu	64(%rdi), %xmm2
+	pxor	64(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(80bytesin256)
+
+	movdqu	80(%rdi), %xmm2
+	pxor	80(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(96bytesin256)
+
+	movdqu	96(%rdi), %xmm2
+	pxor	96(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(112bytesin256)
+
+	movdqu	112(%rdi), %xmm2
+	pxor	112(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(128bytesin256)
+
+	movdqu	128(%rdi), %xmm2
+	pxor	128(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(144bytesin256)
+
+	movdqu	144(%rdi), %xmm2
+	pxor	144(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(160bytesin256)
+
+	movdqu	160(%rdi), %xmm2
+	pxor	160(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(176bytesin256)
+
+	movdqu	176(%rdi), %xmm2
+	pxor	176(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(192bytesin256)
+
+	movdqu	192(%rdi), %xmm2
+	pxor	192(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(208bytesin256)
+
+	movdqu	208(%rdi), %xmm2
+	pxor	208(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(224bytesin256)
+
+	movdqu	224(%rdi), %xmm2
+	pxor	224(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(240bytesin256)
+
+	movdqu	240(%rdi), %xmm2
+	pxor	240(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(256bytesin256)
+
+	add	$256, %rsi
+	add	$256, %rdi
+
+	cmp	$128, %rdx
+	jae	L(less256bytes)
+
+	cmp	$64, %rdx
+	jae	L(less128bytes)
+
+	cmp	$32, %rdx
+	jb	L(less32bytesin256)
+
+	movdqu	(%rdi), %xmm2
+	pxor	(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(16bytesin256)
+
+	movdqu	16(%rdi), %xmm2
+	pxor	16(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(32bytesin256)
+	sub	$32, %rdx
+	add	$32, %rdi
+	add	$32, %rsi
+L(less32bytesin256):
+	add	%rdx, %rsi
+	add	%rdx, %rdi
+	BRANCH_TO_JMPTBL_ENTRY(L(table_64bytes), %rdx, 4)
+
+	ALIGN (4)
+L(512bytesormore):
+#ifdef DATA_CACHE_SIZE_HALF
+	mov	$DATA_CACHE_SIZE_HALF, %r8
+#else
+	mov	__x86_64_data_cache_size_half(%rip), %r8
+#endif
+	mov	%r8, %r9
+	shr	$1, %r8
+	add	%r9, %r8
+	cmp	%r8, %rdx
+	ja	L(L2_L3_cache_unaglined)
+	sub	$64, %rdx
+	ALIGN (4)
+L(64bytesormore_loop):
+	movdqu	(%rdi), %xmm2
+	pxor	(%rsi), %xmm2
+	movdqa	%xmm2, %xmm1
+
+	movdqu	16(%rdi), %xmm3
+	pxor	16(%rsi), %xmm3
+	por	%xmm3, %xmm1
+
+	movdqu	32(%rdi), %xmm4
+	pxor	32(%rsi), %xmm4
+	por	%xmm4, %xmm1
+
+	movdqu	48(%rdi), %xmm5
+	pxor	48(%rsi), %xmm5
+	por	%xmm5, %xmm1
+
+	ptest	%xmm1, %xmm0
+	jnc	L(64bytesormore_loop_end)
+	add	$64, %rsi
+	add	$64, %rdi
+	sub	$64, %rdx
+	jae	L(64bytesormore_loop)
+
+	add	$64, %rdx
+	add	%rdx, %rsi
+	add	%rdx, %rdi
+	BRANCH_TO_JMPTBL_ENTRY(L(table_64bytes), %rdx, 4)
+
+L(L2_L3_cache_unaglined):
+	sub	$64, %rdx
+	ALIGN (4)
+L(L2_L3_unaligned_128bytes_loop):
+	prefetchnta 0x1c0(%rdi)
+	prefetchnta 0x1c0(%rsi)
+	movdqu	(%rdi), %xmm2
+	pxor	(%rsi), %xmm2
+	movdqa	%xmm2, %xmm1
+
+	movdqu	16(%rdi), %xmm3
+	pxor	16(%rsi), %xmm3
+	por	%xmm3, %xmm1
+
+	movdqu	32(%rdi), %xmm4
+	pxor	32(%rsi), %xmm4
+	por	%xmm4, %xmm1
+
+	movdqu	48(%rdi), %xmm5
+	pxor	48(%rsi), %xmm5
+	por	%xmm5, %xmm1
+
+	ptest	%xmm1, %xmm0
+	jnc	L(64bytesormore_loop_end)
+	add	$64, %rsi
+	add	$64, %rdi
+	sub	$64, %rdx
+	jae	L(L2_L3_unaligned_128bytes_loop)
+
+	add	$64, %rdx
+	add	%rdx, %rsi
+	add	%rdx, %rdi
+	BRANCH_TO_JMPTBL_ENTRY(L(table_64bytes), %rdx, 4)
+
+/*
+ * This case is for machines which are sensitive for unaligned instructions.
+ */
+	ALIGN (4)
+L(2aligned):
+	cmp	$128, %rdx
+	ja	L(128bytesormorein2aligned)
+L(less128bytesin2aligned):
+	sub	$64, %rdx
+
+	movdqa	(%rdi), %xmm2
+	pxor	(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(16bytesin256)
+
+	movdqa	16(%rdi), %xmm2
+	pxor	16(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(32bytesin256)
+
+	movdqa	32(%rdi), %xmm2
+	pxor	32(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(48bytesin256)
+
+	movdqa	48(%rdi), %xmm2
+	pxor	48(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(64bytesin256)
+	cmp	$32, %rdx
+	jb	L(less32bytesin64in2alinged)
+
+	movdqa	64(%rdi), %xmm2
+	pxor	64(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(80bytesin256)
+
+	movdqa	80(%rdi), %xmm2
+	pxor	80(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(96bytesin256)
+	sub	$32, %rdx
+	add	$32, %rdi
+	add	$32, %rsi
+L(less32bytesin64in2alinged):
+	add	$64, %rdi
+	add	$64, %rsi
+	add	%rdx, %rsi
+	add	%rdx, %rdi
+	BRANCH_TO_JMPTBL_ENTRY(L(table_64bytes), %rdx, 4)
+
+	ALIGN (4)
+L(128bytesormorein2aligned):
+	cmp	$512, %rdx
+	ja	L(512bytesormorein2aligned)
+	cmp	$256, %rdx
+	ja	L(256bytesormorein2aligned)
+L(less256bytesin2alinged):
+	sub	$128, %rdx
+
+	movdqa	(%rdi), %xmm2
+	pxor	(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(16bytesin256)
+
+	movdqa	16(%rdi), %xmm2
+	pxor	16(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(32bytesin256)
+
+	movdqa	32(%rdi), %xmm2
+	pxor	32(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(48bytesin256)
+
+	movdqa	48(%rdi), %xmm2
+	pxor	48(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(64bytesin256)
+
+	movdqa	64(%rdi), %xmm2
+	pxor	64(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(80bytesin256)
+
+	movdqa	80(%rdi), %xmm2
+	pxor	80(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(96bytesin256)
+
+	movdqa	96(%rdi), %xmm2
+	pxor	96(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(112bytesin256)
+
+	movdqa	112(%rdi), %xmm2
+	pxor	112(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(128bytesin256)
+
+	add	$128, %rsi
+	add	$128, %rdi
+
+	cmp	$64, %rdx
+	jae	L(less128bytesin2aligned)
+
+	cmp	$32, %rdx
+	jb	L(less32bytesin128in2aligned)
+
+	movdqu	(%rdi), %xmm2
+	pxor	(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(16bytesin256)
+
+	movdqu	16(%rdi), %xmm2
+	pxor	16(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(32bytesin256)
+	sub	$32, %rdx
+	add	$32, %rdi
+	add	$32, %rsi
+L(less32bytesin128in2aligned):
+	add	%rdx, %rsi
+	add	%rdx, %rdi
+	BRANCH_TO_JMPTBL_ENTRY(L(table_64bytes), %rdx, 4)
+
+	ALIGN (4)
+L(256bytesormorein2aligned):
+
+	sub	$256, %rdx
+	movdqa	(%rdi), %xmm2
+	pxor	(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(16bytesin256)
+
+	movdqa	16(%rdi), %xmm2
+	pxor	16(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(32bytesin256)
+
+	movdqa	32(%rdi), %xmm2
+	pxor	32(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(48bytesin256)
+
+	movdqa	48(%rdi), %xmm2
+	pxor	48(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(64bytesin256)
+
+	movdqa	64(%rdi), %xmm2
+	pxor	64(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(80bytesin256)
+
+	movdqa	80(%rdi), %xmm2
+	pxor	80(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(96bytesin256)
+
+	movdqa	96(%rdi), %xmm2
+	pxor	96(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(112bytesin256)
+
+	movdqa	112(%rdi), %xmm2
+	pxor	112(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(128bytesin256)
+
+	movdqa	128(%rdi), %xmm2
+	pxor	128(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(144bytesin256)
+
+	movdqa	144(%rdi), %xmm2
+	pxor	144(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(160bytesin256)
+
+	movdqa	160(%rdi), %xmm2
+	pxor	160(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(176bytesin256)
+
+	movdqa	176(%rdi), %xmm2
+	pxor	176(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(192bytesin256)
+
+	movdqa	192(%rdi), %xmm2
+	pxor	192(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(208bytesin256)
+
+	movdqa	208(%rdi), %xmm2
+	pxor	208(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(224bytesin256)
+
+	movdqa	224(%rdi), %xmm2
+	pxor	224(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(240bytesin256)
+
+	movdqa	240(%rdi), %xmm2
+	pxor	240(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(256bytesin256)
+
+	add	$256, %rsi
+	add	$256, %rdi
+
+	cmp	$128, %rdx
+	jae	L(less256bytesin2alinged)
+
+	cmp	$64, %rdx
+	jae	L(less128bytesin2aligned)
+
+	cmp	$32, %rdx
+	jb	L(less32bytesin256in2alinged)
+
+	movdqa	(%rdi), %xmm2
+	pxor	(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(16bytesin256)
+
+	movdqa	16(%rdi), %xmm2
+	pxor	16(%rsi), %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(32bytesin256)
