diff --git a/README b/README
index 1b43ba22d..2dd93ac58 100644
--- a/README
+++ b/README
@@ -42,7 +42,7 @@ Architecture-specific notes:
 		To use NEON instructions, add "-mfpu=neon" to CFLAGS.
 	x86:	The miner checks for SSE2 instructions support at runtime,
 		and uses them if they are available.
-	x86-64:	The miner can take advantage of AVX and XOP instructions,
+	x86-64:	The miner can take advantage of AVX, AVX2 and XOP instructions,
 		but only if both the CPU and the operating system support them.
 		    * Linux supports AVX starting from kernel version 2.6.30.
 		    * FreeBSD supports AVX starting with 9.1-RELEASE.
@@ -50,7 +50,7 @@ Architecture-specific notes:
 		    * Windows supports AVX starting from Windows 7 SP1 and
 		      Windows Server 2008 R2 SP1.
 		The configure script outputs a warning if the assembler
-		cannot compile AVX or XOP instructions. In that case, the miner
+		doesn't support some instruction sets. In that case, the miner
 		can still be built, but unavailable optimizations are left off.
 
 Usage instructions:  Run "minerd --help" to see options.
diff --git a/configure.ac b/configure.ac
index f5260e991..6663ae06c 100644
--- a/configure.ac
+++ b/configure.ac
@@ -77,6 +77,14 @@ then
       AC_MSG_RESULT(no)
       AC_MSG_WARN([The assembler does not support the XOP instruction set.])
     )
+    AC_MSG_CHECKING(whether we can compile AVX2 code)
+    AC_COMPILE_IFELSE([AC_LANG_PROGRAM(,[asm ("vpaddd %ymm0, %ymm1, %ymm2");])],
+      AC_DEFINE(USE_AVX2, 1, [Define to 1 if AVX2 assembly is available.])
+      AC_MSG_RESULT(yes)
+    ,
+      AC_MSG_RESULT(no)
+      AC_MSG_WARN([The assembler does not support the AVX2 instruction set.])
+    )
   ,
     AC_MSG_RESULT(no)
     AC_MSG_WARN([The assembler does not support the AVX instruction set.])
diff --git a/cpu-miner.c b/cpu-miner.c
index 7317692cc..e61eff8bb 100644
--- a/cpu-miner.c
+++ b/cpu-miner.c
@@ -668,7 +668,7 @@ static void *miner_thread(void *userdata)
 	int thr_id = mythr->id;
 	struct work work;
 	uint32_t max_nonce;
-	uint32_t end_nonce = 0xffffffffU / opt_n_threads * (thr_id + 1) - 0x10;