+	sub	$32, %rdx
+	add	$32, %rdi
+	add	$32, %rsi
+L(less32bytesin256in2alinged):
+	add	%rdx, %rsi
+	add	%rdx, %rdi
+	BRANCH_TO_JMPTBL_ENTRY(L(table_64bytes), %rdx, 4)
+
+	ALIGN (4)
+L(512bytesormorein2aligned):
+#ifdef DATA_CACHE_SIZE_HALF
+	mov	$DATA_CACHE_SIZE_HALF, %r8
+#else
+	mov	__x86_64_data_cache_size_half(%rip), %r8
+#endif
+	mov	%r8, %r9
+	shr	$1, %r8
+	add	%r9, %r8
+	cmp	%r8, %rdx
+	ja	L(L2_L3_cache_aglined)
+
+	sub	$64, %rdx
+	ALIGN (4)
+L(64bytesormore_loopin2aligned):
+	movdqa	(%rdi), %xmm2
+	pxor	(%rsi), %xmm2
+	movdqa	%xmm2, %xmm1
+
+	movdqa	16(%rdi), %xmm3
+	pxor	16(%rsi), %xmm3
+	por	%xmm3, %xmm1
+
+	movdqa	32(%rdi), %xmm4
+	pxor	32(%rsi), %xmm4
+	por	%xmm4, %xmm1
+
+	movdqa	48(%rdi), %xmm5
+	pxor	48(%rsi), %xmm5
+	por	%xmm5, %xmm1
+
+	ptest	%xmm1, %xmm0
+	jnc	L(64bytesormore_loop_end)
+	add	$64, %rsi
+	add	$64, %rdi
+	sub	$64, %rdx
+	jae	L(64bytesormore_loopin2aligned)
+
+	add	$64, %rdx
+	add	%rdx, %rsi
+	add	%rdx, %rdi
+	BRANCH_TO_JMPTBL_ENTRY(L(table_64bytes), %rdx, 4)
+L(L2_L3_cache_aglined):
+	sub	$64, %rdx
+	ALIGN (4)
+L(L2_L3_aligned_128bytes_loop):
+	prefetchnta 0x1c0(%rdi)
+	prefetchnta 0x1c0(%rsi)
+	movdqa	(%rdi), %xmm2
+	pxor	(%rsi), %xmm2
+	movdqa	%xmm2, %xmm1
+
+	movdqa	16(%rdi), %xmm3
+	pxor	16(%rsi), %xmm3
+	por	%xmm3, %xmm1
+
+	movdqa	32(%rdi), %xmm4
+	pxor	32(%rsi), %xmm4
+	por	%xmm4, %xmm1
+
+	movdqa	48(%rdi), %xmm5
+	pxor	48(%rsi), %xmm5
+	por	%xmm5, %xmm1
+
+	ptest	%xmm1, %xmm0
+	jnc	L(64bytesormore_loop_end)
+	add	$64, %rsi
+	add	$64, %rdi
+	sub	$64, %rdx
+	jae	L(L2_L3_aligned_128bytes_loop)
+
+	add	$64, %rdx
+	add	%rdx, %rsi
+	add	%rdx, %rdi
+	BRANCH_TO_JMPTBL_ENTRY(L(table_64bytes), %rdx, 4)
+
+
+	ALIGN (4)
+L(64bytesormore_loop_end):
+	add	$16, %rdi
+	add	$16, %rsi
+	ptest	%xmm2, %xmm0
+	jnc	L(16bytes)
+
+	add	$16, %rdi
+	add	$16, %rsi
+	ptest	%xmm3, %xmm0
+	jnc	L(16bytes)
+
+	add	$16, %rdi
+	add	$16, %rsi
+	ptest	%xmm4, %xmm0
+	jnc	L(16bytes)
+
+	add	$16, %rdi
+	add	$16, %rsi
+	jmp	L(16bytes)
+
+L(256bytesin256):
+	add	$256, %rdi
+	add	$256, %rsi
+	jmp	L(16bytes)
+L(240bytesin256):
+	add	$240, %rdi
+	add	$240, %rsi
+	jmp	L(16bytes)
+L(224bytesin256):
+	add	$224, %rdi
+	add	$224, %rsi
+	jmp	L(16bytes)
+L(208bytesin256):
+	add	$208, %rdi
+	add	$208, %rsi
+	jmp	L(16bytes)
+L(192bytesin256):
+	add	$192, %rdi
+	add	$192, %rsi
+	jmp	L(16bytes)
+L(176bytesin256):
+	add	$176, %rdi
+	add	$176, %rsi
+	jmp	L(16bytes)
+L(160bytesin256):
+	add	$160, %rdi
+	add	$160, %rsi
+	jmp	L(16bytes)
+L(144bytesin256):
+	add	$144, %rdi
+	add	$144, %rsi
+	jmp	L(16bytes)
+L(128bytesin256):
+	add	$128, %rdi
+	add	$128, %rsi
+	jmp	L(16bytes)
+L(112bytesin256):
+	add	$112, %rdi
+	add	$112, %rsi
+	jmp	L(16bytes)
+L(96bytesin256):
+	add	$96, %rdi
+	add	$96, %rsi
+	jmp	L(16bytes)
+L(80bytesin256):
+	add	$80, %rdi
+	add	$80, %rsi
+	jmp	L(16bytes)
+L(64bytesin256):
+	add	$64, %rdi
+	add	$64, %rsi
+	jmp	L(16bytes)
+L(48bytesin256):
+	add	$16, %rdi
+	add	$16, %rsi
+L(32bytesin256):
+	add	$16, %rdi
+	add	$16, %rsi
+L(16bytesin256):
+	add	$16, %rdi
+	add	$16, %rsi
+L(16bytes):
+	mov	-16(%rdi), %rax
+	mov	-16(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+L(8bytes):
+	mov	-8(%rdi), %rax
+	mov	-8(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	xor	%eax, %eax
+	ret
+
+	ALIGN (4)
+L(12bytes):
+	mov	-12(%rdi), %rax
+	mov	-12(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+L(4bytes):
+	mov	-4(%rsi), %ecx
+	mov	-4(%rdi), %eax
+	cmp	%eax, %ecx
+	jne	L(diffin4bytes)
+L(0bytes):
+	xor	%eax, %eax
+	ret
+
+#ifndef USE_AS_WMEMCMP
+/* unreal case for wmemcmp */
+	ALIGN (4)
+L(65bytes):
+	movdqu	-65(%rdi), %xmm1
+	movdqu	-65(%rsi), %xmm2
+	mov	$-65, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(49bytes):
+	movdqu	-49(%rdi), %xmm1
+	movdqu	-49(%rsi), %xmm2
+	mov	$-49, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(33bytes):
+	movdqu	-33(%rdi), %xmm1
+	movdqu	-33(%rsi), %xmm2
+	mov	$-33, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(17bytes):
+	mov	-17(%rdi), %rax
+	mov	-17(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+L(9bytes):
+	mov	-9(%rdi), %rax
+	mov	-9(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	movzbl	-1(%rdi), %eax
+	movzbl	-1(%rsi), %edx
+	sub	%edx, %eax
+	ret
+
+	ALIGN (4)
+L(13bytes):
+	mov	-13(%rdi), %rax
+	mov	-13(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	mov	-8(%rdi), %rax
+	mov	-8(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	xor	%eax, %eax
+	ret
+
+	ALIGN (4)
+L(5bytes):
+	mov	-5(%rdi), %eax
+	mov	-5(%rsi), %ecx
+	cmp	%eax, %ecx
+	jne	L(diffin4bytes)
+	movzbl	-1(%rdi), %eax
+	movzbl	-1(%rsi), %edx
+	sub	%edx, %eax
+	ret
+
+	ALIGN (4)
+L(66bytes):
+	movdqu	-66(%rdi), %xmm1
+	movdqu	-66(%rsi), %xmm2
+	mov	$-66, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(50bytes):
+	movdqu	-50(%rdi), %xmm1
+	movdqu	-50(%rsi), %xmm2
+	mov	$-50, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(34bytes):
+	movdqu	-34(%rdi), %xmm1
+	movdqu	-34(%rsi), %xmm2
+	mov	$-34, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(18bytes):
+	mov	-18(%rdi), %rax
+	mov	-18(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+L(10bytes):
+	mov	-10(%rdi), %rax
+	mov	-10(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	movzwl	-2(%rdi), %eax
+	movzwl	-2(%rsi), %ecx
+	cmp	%cl, %al
+	jne	L(end)
+	and	$0xffff, %eax
+	and	$0xffff, %ecx
+	sub	%ecx, %eax
+	ret
+
+	ALIGN (4)
+L(14bytes):
+	mov	-14(%rdi), %rax
+	mov	-14(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	mov	-8(%rdi), %rax
+	mov	-8(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	xor	%eax, %eax
+	ret
+
+	ALIGN (4)
+L(6bytes):
+	mov	-6(%rdi), %eax
+	mov	-6(%rsi), %ecx
+	cmp	%eax, %ecx
+	jne	L(diffin4bytes)
+L(2bytes):
+	movzwl	-2(%rsi), %ecx
+	movzwl	-2(%rdi), %eax
+	cmp	%cl, %al
+	jne	L(end)
+	and	$0xffff, %eax
+	and	$0xffff, %ecx
+	sub	%ecx, %eax
+	ret
+
+	ALIGN (4)
+L(67bytes):
+	movdqu	-67(%rdi), %xmm2
+	movdqu	-67(%rsi), %xmm1
+	mov	$-67, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(51bytes):
+	movdqu	-51(%rdi), %xmm2
+	movdqu	-51(%rsi), %xmm1
+	mov	$-51, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(35bytes):
+	movdqu	-35(%rsi), %xmm1
+	movdqu	-35(%rdi), %xmm2
+	mov	$-35, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(19bytes):
+	mov	-19(%rdi), %rax
+	mov	-19(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+L(11bytes):
+	mov	-11(%rdi), %rax
+	mov	-11(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	mov	-4(%rdi), %eax
+	mov	-4(%rsi), %ecx
+	cmp	%eax, %ecx
+	jne	L(diffin4bytes)
+	xor	%eax, %eax
+	ret
+
+	ALIGN (4)
+L(15bytes):
+	mov	-15(%rdi), %rax
+	mov	-15(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	mov	-8(%rdi), %rax
+	mov	-8(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	xor	%eax, %eax
+	ret
+
+	ALIGN (4)
+L(7bytes):
+	mov	-7(%rdi), %eax
+	mov	-7(%rsi), %ecx
+	cmp	%eax, %ecx
+	jne	L(diffin4bytes)
+	mov	-4(%rdi), %eax
+	mov	-4(%rsi), %ecx
+	cmp	%eax, %ecx
+	jne	L(diffin4bytes)
+	xor	%eax, %eax
+	ret
+
+	ALIGN (4)
+L(3bytes):
+	movzwl	-3(%rdi), %eax
+	movzwl	-3(%rsi), %ecx
+	cmp	%eax, %ecx
+	jne	L(diffin2bytes)
+L(1bytes):
+	movzbl	-1(%rdi), %eax
+	movzbl	-1(%rsi), %ecx
+	sub	%ecx, %eax
+	ret
+#endif
+
+	ALIGN (4)
+L(68bytes):
+	movdqu	-68(%rdi), %xmm2
+	movdqu	-68(%rsi), %xmm1
+	mov	$-68, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(52bytes):
+	movdqu	-52(%rdi), %xmm2
+	movdqu	-52(%rsi), %xmm1
+	mov	$-52, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(36bytes):
+	movdqu	-36(%rdi), %xmm2
+	movdqu	-36(%rsi), %xmm1
+	mov	$-36, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(20bytes):
+	movdqu	-20(%rdi), %xmm2
+	movdqu	-20(%rsi), %xmm1
+	mov	$-20, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+	mov	-4(%rdi), %eax
+	mov	-4(%rsi), %ecx
+	cmp	%eax, %ecx
+	jne	L(diffin4bytes)
+	xor	%eax, %eax
+	ret
+
+#ifndef USE_AS_WMEMCMP
+/* unreal cases for wmemcmp */
+	ALIGN (4)
+L(69bytes):
+	movdqu	-69(%rsi), %xmm1
+	movdqu	-69(%rdi), %xmm2
+	mov	$-69, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(53bytes):
+	movdqu	-53(%rsi), %xmm1
+	movdqu	-53(%rdi), %xmm2