+	uint32_t end_nonce = 0xffffffffU / opt_n_threads * (thr_id + 1) - 0x20;
 	unsigned char *scratchbuf = NULL;
 	char s[16];
 	int i;
diff --git a/miner.h b/miner.h
index 867a1e8a1..4aace442f 100644
--- a/miner.h
+++ b/miner.h
@@ -141,6 +141,13 @@ void sha256_init_4way(uint32_t *state);
 void sha256_transform_4way(uint32_t *state, const uint32_t *block, int swap);
 #endif
 
+#if defined(__x86_64__) && defined(USE_AVX2)
+#define HAVE_SHA256_8WAY 1
+int sha256_use_8way();
+void sha256_init_8way(uint32_t *state);
+void sha256_transform_8way(uint32_t *state, const uint32_t *block, int swap);
+#endif
+
 extern int scanhash_sha256d(int thr_id, uint32_t *pdata,
 	const uint32_t *ptarget, uint32_t max_nonce, unsigned long *hashes_done);
 
diff --git a/scrypt-x64.S b/scrypt-x64.S
index 0b935c582..ab1f3ed5f 100644
--- a/scrypt-x64.S
+++ b/scrypt-x64.S
@@ -1,5 +1,5 @@
 /*
- * Copyright 2011-2012 pooler@litecoinpool.org
+ * Copyright 2011-2013 pooler@litecoinpool.org
  * All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
@@ -39,6 +39,30 @@
 scrypt_best_throughput:
 _scrypt_best_throughput:
 	pushq	%rbx
+#if defined(USE_AVX2)
+	/* Check for AVX and OSXSAVE support */
+	movl	$1, %eax
+	cpuid
+	andl	$0x18000000, %ecx
+	cmpl	$0x18000000, %ecx
+	jne scrypt_best_throughput_no_avx2
+	/* Check for AVX2 support */
+	movl	$7, %eax
+	xorl	%ecx, %ecx
+	cpuid
+	andl	$0x00000020, %ebx
+	cmpl	$0x00000020, %ebx
+	jne scrypt_best_throughput_no_avx2
+	/* Check for XMM and YMM state support */
+	xorl	%ecx, %ecx
+	xgetbv
+	andl	$0x00000006, %eax
+	cmpl	$0x00000006, %eax
+	jne scrypt_best_throughput_no_avx2
+	movl	$6, %eax
+	jmp scrypt_best_throughput_exit
+scrypt_best_throughput_no_avx2:
+#endif
 	/* Check for AuthenticAMD */
 	xorq	%rax, %rax
 	cpuid
@@ -2239,4 +2263,617 @@ scrypt_core_3way_xmm_loop2:
 	scrypt_core_3way_cleanup
 	ret
 
+
+#if defined(USE_AVX2)
+
+.macro salsa8_core_6way_avx2_doubleround
+	vpaddd	%ymm0, %ymm1, %ymm4
+	vpaddd	%ymm8, %ymm9, %ymm6
+	vpaddd	%ymm12, %ymm13, %ymm7
+	vpslld	$7, %ymm4, %ymm5
+	vpsrld	$25, %ymm4, %ymm4
+	vpxor	%ymm5, %ymm3, %ymm3
+	vpxor	%ymm4, %ymm3, %ymm3
+	vpslld	$7, %ymm6, %ymm5
+	vpsrld	$25, %ymm6, %ymm6
+	vpxor	%ymm5, %ymm11, %ymm11
+	vpxor	%ymm6, %ymm11, %ymm11
+	vpslld	$7, %ymm7, %ymm5
+	vpsrld	$25, %ymm7, %ymm7
+	vpxor	%ymm5, %ymm15, %ymm15
+	vpxor	%ymm7, %ymm15, %ymm15
+	
+	vpaddd	%ymm3, %ymm0, %ymm4
+	vpaddd	%ymm11, %ymm8, %ymm6
+	vpaddd	%ymm15, %ymm12, %ymm7
+	vpslld	$9, %ymm4, %ymm5
+	vpsrld	$23, %ymm4, %ymm4
+	vpxor	%ymm5, %ymm2, %ymm2
+	vpxor	%ymm4, %ymm2, %ymm2
+	vpslld	$9, %ymm6, %ymm5
+	vpsrld	$23, %ymm6, %ymm6
+	vpxor	%ymm5, %ymm10, %ymm10
+	vpxor	%ymm6, %ymm10, %ymm10
+	vpslld	$9, %ymm7, %ymm5
+	vpsrld	$23, %ymm7, %ymm7
+	vpxor	%ymm5, %ymm14, %ymm14
+	vpxor	%ymm7, %ymm14, %ymm14
+	
+	vpaddd	%ymm2, %ymm3, %ymm4
+	vpaddd	%ymm10, %ymm11, %ymm6
+	vpaddd	%ymm14, %ymm15, %ymm7
+	vpslld	$13, %ymm4, %ymm5
+	vpsrld	$19, %ymm4, %ymm4
+	vpshufd	$0x93, %ymm3, %ymm3
+	vpshufd	$0x93, %ymm11, %ymm11
+	vpshufd	$0x93, %ymm15, %ymm15
+	vpxor	%ymm5, %ymm1, %ymm1
+	vpxor	%ymm4, %ymm1, %ymm1
+	vpslld	$13, %ymm6, %ymm5
+	vpsrld	$19, %ymm6, %ymm6
+	vpxor	%ymm5, %ymm9, %ymm9
+	vpxor	%ymm6, %ymm9, %ymm9
+	vpslld	$13, %ymm7, %ymm5
+	vpsrld	$19, %ymm7, %ymm7
+	vpxor	%ymm5, %ymm13, %ymm13
+	vpxor	%ymm7, %ymm13, %ymm13
+	
+	vpaddd	%ymm1, %ymm2, %ymm4
+	vpaddd	%ymm9, %ymm10, %ymm6
+	vpaddd	%ymm13, %ymm14, %ymm7
+	vpslld	$18, %ymm4, %ymm5
+	vpsrld	$14, %ymm4, %ymm4
+	vpshufd	$0x4e, %ymm2, %ymm2
+	vpshufd	$0x4e, %ymm10, %ymm10
+	vpshufd	$0x4e, %ymm14, %ymm14
+	vpxor	%ymm5, %ymm0, %ymm0
+	vpxor	%ymm4, %ymm0, %ymm0
+	vpslld	$18, %ymm6, %ymm5
+	vpsrld	$14, %ymm6, %ymm6
+	vpxor	%ymm5, %ymm8, %ymm8
+	vpxor	%ymm6, %ymm8, %ymm8
+	vpslld	$18, %ymm7, %ymm5
+	vpsrld	$14, %ymm7, %ymm7
+	vpxor	%ymm5, %ymm12, %ymm12
+	vpxor	%ymm7, %ymm12, %ymm12
+	
+	vpaddd	%ymm0, %ymm3, %ymm4
+	vpaddd	%ymm8, %ymm11, %ymm6
+	vpaddd	%ymm12, %ymm15, %ymm7
+	vpslld	$7, %ymm4, %ymm5
+	vpsrld	$25, %ymm4, %ymm4
+	vpshufd	$0x39, %ymm1, %ymm1
+	vpxor	%ymm5, %ymm1, %ymm1
+	vpxor	%ymm4, %ymm1, %ymm1
+	vpslld	$7, %ymm6, %ymm5
+	vpsrld	$25, %ymm6, %ymm6
+	vpshufd	$0x39, %ymm9, %ymm9
+	vpxor	%ymm5, %ymm9, %ymm9
+	vpxor	%ymm6, %ymm9, %ymm9
+	vpslld	$7, %ymm7, %ymm5
+	vpsrld	$25, %ymm7, %ymm7
+	vpshufd	$0x39, %ymm13, %ymm13
+	vpxor	%ymm5, %ymm13, %ymm13