+	mov	$-53, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(37bytes):
+	movdqu	-37(%rsi), %xmm1
+	movdqu	-37(%rdi), %xmm2
+	mov	$-37, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(21bytes):
+	movdqu	-21(%rsi), %xmm1
+	movdqu	-21(%rdi), %xmm2
+	mov	$-21, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+	mov	-8(%rdi), %rax
+	mov	-8(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	xor	%eax, %eax
+	ret
+
+	ALIGN (4)
+L(70bytes):
+	movdqu	-70(%rsi), %xmm1
+	movdqu	-70(%rdi), %xmm2
+	mov	$-70, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(54bytes):
+	movdqu	-54(%rsi), %xmm1
+	movdqu	-54(%rdi), %xmm2
+	mov	$-54, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(38bytes):
+	movdqu	-38(%rsi), %xmm1
+	movdqu	-38(%rdi), %xmm2
+	mov	$-38, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(22bytes):
+	movdqu	-22(%rsi), %xmm1
+	movdqu	-22(%rdi), %xmm2
+	mov	$-22, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+	mov	-8(%rdi), %rax
+	mov	-8(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	xor	%eax, %eax
+	ret
+
+	ALIGN (4)
+L(71bytes):
+	movdqu	-71(%rsi), %xmm1
+	movdqu	-71(%rdi), %xmm2
+	mov	$-71, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(55bytes):
+	movdqu	-55(%rdi), %xmm2
+	movdqu	-55(%rsi), %xmm1
+	mov	$-55, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(39bytes):
+	movdqu	-39(%rdi), %xmm2
+	movdqu	-39(%rsi), %xmm1
+	mov	$-39, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(23bytes):
+	movdqu	-23(%rdi), %xmm2
+	movdqu	-23(%rsi), %xmm1
+	mov	$-23, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+	mov	-8(%rdi), %rax
+	mov	-8(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	xor	%eax, %eax
+	ret
+#endif
+
+	ALIGN (4)
+L(72bytes):
+	movdqu	-72(%rsi), %xmm1
+	movdqu	-72(%rdi), %xmm2
+	mov	$-72, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(56bytes):
+	movdqu	-56(%rdi), %xmm2
+	movdqu	-56(%rsi), %xmm1
+	mov	$-56, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(40bytes):
+	movdqu	-40(%rdi), %xmm2
+	movdqu	-40(%rsi), %xmm1
+	mov	$-40, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(24bytes):
+	movdqu	-24(%rdi), %xmm2
+	movdqu	-24(%rsi), %xmm1
+	mov	$-24, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+	mov	-8(%rdi), %rax
+	mov	-8(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	xor	%eax, %eax
+	ret
+
+#ifndef USE_AS_WMEMCMP
+/* unreal cases for wmemcmp */
+	ALIGN (4)
+L(73bytes):
+	movdqu	-73(%rsi), %xmm1
+	movdqu	-73(%rdi), %xmm2
+	mov	$-73, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(57bytes):
+	movdqu	-57(%rdi), %xmm2
+	movdqu	-57(%rsi), %xmm1
+	mov	$-57, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(41bytes):
+	movdqu	-41(%rdi), %xmm2
+	movdqu	-41(%rsi), %xmm1
+	mov	$-41, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(25bytes):
+	movdqu	-25(%rdi), %xmm2
+	movdqu	-25(%rsi), %xmm1
+	mov	$-25, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+	mov	-9(%rdi), %rax
+	mov	-9(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	movzbl	-1(%rdi), %eax
+	movzbl	-1(%rsi), %ecx
+	sub	%ecx, %eax
+	ret
+
+	ALIGN (4)
+L(74bytes):
+	movdqu	-74(%rsi), %xmm1
+	movdqu	-74(%rdi), %xmm2
+	mov	$-74, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(58bytes):
+	movdqu	-58(%rdi), %xmm2
+	movdqu	-58(%rsi), %xmm1
+	mov	$-58, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(42bytes):
+	movdqu	-42(%rdi), %xmm2
+	movdqu	-42(%rsi), %xmm1
+	mov	$-42, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(26bytes):
+	movdqu	-26(%rdi), %xmm2
+	movdqu	-26(%rsi), %xmm1
+	mov	$-26, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+	mov	-10(%rdi), %rax
+	mov	-10(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	movzwl	-2(%rdi), %eax
+	movzwl	-2(%rsi), %ecx
+	jmp	L(diffin2bytes)
+
+	ALIGN (4)
+L(75bytes):
+	movdqu	-75(%rsi), %xmm1
+	movdqu	-75(%rdi), %xmm2
+	mov	$-75, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(59bytes):
+	movdqu	-59(%rdi), %xmm2
+	movdqu	-59(%rsi), %xmm1
+	mov	$-59, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(43bytes):
+	movdqu	-43(%rdi), %xmm2
+	movdqu	-43(%rsi), %xmm1
+	mov	$-43, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(27bytes):
+	movdqu	-27(%rdi), %xmm2
+	movdqu	-27(%rsi), %xmm1
+	mov	$-27, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+	mov	-11(%rdi), %rax
+	mov	-11(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	mov	-4(%rdi), %eax
+	mov	-4(%rsi), %ecx
+	cmp	%eax, %ecx
+	jne	L(diffin4bytes)
+	xor	%eax, %eax
+	ret
+#endif
+	ALIGN (4)
+L(76bytes):
+	movdqu	-76(%rsi), %xmm1
+	movdqu	-76(%rdi), %xmm2
+	mov	$-76, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(60bytes):
+	movdqu	-60(%rdi), %xmm2
+	movdqu	-60(%rsi), %xmm1
+	mov	$-60, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(44bytes):
+	movdqu	-44(%rdi), %xmm2
+	movdqu	-44(%rsi), %xmm1
+	mov	$-44, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(28bytes):
+	movdqu	-28(%rdi), %xmm2
+	movdqu	-28(%rsi), %xmm1
+	mov	$-28, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+	mov	-12(%rdi), %rax
+	mov	-12(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	mov	-4(%rdi), %eax
+	mov	-4(%rsi), %ecx
+	cmp	%eax, %ecx
+	jne	L(diffin4bytes)
+	xor	%eax, %eax
+	ret
+
+#ifndef USE_AS_WMEMCMP
+/* unreal cases for wmemcmp */
+	ALIGN (4)
+L(77bytes):
+	movdqu	-77(%rsi), %xmm1
+	movdqu	-77(%rdi), %xmm2
+	mov	$-77, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(61bytes):
+	movdqu	-61(%rdi), %xmm2
+	movdqu	-61(%rsi), %xmm1
+	mov	$-61, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(45bytes):
+	movdqu	-45(%rdi), %xmm2
+	movdqu	-45(%rsi), %xmm1
+	mov	$-45, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(29bytes):
+	movdqu	-29(%rdi), %xmm2
+	movdqu	-29(%rsi), %xmm1
+	mov	$-29, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+
+	mov	-13(%rdi), %rax
+	mov	-13(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+
+	mov	-8(%rdi), %rax
+	mov	-8(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	xor	%eax, %eax
+	ret
+
+	ALIGN (4)
+L(78bytes):
+	movdqu	-78(%rsi), %xmm1
+	movdqu	-78(%rdi), %xmm2
+	mov	$-78, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(62bytes):
+	movdqu	-62(%rdi), %xmm2
+	movdqu	-62(%rsi), %xmm1
+	mov	$-62, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(46bytes):
+	movdqu	-46(%rdi), %xmm2
+	movdqu	-46(%rsi), %xmm1
+	mov	$-46, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(30bytes):
+	movdqu	-30(%rdi), %xmm2
+	movdqu	-30(%rsi), %xmm1
+	mov	$-30, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+	mov	-14(%rdi), %rax
+	mov	-14(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	mov	-8(%rdi), %rax
+	mov	-8(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	xor	%eax, %eax
+	ret
+
+	ALIGN (4)
+L(79bytes):
+	movdqu	-79(%rsi), %xmm1
+	movdqu	-79(%rdi), %xmm2
+	mov	$-79, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(63bytes):
+	movdqu	-63(%rdi), %xmm2
+	movdqu	-63(%rsi), %xmm1
+	mov	$-63, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(47bytes):
+	movdqu	-47(%rdi), %xmm2
+	movdqu	-47(%rsi), %xmm1
+	mov	$-47, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(31bytes):
+	movdqu	-31(%rdi), %xmm2
+	movdqu	-31(%rsi), %xmm1
+	mov	$-31, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+	mov	-15(%rdi), %rax
+	mov	-15(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	mov	-8(%rdi), %rax
+	mov	-8(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	xor	%eax, %eax
+	ret
+#endif
+	ALIGN (4)
+L(64bytes):
+	movdqu	-64(%rdi), %xmm2
+	movdqu	-64(%rsi), %xmm1
+	mov	$-64, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(48bytes):
+	movdqu	-48(%rdi), %xmm2
+	movdqu	-48(%rsi), %xmm1
+	mov	$-48, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+L(32bytes):
+	movdqu	-32(%rdi), %xmm2
+	movdqu	-32(%rsi), %xmm1
+	mov	$-32, %dl
+	pxor	%xmm1, %xmm2
+	ptest	%xmm2, %xmm0
+	jnc	L(less16bytes)
+
+	mov	-16(%rdi), %rax
+	mov	-16(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+
+	mov	-8(%rdi), %rax
+	mov	-8(%rsi), %rcx
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	xor	%eax, %eax
+	ret
+
+/*
+ * Aligned 8 bytes to avoid 2 branch "taken" in one 16 alinged code block.