+	vpxor	%ymm7, %ymm13, %ymm13
+	
+	vpaddd	%ymm1, %ymm0, %ymm4
+	vpaddd	%ymm9, %ymm8, %ymm6
+	vpaddd	%ymm13, %ymm12, %ymm7
+	vpslld	$9, %ymm4, %ymm5
+	vpsrld	$23, %ymm4, %ymm4
+	vpxor	%ymm5, %ymm2, %ymm2
+	vpxor	%ymm4, %ymm2, %ymm2
+	vpslld	$9, %ymm6, %ymm5
+	vpsrld	$23, %ymm6, %ymm6
+	vpxor	%ymm5, %ymm10, %ymm10
+	vpxor	%ymm6, %ymm10, %ymm10
+	vpslld	$9, %ymm7, %ymm5
+	vpsrld	$23, %ymm7, %ymm7
+	vpxor	%ymm5, %ymm14, %ymm14
+	vpxor	%ymm7, %ymm14, %ymm14
+	
+	vpaddd	%ymm2, %ymm1, %ymm4
+	vpaddd	%ymm10, %ymm9, %ymm6
+	vpaddd	%ymm14, %ymm13, %ymm7
+	vpslld	$13, %ymm4, %ymm5
+	vpsrld	$19, %ymm4, %ymm4
+	vpshufd	$0x93, %ymm1, %ymm1
+	vpshufd	$0x93, %ymm9, %ymm9
+	vpshufd	$0x93, %ymm13, %ymm13
+	vpxor	%ymm5, %ymm3, %ymm3
+	vpxor	%ymm4, %ymm3, %ymm3
+	vpslld	$13, %ymm6, %ymm5
+	vpsrld	$19, %ymm6, %ymm6
+	vpxor	%ymm5, %ymm11, %ymm11
+	vpxor	%ymm6, %ymm11, %ymm11
+	vpslld	$13, %ymm7, %ymm5
+	vpsrld	$19, %ymm7, %ymm7
+	vpxor	%ymm5, %ymm15, %ymm15
+	vpxor	%ymm7, %ymm15, %ymm15
+	
+	vpaddd	%ymm3, %ymm2, %ymm4
+	vpaddd	%ymm11, %ymm10, %ymm6
+	vpaddd	%ymm15, %ymm14, %ymm7
+	vpslld	$18, %ymm4, %ymm5
+	vpsrld	$14, %ymm4, %ymm4
+	vpshufd	$0x4e, %ymm2, %ymm2
+	vpshufd	$0x4e, %ymm10, %ymm10
+	vpxor	%ymm5, %ymm0, %ymm0
+	vpxor	%ymm4, %ymm0, %ymm0
+	vpslld	$18, %ymm6, %ymm5
+	vpsrld	$14, %ymm6, %ymm6
+	vpshufd	$0x4e, %ymm14, %ymm14
+	vpshufd	$0x39, %ymm11, %ymm11
+	vpxor	%ymm5, %ymm8, %ymm8
+	vpxor	%ymm6, %ymm8, %ymm8
+	vpslld	$18, %ymm7, %ymm5
+	vpsrld	$14, %ymm7, %ymm7
+	vpshufd	$0x39, %ymm3, %ymm3
+	vpshufd	$0x39, %ymm15, %ymm15
+	vpxor	%ymm5, %ymm12, %ymm12
+	vpxor	%ymm7, %ymm12, %ymm12
+.endm
+
+.macro salsa8_core_6way_avx2
+	salsa8_core_6way_avx2_doubleround
+	salsa8_core_6way_avx2_doubleround
+	salsa8_core_6way_avx2_doubleround
+	salsa8_core_6way_avx2_doubleround
+.endm
+	
+	.text
+	.p2align 6
+	.globl scrypt_core_6way
+	.globl _scrypt_core_6way
+scrypt_core_6way:
+_scrypt_core_6way:
+	pushq	%rbx
+	pushq	%rbp
+#if defined(WIN64)
+	subq	$176, %rsp
+	vmovdqa	%xmm6, 8(%rsp)
+	vmovdqa	%xmm7, 24(%rsp)
+	vmovdqa	%xmm8, 40(%rsp)
+	vmovdqa	%xmm9, 56(%rsp)
+	vmovdqa	%xmm10, 72(%rsp)
+	vmovdqa	%xmm11, 88(%rsp)