+ */
+	ALIGN (3)
+L(less16bytes):
+	movsbq	%dl, %rdx
+	mov	(%rsi, %rdx), %rcx
+	mov	(%rdi, %rdx), %rax
+	cmp	%rax, %rcx
+	jne	L(diffin8bytes)
+	mov	8(%rsi, %rdx), %rcx
+	mov	8(%rdi, %rdx), %rax
+L(diffin8bytes):
+	cmp	%eax, %ecx
+	jne	L(diffin4bytes)
+	shr	$32, %rcx
+	shr	$32, %rax
+
+#ifdef USE_AS_WMEMCMP
+/* for wmemcmp */
+	cmp	%eax, %ecx
+	jne	L(diffin4bytes)
+	xor	%eax, %eax
+	ret
+#endif
+
+L(diffin4bytes):
+#ifndef USE_AS_WMEMCMP
+	cmp	%cx, %ax
+	jne	L(diffin2bytes)
+	shr	$16, %ecx
+	shr	$16, %eax
+L(diffin2bytes):
+	cmp	%cl, %al
+	jne	L(end)
+	and	$0xffff, %eax
+	and	$0xffff, %ecx
+	sub	%ecx, %eax
+	ret
+#else
+
+/* for wmemcmp */
+	mov	$1, %eax
+	jl	L(nequal_bigger)
+	neg	%eax
+	ret
+
+	ALIGN (4)
+L(nequal_bigger):
+	ret
+
+L(unreal_case):
+	xor	%eax, %eax
+	ret
+#endif
+
+	ALIGN (4)
+L(end):
+	and	$0xff, %eax
+	and	$0xff, %ecx
+	sub	%ecx, %eax
+	ret
+
+END (MEMCMP)
+
+	.section .rodata.sse4.1,"a",@progbits
+	ALIGN (3)
+#ifndef USE_AS_WMEMCMP
+L(table_64bytes):
+	.int	JMPTBL (L(0bytes), L(table_64bytes))
+	.int	JMPTBL (L(1bytes), L(table_64bytes))
+	.int	JMPTBL (L(2bytes), L(table_64bytes))
+	.int	JMPTBL (L(3bytes), L(table_64bytes))
+	.int	JMPTBL (L(4bytes), L(table_64bytes))
+	.int	JMPTBL (L(5bytes), L(table_64bytes))
+	.int	JMPTBL (L(6bytes), L(table_64bytes))
+	.int	JMPTBL (L(7bytes), L(table_64bytes))
+	.int	JMPTBL (L(8bytes), L(table_64bytes))
+	.int	JMPTBL (L(9bytes), L(table_64bytes))
+	.int	JMPTBL (L(10bytes), L(table_64bytes))
+	.int	JMPTBL (L(11bytes), L(table_64bytes))
+	.int	JMPTBL (L(12bytes), L(table_64bytes))
+	.int	JMPTBL (L(13bytes), L(table_64bytes))
+	.int	JMPTBL (L(14bytes), L(table_64bytes))
+	.int	JMPTBL (L(15bytes), L(table_64bytes))
+	.int	JMPTBL (L(16bytes), L(table_64bytes))
+	.int	JMPTBL (L(17bytes), L(table_64bytes))
+	.int	JMPTBL (L(18bytes), L(table_64bytes))
+	.int	JMPTBL (L(19bytes), L(table_64bytes))
+	.int	JMPTBL (L(20bytes), L(table_64bytes))
+	.int	JMPTBL (L(21bytes), L(table_64bytes))
+	.int	JMPTBL (L(22bytes), L(table_64bytes))
+	.int	JMPTBL (L(23bytes), L(table_64bytes))
+	.int	JMPTBL (L(24bytes), L(table_64bytes))
+	.int	JMPTBL (L(25bytes), L(table_64bytes))
+	.int	JMPTBL (L(26bytes), L(table_64bytes))
+	.int	JMPTBL (L(27bytes), L(table_64bytes))
+	.int	JMPTBL (L(28bytes), L(table_64bytes))
+	.int	JMPTBL (L(29bytes), L(table_64bytes))
+	.int	JMPTBL (L(30bytes), L(table_64bytes))
+	.int	JMPTBL (L(31bytes), L(table_64bytes))
+	.int	JMPTBL (L(32bytes), L(table_64bytes))
+	.int	JMPTBL (L(33bytes), L(table_64bytes))
+	.int	JMPTBL (L(34bytes), L(table_64bytes))
+	.int	JMPTBL (L(35bytes), L(table_64bytes))
+	.int	JMPTBL (L(36bytes), L(table_64bytes))
+	.int	JMPTBL (L(37bytes), L(table_64bytes))
+	.int	JMPTBL (L(38bytes), L(table_64bytes))
+	.int	JMPTBL (L(39bytes), L(table_64bytes))
+	.int	JMPTBL (L(40bytes), L(table_64bytes))
+	.int	JMPTBL (L(41bytes), L(table_64bytes))
+	.int	JMPTBL (L(42bytes), L(table_64bytes))
+	.int	JMPTBL (L(43bytes), L(table_64bytes))
+	.int	JMPTBL (L(44bytes), L(table_64bytes))
+	.int	JMPTBL (L(45bytes), L(table_64bytes))
+	.int	JMPTBL (L(46bytes), L(table_64bytes))
+	.int	JMPTBL (L(47bytes), L(table_64bytes))
+	.int	JMPTBL (L(48bytes), L(table_64bytes))
+	.int	JMPTBL (L(49bytes), L(table_64bytes))
+	.int	JMPTBL (L(50bytes), L(table_64bytes))
+	.int	JMPTBL (L(51bytes), L(table_64bytes))
+	.int	JMPTBL (L(52bytes), L(table_64bytes))
+	.int	JMPTBL (L(53bytes), L(table_64bytes))
+	.int	JMPTBL (L(54bytes), L(table_64bytes))
+	.int	JMPTBL (L(55bytes), L(table_64bytes))
+	.int	JMPTBL (L(56bytes), L(table_64bytes))
+	.int	JMPTBL (L(57bytes), L(table_64bytes))
+	.int	JMPTBL (L(58bytes), L(table_64bytes))
+	.int	JMPTBL (L(59bytes), L(table_64bytes))
+	.int	JMPTBL (L(60bytes), L(table_64bytes))
+	.int	JMPTBL (L(61bytes), L(table_64bytes))
+	.int	JMPTBL (L(62bytes), L(table_64bytes))
+	.int	JMPTBL (L(63bytes), L(table_64bytes))
+	.int	JMPTBL (L(64bytes), L(table_64bytes))
+	.int	JMPTBL (L(65bytes), L(table_64bytes))
+	.int	JMPTBL (L(66bytes), L(table_64bytes))
+	.int	JMPTBL (L(67bytes), L(table_64bytes))
+	.int	JMPTBL (L(68bytes), L(table_64bytes))
+	.int	JMPTBL (L(69bytes), L(table_64bytes))
+	.int	JMPTBL (L(70bytes), L(table_64bytes))
+	.int	JMPTBL (L(71bytes), L(table_64bytes))
+	.int	JMPTBL (L(72bytes), L(table_64bytes))
+	.int	JMPTBL (L(73bytes), L(table_64bytes))
+	.int	JMPTBL (L(74bytes), L(table_64bytes))
+	.int	JMPTBL (L(75bytes), L(table_64bytes))
+	.int	JMPTBL (L(76bytes), L(table_64bytes))
+	.int	JMPTBL (L(77bytes), L(table_64bytes))
+	.int	JMPTBL (L(78bytes), L(table_64bytes))
+	.int	JMPTBL (L(79bytes), L(table_64bytes))
+#else
+L(table_64bytes):
+	.int	JMPTBL (L(0bytes), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(4bytes), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(8bytes), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(12bytes), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(16bytes), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(20bytes), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(24bytes), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(28bytes), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(32bytes), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(36bytes), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(40bytes), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(44bytes), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(48bytes), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(52bytes), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(56bytes), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(60bytes), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(64bytes), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(68bytes), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(72bytes), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(76bytes), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+	.int	JMPTBL (L(unreal_case), L(table_64bytes))
+#endif
diff --git a/libc/arch-x86_64/string/ssse3-strcmp-slm.S b/libc/arch-x86_64/string/ssse3-strcmp-slm.S
new file mode 100644
index 0000000..0dd8c27
--- /dev/null
+++ b/libc/arch-x86_64/string/ssse3-strcmp-slm.S
@@ -0,0 +1,1925 @@
+/*
+Copyright (c) 2014, Intel Corporation
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+    * Redistributions of source code must retain the above copyright notice,
+    * this list of conditions and the following disclaimer.
+
+    * Redistributions in binary form must reproduce the above copyright notice,
+    * this list of conditions and the following disclaimer in the documentation
+    * and/or other materials provided with the distribution.
+
+    * Neither the name of Intel Corporation nor the names of its contributors
+    * may be used to endorse or promote products derived from this software
+    * without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#ifdef USE_AS_STRNCMP
+/* Since the counter, %r11, is unsigned, we branch to strcmp_exitz
+   if the new counter > the old one or is 0.  */
+#define UPDATE_STRNCMP_COUNTER				\
+	/* calculate left number to compare */		\
+	lea	-16(%rcx, %r11), %r9;			\
+	cmp	%r9, %r11;				\
+	jb	L(strcmp_exitz);			\
+	test	%r9, %r9;				\
+	je	L(strcmp_exitz);			\
+	mov	%r9, %r11
+
+#else
+#define UPDATE_STRNCMP_COUNTER
+#ifndef STRCMP
+#define STRCMP		strcmp
+#endif
+#endif
+
+#ifndef L
+# define L(label)	.L##label
+#endif
+
+#ifndef cfi_startproc
+# define cfi_startproc			.cfi_startproc
+#endif
+
+#ifndef cfi_endproc
+# define cfi_endproc			.cfi_endproc
+#endif
+
+#ifndef ENTRY
+# define ENTRY(name)			\
+	.type name,  @function; 	\
+	.globl name;			\
+	.p2align 4;			\
+name:					\
+	cfi_startproc
+#endif
+
+#ifndef END
+# define END(name)			\
+	cfi_endproc;			\
+	.size name, .-name
+#endif
+#define RETURN ret
+	.section .text.ssse3,"ax",@progbits
+ENTRY (STRCMP)
+/*
+ * This implementation uses SSE to compare up to 16 bytes at a time.
+ */
+#ifdef USE_AS_STRNCMP
+	test	%rdx, %rdx
+	je	L(strcmp_exitz)
+	cmp	$1, %rdx
+	je	L(Byte0)
+	mov	%rdx, %r11
+#endif
+	mov	%esi, %ecx
+	mov	%edi, %eax
+/* Use 64bit AND here to avoid long NOP padding.  */
+	and	$0x3f, %rcx		/* rsi alignment in cache line */
+	and	$0x3f, %rax		/* rdi alignment in cache line */
+	cmp	$0x30, %ecx
+	ja	L(crosscache)	/* rsi: 16-byte load will cross cache line */
+	cmp	$0x30, %eax
+	ja	L(crosscache)	/* rdi: 16-byte load will cross cache line */
+	movlpd	(%rdi), %xmm1
+	movlpd	(%rsi), %xmm2
+	movhpd	8(%rdi), %xmm1
+	movhpd	8(%rsi), %xmm2
+	pxor	%xmm0, %xmm0		/* clear %xmm0 for null char checks */
+	pcmpeqb	%xmm1, %xmm0		/* Any null chars? */
+	pcmpeqb	%xmm2, %xmm1		/* compare first 16 bytes for equality */
+	psubb	%xmm0, %xmm1		/* packed sub of comparison results*/
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx		/* if first 16 bytes are same, edx == 0xffff */
+	jnz	L(less16bytes)	/* If not, find different value or null char */
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)	/* finish comparision */
+#endif
+	add	$16, %rsi		/* prepare to search next 16 bytes */
+	add	$16, %rdi		/* prepare to search next 16 bytes */
+
+	/*
+	 * Determine source and destination string offsets from 16-byte alignment.
+	 * Use relative offset difference between the two to determine which case
+	 * below to use.
+	 */
+	.p2align 4
+L(crosscache):
+	and	$0xfffffffffffffff0, %rsi	/* force %rsi is 16 byte aligned */
+	and	$0xfffffffffffffff0, %rdi	/* force %rdi is 16 byte aligned */
+	mov	$0xffff, %edx			/* for equivalent offset */
+	xor	%r8d, %r8d
+	and	$0xf, %ecx			/* offset of rsi */
+	and	$0xf, %eax			/* offset of rdi */
+	cmp	%eax, %ecx
+	je	L(ashr_0)			/* rsi and rdi relative offset same */
+	ja	L(bigger)
+	mov	%edx, %r8d			/* r8d is offset flag for exit tail */
+	xchg	%ecx, %eax
+	xchg	%rsi, %rdi
+L(bigger):
+	lea	15(%rax), %r9
+	sub	%rcx, %r9
+	lea	L(unaligned_table)(%rip), %r10
+	movslq	(%r10, %r9,4), %r9
+	lea	(%r10, %r9), %r10
+	jmp	*%r10				/* jump to corresponding case */
+
+/*
+ * The following cases will be handled by ashr_0
+ *  rcx(offset of rsi)  rax(offset of rdi)  relative offset  corresponding case
+ *        n(0~15)            n(0~15)           15(15+ n-n)         ashr_0
+ */
+	.p2align 4
+L(ashr_0):
+
+	movdqa	(%rsi), %xmm1
+	pxor	%xmm0, %xmm0			/* clear %xmm0 for null char check */
+	pcmpeqb	%xmm1, %xmm0			/* Any null chars? */
+	pcmpeqb	(%rdi), %xmm1			/* compare 16 bytes for equality */
+	psubb	%xmm0, %xmm1			/* packed sub of comparison results*/
+	pmovmskb %xmm1, %r9d
+	shr	%cl, %edx			/* adjust 0xffff for offset */
+	shr	%cl, %r9d			/* adjust for 16-byte offset */
+	sub	%r9d, %edx
+	/*
+	 * edx must be the same with r9d if in left byte (16-rcx) is equal to
+	 * the start from (16-rax) and no null char was seen.
+	 */
+	jne	L(less32bytes)		/* mismatch or null char */
+	UPDATE_STRNCMP_COUNTER
+	mov	$16, %rcx
+	mov	$16, %r9
+	pxor	%xmm0, %xmm0			/* clear xmm0, may have changed above */
+
+	/*
+	 * Now both strings are aligned at 16-byte boundary. Loop over strings
+	 * checking 32-bytes per iteration.