+	vmovdqa	%xmm12, 104(%rsp)
+	vmovdqa	%xmm13, 120(%rsp)
+	vmovdqa	%xmm14, 136(%rsp)
+	vmovdqa	%xmm15, 152(%rsp)
+	pushq	%rdi
+	pushq	%rsi
+	movq	%rcx, %rdi
+	movq	%rdx, %rsi
+#endif
+	movq	%rsp, %rdx
+	subq	$768, %rsp
+	andq	$-128, %rsp
+	
+.macro scrypt_core_6way_cleanup
+	movq	%rdx, %rsp
+#if defined(WIN64)
+	popq	%rsi
+	popq	%rdi
+	vmovdqa	8(%rsp), %xmm6
+	vmovdqa	24(%rsp), %xmm7
+	vmovdqa	40(%rsp), %xmm8
+	vmovdqa	56(%rsp), %xmm9
+	vmovdqa	72(%rsp), %xmm10
+	vmovdqa	88(%rsp), %xmm11
+	vmovdqa	104(%rsp), %xmm12
+	vmovdqa	120(%rsp), %xmm13
+	vmovdqa	136(%rsp), %xmm14
+	vmovdqa	152(%rsp), %xmm15
+	addq	$176, %rsp
+#endif
+	popq	%rbp
+	popq	%rbx
+.endm
+
+.macro scrypt_shuffle_pack2 src, so, dest, do
+	vmovdqa	\so+0*16(\src), %xmm0
+	vmovdqa	\so+1*16(\src), %xmm1
+	vmovdqa	\so+2*16(\src), %xmm2
+	vmovdqa	\so+3*16(\src), %xmm3
+	vinserti128	$1, \so+128+0*16(\src), %ymm0, %ymm0
+	vinserti128	$1, \so+128+1*16(\src), %ymm1, %ymm1
+	vinserti128	$1, \so+128+2*16(\src), %ymm2, %ymm2
+	vinserti128	$1, \so+128+3*16(\src), %ymm3, %ymm3
+	vpblendd	$0x33, %ymm0, %ymm2, %ymm4
+	vpblendd	$0xcc, %ymm1, %ymm3, %ymm5
+	vpblendd	$0x33, %ymm2, %ymm0, %ymm6
+	vpblendd	$0xcc, %ymm3, %ymm1, %ymm7
+	vpblendd	$0x55, %ymm7, %ymm6, %ymm3
+	vpblendd	$0x55, %ymm6, %ymm5, %ymm2
+	vpblendd	$0x55, %ymm5, %ymm4, %ymm1
+	vpblendd	$0x55, %ymm4, %ymm7, %ymm0
+	vmovdqa	%ymm0, \do+0*32(\dest)
+	vmovdqa	%ymm1, \do+1*32(\dest)
+	vmovdqa	%ymm2, \do+2*32(\dest)
+	vmovdqa	%ymm3, \do+3*32(\dest)
+.endm
+
+.macro scrypt_shuffle_unpack2 src, so, dest, do
+	vmovdqa	\so+0*32(\src), %ymm0
+	vmovdqa	\so+1*32(\src), %ymm1
+	vmovdqa	\so+2*32(\src), %ymm2
+	vmovdqa	\so+3*32(\src), %ymm3
+	vpblendd	$0x33, %ymm0, %ymm2, %ymm4
+	vpblendd	$0xcc, %ymm1, %ymm3, %ymm5
+	vpblendd	$0x33, %ymm2, %ymm0, %ymm6
+	vpblendd	$0xcc, %ymm3, %ymm1, %ymm7
+	vpblendd	$0x55, %ymm7, %ymm6, %ymm3
+	vpblendd	$0x55, %ymm6, %ymm5, %ymm2
+	vpblendd	$0x55, %ymm5, %ymm4, %ymm1
+	vpblendd	$0x55, %ymm4, %ymm7, %ymm0
+	vmovdqa	%xmm0, \do+0*16(\dest)
+	vmovdqa	%xmm1, \do+1*16(\dest)
+	vmovdqa	%xmm2, \do+2*16(\dest)
+	vmovdqa	%xmm3, \do+3*16(\dest)
+	vextracti128	$1, %ymm0, \do+128+0*16(\dest)