+	 */
+	.p2align 4
+L(loop_ashr_0):
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)		/* mismatch or null char seen */
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+	add	$16, %rcx
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+	add	$16, %rcx
+	jmp	L(loop_ashr_0)
+
+/*
+ * The following cases will be handled by ashr_1
+ * rcx(offset of rsi)  rax(offset of rdi)   relative offset   	corresponding case
+ *        n(15)            n -15            0(15 +(n-15) - n)         ashr_1
+ */
+	.p2align 4
+L(ashr_1):
+	pxor	%xmm0, %xmm0
+	movdqa	(%rdi), %xmm2
+	movdqa	(%rsi), %xmm1
+	pcmpeqb	%xmm1, %xmm0		/* Any null chars? */
+	pslldq	$15, %xmm2		/* shift first string to align with second */
+	pcmpeqb	%xmm1, %xmm2		/* compare 16 bytes for equality */
+	psubb	%xmm0, %xmm2		/* packed sub of comparison results*/
+	pmovmskb %xmm2, %r9d
+	shr	%cl, %edx		/* adjust 0xffff for offset */
+	shr	%cl, %r9d		/* adjust for 16-byte offset */
+	sub	%r9d, %edx
+	jnz	L(less32bytes)	/* mismatch or null char seen */
+	movdqa	(%rdi), %xmm3
+	UPDATE_STRNCMP_COUNTER
+
+	pxor	%xmm0, %xmm0
+	mov	$16, %rcx		/* index for loads*/
+	mov	$1, %r9d		/* byte position left over from less32bytes case */
+	/*
+	 * Setup %r10 value allows us to detect crossing a page boundary.
+	 * When %r10 goes positive we have crossed a page boundary and
+	 * need to do a nibble.
+	 */
+	lea	1(%rdi), %r10
+	and	$0xfff, %r10		/* offset into 4K page */
+	sub	$0x1000, %r10		/* subtract 4K pagesize */
+
+	.p2align 4
+L(loop_ashr_1):
+	add	$16, %r10
+	jg	L(nibble_ashr_1)	/* cross page boundary */
+
+L(gobble_ashr_1):
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4		 /* store for next cycle */
+
+	palignr $1, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+
+	add	$16, %r10
+	jg	L(nibble_ashr_1)	/* cross page boundary */
+
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4		/* store for next cycle */
+
+	palignr $1, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+	jmp	L(loop_ashr_1)
+
+	/*
+	 * Nibble avoids loads across page boundary. This is to avoid a potential
+	 * access into unmapped memory.
+	 */
+	.p2align 4
+L(nibble_ashr_1):
+	pcmpeqb	%xmm3, %xmm0		 /* check nibble for null char*/
+	pmovmskb %xmm0, %edx
+	test	$0xfffe, %edx
+	jnz	L(ashr_1_exittail)	/* find null char*/
+
+#ifdef USE_AS_STRNCMP
+	cmp	$14, %r11
+	jbe	L(ashr_1_exittail)
+#endif
+
+	pxor	%xmm0, %xmm0
+	sub	$0x1000, %r10		/* substract 4K from %r10 */
+	jmp	L(gobble_ashr_1)
+
+	/*
+	 * Once find null char, determine if there is a string mismatch
+	 * before the null char.
+	 */
+	.p2align 4
+L(ashr_1_exittail):
+	movdqa	(%rsi, %rcx), %xmm1
+	psrldq	$1, %xmm0
+	psrldq	$1, %xmm3
+	jmp	L(aftertail)
+
+/*
+ * The following cases will be handled by ashr_2
+ * rcx(offset of rsi)  rax(offset of rdi)   relative offset   	corresponding case
+ *        n(14~15)            n -14         1(15 +(n-14) - n)         ashr_2
+ */
+	.p2align 4
+L(ashr_2):
+	pxor	%xmm0, %xmm0
+	movdqa	(%rdi), %xmm2
+	movdqa	(%rsi), %xmm1
+	pcmpeqb	%xmm1, %xmm0
+	pslldq	$14, %xmm2
+	pcmpeqb	%xmm1, %xmm2
+	psubb	%xmm0, %xmm2
+	pmovmskb %xmm2, %r9d
+	shr	%cl, %edx
+	shr	%cl, %r9d
+	sub	%r9d, %edx
+	jnz	L(less32bytes)
+	movdqa	(%rdi), %xmm3
+	UPDATE_STRNCMP_COUNTER
+
+	pxor	%xmm0, %xmm0
+	mov	$16, %rcx	/* index for loads */
+	mov	$2, %r9d	/* byte position left over from less32bytes case */
+	/*
+	 * Setup %r10 value allows us to detect crossing a page boundary.
+	 * When %r10 goes positive we have crossed a page boundary and
+	 * need to do a nibble.
+	 */
+	lea	2(%rdi), %r10
+	and	$0xfff, %r10	/* offset into 4K page */
+	sub	$0x1000, %r10	/* subtract 4K pagesize */
+
+	.p2align 4
+L(loop_ashr_2):
+	add	$16, %r10
+	jg	L(nibble_ashr_2)
+
+L(gobble_ashr_2):
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $2, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+
+	add	$16, %r10
+	jg	L(nibble_ashr_2)	/* cross page boundary */
+
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $2, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+	jmp	L(loop_ashr_2)
+
+	.p2align 4
+L(nibble_ashr_2):
+	pcmpeqb	%xmm3, %xmm0		/* check nibble for null char */
+	pmovmskb %xmm0, %edx
+	test	$0xfffc, %edx
+	jnz	L(ashr_2_exittail)
+
+#ifdef USE_AS_STRNCMP
+	cmp	$13, %r11
+	jbe	L(ashr_2_exittail)
+#endif
+
+	pxor	%xmm0, %xmm0
+	sub	$0x1000, %r10
+	jmp	L(gobble_ashr_2)
+
+	.p2align 4
+L(ashr_2_exittail):
+	movdqa	(%rsi, %rcx), %xmm1
+	psrldq	$2, %xmm0
+	psrldq	$2, %xmm3
+	jmp	L(aftertail)
+
+/*
+ * The following cases will be handled by ashr_3
+ *  rcx(offset of rsi)  rax(offset of rdi)  relative offset	 corresponding case
+ *        n(13~15)            n -13         2(15 +(n-13) - n)         ashr_3
+ */
+	.p2align 4
+L(ashr_3):
+	pxor	%xmm0, %xmm0
+	movdqa	(%rdi), %xmm2
+	movdqa	(%rsi), %xmm1
+	pcmpeqb	%xmm1, %xmm0
+	pslldq	$13, %xmm2
+	pcmpeqb	%xmm1, %xmm2
+	psubb	%xmm0, %xmm2
+	pmovmskb %xmm2, %r9d
+	shr	%cl, %edx
+	shr	%cl, %r9d
+	sub	%r9d, %edx
+	jnz	L(less32bytes)
+	movdqa	(%rdi), %xmm3
+
+	UPDATE_STRNCMP_COUNTER
+
+	pxor	%xmm0, %xmm0
+	mov	$16, %rcx	/* index for loads */
+	mov	$3, %r9d	/* byte position left over from less32bytes case */
+	/*
+	 * Setup %r10 value allows us to detect crossing a page boundary.
+	 * When %r10 goes positive we have crossed a page boundary and
+	 * need to do a nibble.
+	 */
+	lea	3(%rdi), %r10
+	and	$0xfff, %r10	/* offset into 4K page */
+	sub	$0x1000, %r10	/* subtract 4K pagesize */
+
+	.p2align 4
+L(loop_ashr_3):
+	add	$16, %r10
+	jg	L(nibble_ashr_3)
+
+L(gobble_ashr_3):
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $3, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+
+	add	$16, %r10
+	jg	L(nibble_ashr_3)	/* cross page boundary */
+
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $3, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+	jmp	L(loop_ashr_3)
+
+	.p2align 4
+L(nibble_ashr_3):
+	pcmpeqb	%xmm3, %xmm0		/* check nibble for null char */
+	pmovmskb %xmm0, %edx
+	test	$0xfff8, %edx
+	jnz	L(ashr_3_exittail)
+
+#ifdef USE_AS_STRNCMP
+	cmp	$12, %r11
+	jbe	L(ashr_3_exittail)
+#endif
+
+	pxor	%xmm0, %xmm0
+	sub	$0x1000, %r10
+	jmp	L(gobble_ashr_3)
+
+	.p2align 4
+L(ashr_3_exittail):
+	movdqa	(%rsi, %rcx), %xmm1
+	psrldq	$3, %xmm0
+	psrldq	$3, %xmm3
+	jmp	L(aftertail)
+
+/*
+ * The following cases will be handled by ashr_4
+ *  rcx(offset of rsi)  rax(offset of rdi)  relative offset	 corresponding case
+ *        n(12~15)            n -12         3(15 +(n-12) - n)         ashr_4
+ */
+	.p2align 4
+L(ashr_4):
+	pxor	%xmm0, %xmm0
+	movdqa	(%rdi), %xmm2
+	movdqa	(%rsi), %xmm1
+	pcmpeqb	%xmm1, %xmm0
+	pslldq	$12, %xmm2
+	pcmpeqb	%xmm1, %xmm2
+	psubb	%xmm0, %xmm2
+	pmovmskb %xmm2, %r9d
+	shr	%cl, %edx
+	shr	%cl, %r9d
+	sub	%r9d, %edx
+	jnz	L(less32bytes)
+	movdqa	(%rdi), %xmm3
+
+	UPDATE_STRNCMP_COUNTER
+
+	pxor	%xmm0, %xmm0
+	mov	$16, %rcx	/* index for loads */
+	mov	$4, %r9d	/* byte position left over from less32bytes case */
+	/*
+	 * Setup %r10 value allows us to detect crossing a page boundary.
+	 * When %r10 goes positive we have crossed a page boundary and
+	 * need to do a nibble.
+	 */
+	lea	4(%rdi), %r10
+	and	$0xfff, %r10	/* offset into 4K page */
+	sub	$0x1000, %r10	/* subtract 4K pagesize */
+
+	.p2align 4
+L(loop_ashr_4):
+	add	$16, %r10
+	jg	L(nibble_ashr_4)
+
+L(gobble_ashr_4):
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $4, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+
+	add	$16, %r10
+	jg	L(nibble_ashr_4)	/* cross page boundary */
+
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $4, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+	jmp	L(loop_ashr_4)
+
+	.p2align 4
+L(nibble_ashr_4):
+	pcmpeqb	%xmm3, %xmm0		/* check nibble for null char */
+	pmovmskb %xmm0, %edx
+	test	$0xfff0, %edx
+	jnz	L(ashr_4_exittail)
+
+#ifdef USE_AS_STRNCMP
+	cmp	$11, %r11
+	jbe	L(ashr_4_exittail)
+#endif
+
+	pxor	%xmm0, %xmm0
+	sub	$0x1000, %r10
+	jmp	L(gobble_ashr_4)
+
+	.p2align 4
+L(ashr_4_exittail):
+	movdqa	(%rsi, %rcx), %xmm1
+	psrldq	$4, %xmm0
+	psrldq	$4, %xmm3
+	jmp	L(aftertail)
+
+/*
+ * The following cases will be handled by ashr_5
+ *  rcx(offset of rsi)  rax(offset of rdi)        relative offset      corresponding case
+ *        n(11~15)          n - 11      	  4(15 +(n-11) - n)         ashr_5
+ */
+	.p2align 4
+L(ashr_5):
+	pxor	%xmm0, %xmm0
+	movdqa	(%rdi), %xmm2
+	movdqa	(%rsi), %xmm1
+	pcmpeqb	%xmm1, %xmm0
+	pslldq	$11, %xmm2
+	pcmpeqb	%xmm1, %xmm2
+	psubb	%xmm0, %xmm2
+	pmovmskb %xmm2, %r9d
+	shr	%cl, %edx
+	shr	%cl, %r9d
+	sub	%r9d, %edx
+	jnz	L(less32bytes)
+	movdqa	(%rdi), %xmm3
+
+	UPDATE_STRNCMP_COUNTER
+
+	pxor	%xmm0, %xmm0
+	mov	$16, %rcx	/* index for loads */
+	mov	$5, %r9d	/* byte position left over from less32bytes case */
+	/*
+	 * Setup %r10 value allows us to detect crossing a page boundary.
+	 * When %r10 goes positive we have crossed a page boundary and
+	 * need to do a nibble.
+	 */
+	lea	5(%rdi), %r10
+	and	$0xfff, %r10	/* offset into 4K page */
+	sub	$0x1000, %r10	/* subtract 4K pagesize */
+
+	.p2align 4
+L(loop_ashr_5):
+	add	$16, %r10
+	jg	L(nibble_ashr_5)
+
+L(gobble_ashr_5):
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $5, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+
+	add	$16, %r10
+	jg	L(nibble_ashr_5)	/* cross page boundary */
+
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $5, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+	jmp	L(loop_ashr_5)
+
+	.p2align 4
+L(nibble_ashr_5):
+	pcmpeqb	%xmm3, %xmm0		/* check nibble for null char */
+	pmovmskb %xmm0, %edx
+	test	$0xffe0, %edx
+	jnz	L(ashr_5_exittail)
+
+#ifdef USE_AS_STRNCMP
+	cmp	$10, %r11
+	jbe	L(ashr_5_exittail)
+#endif
+
+	pxor	%xmm0, %xmm0
+	sub	$0x1000, %r10
+	jmp	L(gobble_ashr_5)
+
+	.p2align 4
+L(ashr_5_exittail):
+	movdqa	(%rsi, %rcx), %xmm1
+	psrldq	$5, %xmm0
+	psrldq	$5, %xmm3
+	jmp	L(aftertail)
+
+/*
+ * The following cases will be handled by ashr_6
+ *  rcx(offset of rsi)  rax(offset of rdi)        relative offset      corresponding case
+ *        n(10~15)          n - 10      	  5(15 +(n-10) - n)         ashr_6
+ */
+	.p2align 4
+L(ashr_6):
+	pxor	%xmm0, %xmm0
+	movdqa	(%rdi), %xmm2
+	movdqa	(%rsi), %xmm1
+	pcmpeqb	%xmm1, %xmm0
+	pslldq	$10, %xmm2
+	pcmpeqb	%xmm1, %xmm2
+	psubb	%xmm0, %xmm2
+	pmovmskb %xmm2, %r9d
+	shr	%cl, %edx
+	shr	%cl, %r9d
+	sub	%r9d, %edx
+	jnz	L(less32bytes)
+	movdqa	(%rdi), %xmm3
+
+	UPDATE_STRNCMP_COUNTER
+
+	pxor	%xmm0, %xmm0
+	mov	$16, %rcx	/* index for loads */
+	mov	$6, %r9d	/* byte position left over from less32bytes case */
+	/*
+	 * Setup %r10 value allows us to detect crossing a page boundary.
+	 * When %r10 goes positive we have crossed a page boundary and
+	 * need to do a nibble.
+	 */
+	lea	6(%rdi), %r10
+	and	$0xfff, %r10	/* offset into 4K page */
+	sub	$0x1000, %r10	/* subtract 4K pagesize */
+
+	.p2align 4
+L(loop_ashr_6):
+	add	$16, %r10
+	jg	L(nibble_ashr_6)
+
+L(gobble_ashr_6):
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $6, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+
+	add	$16, %r10
+	jg	L(nibble_ashr_6)	/* cross page boundary */
+
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $6, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+	jmp	L(loop_ashr_6)
+
+	.p2align 4
+L(nibble_ashr_6):
+	pcmpeqb	%xmm3, %xmm0		/* check nibble for null char */
+	pmovmskb %xmm0, %edx
+	test	$0xffc0, %edx
+	jnz	L(ashr_6_exittail)
+
+#ifdef USE_AS_STRNCMP
+	cmp	$9, %r11
+	jbe	L(ashr_6_exittail)
+#endif
+
+	pxor	%xmm0, %xmm0
+	sub	$0x1000, %r10
+	jmp	L(gobble_ashr_6)
+
+	.p2align 4
+L(ashr_6_exittail):
+	movdqa	(%rsi, %rcx), %xmm1
+	psrldq	$6, %xmm0
+	psrldq	$6, %xmm3
+	jmp	L(aftertail)
+
+/*
+ * The following cases will be handled by ashr_7
+ *  rcx(offset of rsi)  rax(offset of rdi)        relative offset      corresponding case
+ *        n(9~15)          n - 9      	        6(15 +(n - 9) - n)         ashr_7
+ */
+	.p2align 4
+L(ashr_7):
+	pxor	%xmm0, %xmm0
+	movdqa	(%rdi), %xmm2
+	movdqa	(%rsi), %xmm1
+	pcmpeqb	%xmm1, %xmm0
+	pslldq	$9, %xmm2
+	pcmpeqb	%xmm1, %xmm2
+	psubb	%xmm0, %xmm2
+	pmovmskb %xmm2, %r9d
+	shr	%cl, %edx
+	shr	%cl, %r9d
+	sub	%r9d, %edx
+	jnz	L(less32bytes)
+	movdqa	(%rdi), %xmm3
+
+	UPDATE_STRNCMP_COUNTER
+
+	pxor	%xmm0, %xmm0
+	mov	$16, %rcx	/* index for loads */
+	mov	$7, %r9d	/* byte position left over from less32bytes case */
+	/*
+	 * Setup %r10 value allows us to detect crossing a page boundary.
+	 * When %r10 goes positive we have crossed a page boundary and
+	 * need to do a nibble.
+	 */
+	lea	7(%rdi), %r10
+	and	$0xfff, %r10	/* offset into 4K page */
+	sub	$0x1000, %r10	/* subtract 4K pagesize */
+
+	.p2align 4
+L(loop_ashr_7):
+	add	$16, %r10
+	jg	L(nibble_ashr_7)
+
+L(gobble_ashr_7):
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $7, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+
+	add	$16, %r10
+	jg	L(nibble_ashr_7)	/* cross page boundary */
+
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $7, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+	jmp	L(loop_ashr_7)
+
+	.p2align 4
+L(nibble_ashr_7):
+	pcmpeqb	%xmm3, %xmm0		/* check nibble for null char */
+	pmovmskb %xmm0, %edx
+	test	$0xff80, %edx
+	jnz	L(ashr_7_exittail)
+
+#ifdef USE_AS_STRNCMP
+	cmp	$8, %r11
+	jbe	L(ashr_7_exittail)
+#endif
+
+	pxor	%xmm0, %xmm0
+	sub	$0x1000, %r10
+	jmp	L(gobble_ashr_7)
+
+	.p2align 4
+L(ashr_7_exittail):
+	movdqa	(%rsi, %rcx), %xmm1
+	psrldq	$7, %xmm0
+	psrldq	$7, %xmm3
+	jmp	L(aftertail)
+
+/*
+ *  The following cases will be handled by ashr_8
+ *  rcx(offset of rsi)  rax(offset of rdi)        relative offset	 corresponding case
+ *        n(8~15)          n - 8      	        7(15 +(n - 8) - n)         ashr_8
+ */
+	.p2align 4
+L(ashr_8):
+	pxor	%xmm0, %xmm0
+	movdqa	(%rdi), %xmm2
+	movdqa	(%rsi), %xmm1
+	pcmpeqb	%xmm1, %xmm0
+	pslldq	$8, %xmm2
+	pcmpeqb	%xmm1, %xmm2
+	psubb	%xmm0, %xmm2
+	pmovmskb %xmm2, %r9d
+	shr	%cl, %edx
+	shr	%cl, %r9d
+	sub	%r9d, %edx
+	jnz	L(less32bytes)
+	movdqa	(%rdi), %xmm3
+
+	UPDATE_STRNCMP_COUNTER
+
+	pxor	%xmm0, %xmm0
+	mov	$16, %rcx	/* index for loads */
+	mov	$8, %r9d	/* byte position left over from less32bytes case */
+	/*
+	 * Setup %r10 value allows us to detect crossing a page boundary.
+	 * When %r10 goes positive we have crossed a page boundary and
+	 * need to do a nibble.
+	 */
+	lea	8(%rdi), %r10
+	and	$0xfff, %r10	/* offset into 4K page */
+	sub	$0x1000, %r10	/* subtract 4K pagesize */
+
+	.p2align 4
+L(loop_ashr_8):
+	add	$16, %r10
+	jg	L(nibble_ashr_8)
+
+L(gobble_ashr_8):
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $8, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+
+	add	$16, %r10
+	jg	L(nibble_ashr_8)	/* cross page boundary */
+
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $8, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+	jmp	L(loop_ashr_8)
+
+	.p2align 4
+L(nibble_ashr_8):
+	pcmpeqb	%xmm3, %xmm0		/* check nibble for null char */
+	pmovmskb %xmm0, %edx
+	test	$0xff00, %edx
+	jnz	L(ashr_8_exittail)
+
+#ifdef USE_AS_STRNCMP
+	cmp	$7, %r11
+	jbe	L(ashr_8_exittail)
+#endif
+
+	pxor	%xmm0, %xmm0
+	sub	$0x1000, %r10
+	jmp	L(gobble_ashr_8)
+
+	.p2align 4
+L(ashr_8_exittail):
+	movdqa	(%rsi, %rcx), %xmm1
+	psrldq	$8, %xmm0
+	psrldq	$8, %xmm3
+	jmp	L(aftertail)
+
+/*
+ *  The following cases will be handled by ashr_9
+ *  rcx(offset of rsi)  rax(offset of rdi)        relative offset	 corresponding case
+ *        n(7~15)          n - 7      	        8(15 +(n - 7) - n)         ashr_9
+ */
+	.p2align 4
+L(ashr_9):
+	pxor	%xmm0, %xmm0
+	movdqa	(%rdi), %xmm2
+	movdqa	(%rsi), %xmm1
+	pcmpeqb	%xmm1, %xmm0
+	pslldq	$7, %xmm2
+	pcmpeqb	%xmm1, %xmm2
+	psubb	%xmm0, %xmm2
+	pmovmskb %xmm2, %r9d
+	shr	%cl, %edx
+	shr	%cl, %r9d
+	sub	%r9d, %edx
+	jnz	L(less32bytes)
+	movdqa	(%rdi), %xmm3
+
+	UPDATE_STRNCMP_COUNTER
+
+	pxor	%xmm0, %xmm0
+	mov	$16, %rcx	/* index for loads */
+	mov	$9, %r9d	/* byte position left over from less32bytes case */
+	/*
+	 * Setup %r10 value allows us to detect crossing a page boundary.
+	 * When %r10 goes positive we have crossed a page boundary and
+	 * need to do a nibble.
+	 */
+	lea	9(%rdi), %r10
+	and	$0xfff, %r10	/* offset into 4K page */
+	sub	$0x1000, %r10	/* subtract 4K pagesize */
+
+	.p2align 4
+L(loop_ashr_9):
+	add	$16, %r10
+	jg	L(nibble_ashr_9)
+
+L(gobble_ashr_9):
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $9, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+
+	add	$16, %r10
+	jg	L(nibble_ashr_9)	/* cross page boundary */
+
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $9, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3		/* store for next cycle */
+	jmp	L(loop_ashr_9)
+
+	.p2align 4
+L(nibble_ashr_9):
+	pcmpeqb	%xmm3, %xmm0		/* check nibble for null char */
+	pmovmskb %xmm0, %edx
+	test	$0xfe00, %edx
+	jnz	L(ashr_9_exittail)
+
+#ifdef USE_AS_STRNCMP
+	cmp	$6, %r11
+	jbe	L(ashr_9_exittail)
+#endif
+
+	pxor	%xmm0, %xmm0
+	sub	$0x1000, %r10
+	jmp	L(gobble_ashr_9)
+
+	.p2align 4
+L(ashr_9_exittail):
+	movdqa	(%rsi, %rcx), %xmm1
+	psrldq	$9, %xmm0
+	psrldq	$9, %xmm3
+	jmp	L(aftertail)
+
+/*
+ *  The following cases will be handled by ashr_10
+ *  rcx(offset of rsi)  rax(offset of rdi)        relative offset	 corresponding case
+ *        n(6~15)          n - 6      	        9(15 +(n - 6) - n)         ashr_10
+ */
+	.p2align 4
+L(ashr_10):
+	pxor	%xmm0, %xmm0
+	movdqa	(%rdi), %xmm2
+	movdqa	(%rsi), %xmm1
+	pcmpeqb	%xmm1, %xmm0
+	pslldq	$6, %xmm2
+	pcmpeqb	%xmm1, %xmm2
+	psubb	%xmm0, %xmm2
+	pmovmskb %xmm2, %r9d
+	shr	%cl, %edx
+	shr	%cl, %r9d
+	sub	%r9d, %edx
+	jnz	L(less32bytes)
+	movdqa	(%rdi), %xmm3
+
+	UPDATE_STRNCMP_COUNTER
+
+	pxor	%xmm0, %xmm0
+	mov	$16, %rcx	/* index for loads */
+	mov	$10, %r9d	/* byte position left over from less32bytes case */
+	/*
+	 * Setup %r10 value allows us to detect crossing a page boundary.
+	 * When %r10 goes positive we have crossed a page boundary and
+	 * need to do a nibble.
+	 */
+	lea	10(%rdi), %r10
+	and	$0xfff, %r10	/* offset into 4K page */
+	sub	$0x1000, %r10	/* subtract 4K pagesize */
+
+	.p2align 4
+L(loop_ashr_10):
+	add	$16, %r10
+	jg	L(nibble_ashr_10)
+
+L(gobble_ashr_10):
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $10, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+
+	add	$16, %r10
+	jg	L(nibble_ashr_10)	/* cross page boundary */
+
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $10, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+	jmp	L(loop_ashr_10)
+
+	.p2align 4
+L(nibble_ashr_10):
+	pcmpeqb	%xmm3, %xmm0		/* check nibble for null char */
+	pmovmskb %xmm0, %edx
+	test	$0xfc00, %edx
+	jnz	L(ashr_10_exittail)
+
+#ifdef USE_AS_STRNCMP
+	cmp	$5, %r11
+	jbe	L(ashr_10_exittail)
+#endif
+
+	pxor	%xmm0, %xmm0
+	sub	$0x1000, %r10
+	jmp	L(gobble_ashr_10)
+
+	.p2align 4
+L(ashr_10_exittail):
+	movdqa	(%rsi, %rcx), %xmm1
+	psrldq	$10, %xmm0
+	psrldq	$10, %xmm3
+	jmp	L(aftertail)
+
+/*
+ *  The following cases will be handled by ashr_11
+ *  rcx(offset of rsi)  rax(offset of rdi)        relative offset	 corresponding case
+ *        n(5~15)          n - 5      	        10(15 +(n - 5) - n)         ashr_11
+ */
+	.p2align 4
+L(ashr_11):
+	pxor	%xmm0, %xmm0
+	movdqa	(%rdi), %xmm2
+	movdqa	(%rsi), %xmm1
+	pcmpeqb	%xmm1, %xmm0
+	pslldq	$5, %xmm2
+	pcmpeqb	%xmm1, %xmm2
+	psubb	%xmm0, %xmm2
+	pmovmskb %xmm2, %r9d
+	shr	%cl, %edx
+	shr	%cl, %r9d
+	sub	%r9d, %edx
+	jnz	L(less32bytes)
+	movdqa	(%rdi), %xmm3
+
+	UPDATE_STRNCMP_COUNTER
+
+	pxor	%xmm0, %xmm0
+	mov	$16, %rcx	/* index for loads */
+	mov	$11, %r9d	/* byte position left over from less32bytes case */
+	/*
+	 * Setup %r10 value allows us to detect crossing a page boundary.
+	 * When %r10 goes positive we have crossed a page boundary and
+	 * need to do a nibble.
+	 */
+	lea	11(%rdi), %r10
+	and	$0xfff, %r10	/* offset into 4K page */
+	sub	$0x1000, %r10	/* subtract 4K pagesize */
+
+	.p2align 4
+L(loop_ashr_11):
+	add	$16, %r10
+	jg	L(nibble_ashr_11)
+
+L(gobble_ashr_11):
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $11, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+
+	add	$16, %r10
+	jg	L(nibble_ashr_11)	/* cross page boundary */
+
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $11, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+	jmp	L(loop_ashr_11)
+
+	.p2align 4
+L(nibble_ashr_11):
+	pcmpeqb	%xmm3, %xmm0		/* check nibble for null char */
+	pmovmskb %xmm0, %edx
+	test	$0xf800, %edx
+	jnz	L(ashr_11_exittail)
+
+#ifdef USE_AS_STRNCMP
+	cmp	$4, %r11
+	jbe	L(ashr_11_exittail)
+#endif
+
+	pxor	%xmm0, %xmm0
+	sub	$0x1000, %r10
+	jmp	L(gobble_ashr_11)
+
+	.p2align 4
+L(ashr_11_exittail):
+	movdqa	(%rsi, %rcx), %xmm1
+	psrldq	$11, %xmm0
+	psrldq	$11, %xmm3
+	jmp	L(aftertail)
+
+/*
+ *  The following cases will be handled by ashr_12
+ *  rcx(offset of rsi)  rax(offset of rdi)        relative offset	 corresponding case
+ *        n(4~15)          n - 4      	        11(15 +(n - 4) - n)         ashr_12
+ */
+	.p2align 4
+L(ashr_12):
+	pxor	%xmm0, %xmm0
+	movdqa	(%rdi), %xmm2
+	movdqa	(%rsi), %xmm1
+	pcmpeqb	%xmm1, %xmm0
+	pslldq	$4, %xmm2
+	pcmpeqb	%xmm1, %xmm2
+	psubb	%xmm0, %xmm2
+	pmovmskb %xmm2, %r9d
+	shr	%cl, %edx
+	shr	%cl, %r9d
+	sub	%r9d, %edx
+	jnz	L(less32bytes)
+	movdqa	(%rdi), %xmm3
+
+	UPDATE_STRNCMP_COUNTER
+
+	pxor	%xmm0, %xmm0
+	mov	$16, %rcx	/* index for loads */
+	mov	$12, %r9d	/* byte position left over from less32bytes case */
+	/*
+	 * Setup %r10 value allows us to detect crossing a page boundary.
+	 * When %r10 goes positive we have crossed a page boundary and
+	 * need to do a nibble.
+	 */
+	lea	12(%rdi), %r10
+	and	$0xfff, %r10	/* offset into 4K page */
+	sub	$0x1000, %r10	/* subtract 4K pagesize */
+
+	.p2align 4
+L(loop_ashr_12):
+	add	$16, %r10
+	jg	L(nibble_ashr_12)
+
+L(gobble_ashr_12):
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $12, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+
+	add	$16, %r10
+	jg	L(nibble_ashr_12)	/* cross page boundary */
+
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $12, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+	jmp	L(loop_ashr_12)
+
+	.p2align 4
+L(nibble_ashr_12):
+	pcmpeqb	%xmm3, %xmm0		/* check nibble for null char */
+	pmovmskb %xmm0, %edx
+	test	$0xf000, %edx
+	jnz	L(ashr_12_exittail)
+
+#ifdef USE_AS_STRNCMP
+	cmp	$3, %r11
+	jbe	L(ashr_12_exittail)
+#endif
+
+	pxor	%xmm0, %xmm0
+	sub	$0x1000, %r10
+	jmp	L(gobble_ashr_12)
+
+	.p2align 4
+L(ashr_12_exittail):
+	movdqa	(%rsi, %rcx), %xmm1
+	psrldq	$12, %xmm0
+	psrldq	$12, %xmm3
+	jmp	L(aftertail)
+
+/*
+ *  The following cases will be handled by ashr_13
+ *  rcx(offset of rsi)  rax(offset of rdi)        relative offset	 corresponding case
+ *        n(3~15)          n - 3      	        12(15 +(n - 3) - n)         ashr_13
+ */
+	.p2align 4
+L(ashr_13):
+	pxor	%xmm0, %xmm0
+	movdqa	(%rdi), %xmm2
+	movdqa	(%rsi), %xmm1
+	pcmpeqb	%xmm1, %xmm0
+	pslldq	$3, %xmm2
+	pcmpeqb	%xmm1, %xmm2
+	psubb	%xmm0, %xmm2
+	pmovmskb %xmm2, %r9d
+	shr	%cl, %edx
+	shr	%cl, %r9d
+	sub	%r9d, %edx
+	jnz	L(less32bytes)
+	movdqa	(%rdi), %xmm3
+
+	UPDATE_STRNCMP_COUNTER
+
+	pxor	%xmm0, %xmm0
+	mov	$16, %rcx	/* index for loads */
+	mov	$13, %r9d	/* byte position left over from less32bytes case */
+	/*
+	 * Setup %r10 value allows us to detect crossing a page boundary.
+	 * When %r10 goes positive we have crossed a page boundary and
+	 * need to do a nibble.
+	 */
+	lea	13(%rdi), %r10
+	and	$0xfff, %r10	/* offset into 4K page */
+	sub	$0x1000, %r10	/* subtract 4K pagesize */
+
+	.p2align 4
+L(loop_ashr_13):
+	add	$16, %r10
+	jg	L(nibble_ashr_13)
+
+L(gobble_ashr_13):
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $13, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+
+	add	$16, %r10
+	jg	L(nibble_ashr_13)	/* cross page boundary */
+
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $13, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+	jmp	L(loop_ashr_13)
+
+	.p2align 4
+L(nibble_ashr_13):
+	pcmpeqb	%xmm3, %xmm0		/* check nibble for null char */
+	pmovmskb %xmm0, %edx
+	test	$0xe000, %edx
+	jnz	L(ashr_13_exittail)
+
+#ifdef USE_AS_STRNCMP
+	cmp	$2, %r11
+	jbe	L(ashr_13_exittail)
+#endif
+
+	pxor	%xmm0, %xmm0
+	sub	$0x1000, %r10
+	jmp	L(gobble_ashr_13)
+
+	.p2align 4
+L(ashr_13_exittail):
+	movdqa	(%rsi, %rcx), %xmm1
+	psrldq  $13, %xmm0
+	psrldq  $13, %xmm3
+	jmp	L(aftertail)
+
+/*
+ *  The following cases will be handled by ashr_14
+ *  rcx(offset of rsi)  rax(offset of rdi)        relative offset	 corresponding case
+ *        n(2~15)          n - 2      	        13(15 +(n - 2) - n)         ashr_14
+ */
+	.p2align 4
+L(ashr_14):
+	pxor	%xmm0, %xmm0
+	movdqa	(%rdi), %xmm2
+	movdqa	(%rsi), %xmm1
+	pcmpeqb	%xmm1, %xmm0
+	pslldq  $2, %xmm2
+	pcmpeqb	%xmm1, %xmm2
+	psubb	%xmm0, %xmm2
+	pmovmskb %xmm2, %r9d
+	shr	%cl, %edx
+	shr	%cl, %r9d
+	sub	%r9d, %edx
+	jnz	L(less32bytes)
+	movdqa	(%rdi), %xmm3
+
+	UPDATE_STRNCMP_COUNTER
+
+	pxor	%xmm0, %xmm0
+	mov	$16, %rcx	/* index for loads */
+	mov	$14, %r9d	/* byte position left over from less32bytes case */
+	/*
+	 * Setup %r10 value allows us to detect crossing a page boundary.
+	 * When %r10 goes positive we have crossed a page boundary and
+	 * need to do a nibble.
+	 */
+	lea	14(%rdi), %r10
+	and	$0xfff, %r10	/* offset into 4K page */
+	sub	$0x1000, %r10	/* subtract 4K pagesize */
+
+	.p2align 4
+L(loop_ashr_14):
+	add	$16, %r10
+	jg	L(nibble_ashr_14)
+
+L(gobble_ashr_14):
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $14, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+
+	add	$16, %r10
+	jg	L(nibble_ashr_14)	/* cross page boundary */
+
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $14, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+	jmp	L(loop_ashr_14)
+
+	.p2align 4
+L(nibble_ashr_14):
+	pcmpeqb	%xmm3, %xmm0		/* check nibble for null char */
+	pmovmskb %xmm0, %edx
+	test	$0xc000, %edx
+	jnz	L(ashr_14_exittail)
+
+#ifdef USE_AS_STRNCMP
+	cmp	$1, %r11
+	jbe	L(ashr_14_exittail)
+#endif
+
+	pxor	%xmm0, %xmm0
+	sub	$0x1000, %r10
+	jmp	L(gobble_ashr_14)
+
+	.p2align 4
+L(ashr_14_exittail):
+	movdqa	(%rsi, %rcx), %xmm1
+	psrldq	$14, %xmm0
+	psrldq	$14, %xmm3
+	jmp	L(aftertail)
+
+/*
+ *  The following cases will be handled by ashr_15
+ *  rcx(offset of rsi)  rax(offset of rdi)        relative offset	 corresponding case
+ *        n(1~15)          n - 1      	        14(15 +(n - 1) - n)         ashr_15
+ */
+	.p2align 4
+L(ashr_15):
+	pxor	%xmm0, %xmm0
+	movdqa	(%rdi), %xmm2
+	movdqa	(%rsi), %xmm1
+	pcmpeqb	%xmm1, %xmm0
+	pslldq	$1, %xmm2
+	pcmpeqb	%xmm1, %xmm2
+	psubb	%xmm0, %xmm2
+	pmovmskb %xmm2, %r9d
+	shr	%cl, %edx
+	shr	%cl, %r9d
+	sub	%r9d, %edx
+	jnz	L(less32bytes)
+
+	movdqa	(%rdi), %xmm3
+
+	UPDATE_STRNCMP_COUNTER
+
+	pxor	%xmm0, %xmm0
+	mov	$16, %rcx	/* index for loads */
+	mov	$15, %r9d	/* byte position left over from less32bytes case */
+	/*
+	 * Setup %r10 value allows us to detect crossing a page boundary.
+	 * When %r10 goes positive we have crossed a page boundary and
+	 * need to do a nibble.
+	 */
+	lea	15(%rdi), %r10
+	and	$0xfff, %r10	/* offset into 4K page */
+
+	sub	$0x1000, %r10	/* subtract 4K pagesize */
+
+	.p2align 4
+L(loop_ashr_15):
+	add	$16, %r10
+	jg	L(nibble_ashr_15)
+
+L(gobble_ashr_15):
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $15, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+
+	add	$16, %r10
+	jg	L(nibble_ashr_15)	/* cross page boundary */
+
+	movdqa	(%rsi, %rcx), %xmm1
+	movdqa	(%rdi, %rcx), %xmm2
+	movdqa	%xmm2, %xmm4
+
+	palignr $15, %xmm3, %xmm2        /* merge into one 16byte value */
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm2, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	sub	$0xffff, %edx
+	jnz	L(exit)
+
+#ifdef USE_AS_STRNCMP
+	sub	$16, %r11
+	jbe	L(strcmp_exitz)
+#endif
+
+	add	$16, %rcx
+	movdqa	%xmm4, %xmm3
+	jmp	L(loop_ashr_15)
+
+	.p2align 4
+L(nibble_ashr_15):
+	pcmpeqb	%xmm3, %xmm0		/* check nibble for null char */
+	pmovmskb %xmm0, %edx
+	test	$0x8000, %edx
+	jnz	L(ashr_15_exittail)
+
+#ifdef USE_AS_STRNCMP
+	test	%r11, %r11
+	je	L(ashr_15_exittail)
+#endif
+
+	pxor	%xmm0, %xmm0
+	sub	$0x1000, %r10
+	jmp	L(gobble_ashr_15)
+
+	.p2align 4
+L(ashr_15_exittail):
+	movdqa	(%rsi, %rcx), %xmm1
+	psrldq	$15, %xmm3
+	psrldq	$15, %xmm0
+
+	.p2align 4
+L(aftertail):
+	pcmpeqb	%xmm3, %xmm1
+	psubb	%xmm0, %xmm1
+	pmovmskb %xmm1, %edx
+	not	%edx
+
+	.p2align 4
+L(exit):
+	lea	-16(%r9, %rcx), %rax	/* locate the exact offset for rdi */
+L(less32bytes):
+	lea	(%rdi, %rax), %rdi	/* locate the exact address for first operand(rdi) */
+	lea	(%rsi, %rcx), %rsi	/* locate the exact address for second operand(rsi) */
+	test	%r8d, %r8d
+	jz	L(ret)
+	xchg	%rsi, %rdi		/* recover original order according to flag(%r8d) */
+
+	.p2align 4
+L(ret):
+L(less16bytes):
+	bsf	%rdx, %rdx		/* find and store bit index in %rdx */
+
+#ifdef USE_AS_STRNCMP
+	sub	%rdx, %r11
+	jbe	L(strcmp_exitz)
+#endif
+	movzbl	(%rsi, %rdx), %ecx
+	movzbl	(%rdi, %rdx), %eax
+
+	sub	%ecx, %eax
+	ret
+
+L(strcmp_exitz):
+	xor	%eax, %eax
+	ret
+
+	.p2align 4
+L(Byte0):
+	movzx	(%rsi), %ecx
+	movzx	(%rdi), %eax
+
+	sub	%ecx, %eax
+	ret
+END (STRCMP)
+
+	.section .rodata,"a",@progbits
+	.p2align 3
+L(unaligned_table):
+	.int	L(ashr_1) - L(unaligned_table)
+	.int	L(ashr_2) - L(unaligned_table)
+	.int	L(ashr_3) - L(unaligned_table)
+	.int	L(ashr_4) - L(unaligned_table)
+	.int	L(ashr_5) - L(unaligned_table)
+	.int	L(ashr_6) - L(unaligned_table)
+	.int	L(ashr_7) - L(unaligned_table)
+	.int	L(ashr_8) - L(unaligned_table)
+	.int	L(ashr_9) - L(unaligned_table)
+	.int	L(ashr_10) - L(unaligned_table)
+	.int	L(ashr_11) - L(unaligned_table)
+	.int	L(ashr_12) - L(unaligned_table)
+	.int	L(ashr_13) - L(unaligned_table)
+	.int	L(ashr_14) - L(unaligned_table)
+	.int	L(ashr_15) - L(unaligned_table)
+	.int	L(ashr_0) - L(unaligned_table)
diff --git a/libc/arch-x86_64/string/ssse3-strncmp-slm.S b/libc/arch-x86_64/string/ssse3-strncmp-slm.S
new file mode 100644
index 0000000..0e40775
--- /dev/null
+++ b/libc/arch-x86_64/string/ssse3-strncmp-slm.S
@@ -0,0 +1,33 @@
+/*
+Copyright (c) 2014, Intel Corporation
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+    * Redistributions of source code must retain the above copyright notice,
+    * this list of conditions and the following disclaimer.
+
+    * Redistributions in binary form must reproduce the above copyright notice,
+    * this list of conditions and the following disclaimer in the documentation
+    * and/or other materials provided with the distribution.
+
+    * Neither the name of Intel Corporation nor the names of its contributors
+    * may be used to endorse or promote products derived from this software
+    * without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#define USE_AS_STRNCMP
+#define STRCMP		strncmp
+#include "ssse3-strcmp-slm.S"
diff --git a/libc/arch-x86_64/x86_64.mk b/libc/arch-x86_64/x86_64.mk
index f0010f3..2bcf432 100644
--- a/libc/arch-x86_64/x86_64.mk
+++ b/libc/arch-x86_64/x86_64.mk
@@ -3,11 +3,7 @@
 libc_common_src_files_x86_64 := \
     bionic/index.cpp \
     bionic/memchr.c \
-    bionic/memcmp.c \
-    bionic/memcpy.cpp \
-    bionic/memmove.c \
     bionic/memrchr.c \
-    bionic/memset.c \
     bionic/strchr.cpp \
     bionic/strnlen.c \
     bionic/strrchr.cpp \
@@ -18,18 +14,8 @@
     upstream-freebsd/lib/libc/string/wcslen.c \
     upstream-freebsd/lib/libc/string/wcsrchr.c \
     upstream-freebsd/lib/libc/string/wmemcmp.c \
-    upstream-openbsd/lib/libc/string/bcopy.c \
-    upstream-openbsd/lib/libc/string/stpcpy.c \
-    upstream-openbsd/lib/libc/string/stpncpy.c \
-    upstream-openbsd/lib/libc/string/strcat.c \
-    upstream-openbsd/lib/libc/string/strcmp.c \
-    upstream-openbsd/lib/libc/string/strcpy.c \
     upstream-openbsd/lib/libc/string/strlcat.c \
     upstream-openbsd/lib/libc/string/strlcpy.c \
-    upstream-openbsd/lib/libc/string/strlen.c \
-    upstream-openbsd/lib/libc/string/strncat.c \
-    upstream-openbsd/lib/libc/string/strncmp.c \
-    upstream-openbsd/lib/libc/string/strncpy.c \
 
 # Fortify implementations of libc functions.
 libc_common_src_files_x86_64 += \
@@ -55,6 +41,23 @@
     arch-x86_64/bionic/vfork.S \
     bionic/__memcmp16.cpp \
 
+libc_bionic_src_files_x86_64 += \
+    arch-x86_64/string/sse2-bcopy-slm.S \
+    arch-x86_64/string/sse2-bzero-slm.S \
+    arch-x86_64/string/sse2-memcpy-slm.S \
+    arch-x86_64/string/sse2-memmove-slm.S \
+    arch-x86_64/string/sse2-memset-slm.S \
+    arch-x86_64/string/sse2-stpcpy-slm.S \
+    arch-x86_64/string/sse2-stpncpy-slm.S \
+    arch-x86_64/string/sse2-strcat-slm.S \
+    arch-x86_64/string/sse2-strcpy-slm.S \
+    arch-x86_64/string/sse2-strlen-slm.S \
+    arch-x86_64/string/sse2-strncat-slm.S \
+    arch-x86_64/string/sse2-strncpy-slm.S \
+    arch-x86_64/string/sse4-memcmp-slm.S \
+    arch-x86_64/string/ssse3-strcmp-slm.S \
+    arch-x86_64/string/ssse3-strncmp-slm.S \
+
 libc_crt_target_cflags_x86_64 += \
     -m64 \
     -I$(LOCAL_PATH)/arch-x86_64/include