+	vextracti128	$1, %ymm1, \do+128+1*16(\dest)
+	vextracti128	$1, %ymm2, \do+128+2*16(\dest)
+	vextracti128	$1, %ymm3, \do+128+3*16(\dest)
+.endm
+	
+scrypt_core_6way_avx2:
+	scrypt_shuffle_pack2 %rdi, 0*256+0, %rsp, 0*128
+	scrypt_shuffle_pack2 %rdi, 0*256+64, %rsp, 1*128
+	scrypt_shuffle_pack2 %rdi, 1*256+0, %rsp, 2*128
+	scrypt_shuffle_pack2 %rdi, 1*256+64, %rsp, 3*128
+	scrypt_shuffle_pack2 %rdi, 2*256+0, %rsp, 4*128
+	scrypt_shuffle_pack2 %rdi, 2*256+64, %rsp, 5*128
+	
+	vmovdqa	0*256+4*32(%rsp), %ymm0
+	vmovdqa	0*256+5*32(%rsp), %ymm1
+	vmovdqa	0*256+6*32(%rsp), %ymm2
+	vmovdqa	0*256+7*32(%rsp), %ymm3
+	vmovdqa	1*256+4*32(%rsp), %ymm8
+	vmovdqa	1*256+5*32(%rsp), %ymm9
+	vmovdqa	1*256+6*32(%rsp), %ymm10
+	vmovdqa	1*256+7*32(%rsp), %ymm11
+	vmovdqa	2*256+4*32(%rsp), %ymm12
+	vmovdqa	2*256+5*32(%rsp), %ymm13
+	vmovdqa	2*256+6*32(%rsp), %ymm14
+	vmovdqa	2*256+7*32(%rsp), %ymm15
+	
+	movq	%rsi, %rbx
+	leaq	6*131072(%rsi), %rax
+scrypt_core_6way_avx2_loop1:
+	vmovdqa	%ymm0, 0*256+4*32(%rbx)
+	vmovdqa	%ymm1, 0*256+5*32(%rbx)
+	vmovdqa	%ymm2, 0*256+6*32(%rbx)
+	vmovdqa	%ymm3, 0*256+7*32(%rbx)
+	vpxor	0*256+0*32(%rsp), %ymm0, %ymm0
+	vpxor	0*256+1*32(%rsp), %ymm1, %ymm1
+	vpxor	0*256+2*32(%rsp), %ymm2, %ymm2
+	vpxor	0*256+3*32(%rsp), %ymm3, %ymm3
+	vmovdqa	%ymm8, 1*256+4*32(%rbx)
+	vmovdqa	%ymm9, 1*256+5*32(%rbx)
+	vmovdqa	%ymm10, 1*256+6*32(%rbx)
+	vmovdqa	%ymm11, 1*256+7*32(%rbx)
+	vpxor	1*256+0*32(%rsp), %ymm8, %ymm8
+	vpxor	1*256+1*32(%rsp), %ymm9, %ymm9
+	vpxor	1*256+2*32(%rsp), %ymm10, %ymm10
+	vpxor	1*256+3*32(%rsp), %ymm11, %ymm11
+	vmovdqa	%ymm12, 2*256+4*32(%rbx)
+	vmovdqa	%ymm13, 2*256+5*32(%rbx)
+	vmovdqa	%ymm14, 2*256+6*32(%rbx)
+	vmovdqa	%ymm15, 2*256+7*32(%rbx)
+	vpxor	2*256+0*32(%rsp), %ymm12, %ymm12
+	vpxor	2*256+1*32(%rsp), %ymm13, %ymm13
+	vpxor	2*256+2*32(%rsp), %ymm14, %ymm14
+	vpxor	2*256+3*32(%rsp), %ymm15, %ymm15
+	vmovdqa	%ymm0, 0*256+0*32(%rbx)
+	vmovdqa	%ymm1, 0*256+1*32(%rbx)
+	vmovdqa	%ymm2, 0*256+2*32(%rbx)
+	vmovdqa	%ymm3, 0*256+3*32(%rbx)
+	vmovdqa	%ymm8, 1*256+0*32(%rbx)
+	vmovdqa	%ymm9, 1*256+1*32(%rbx)
+	vmovdqa	%ymm10, 1*256+2*32(%rbx)
+	vmovdqa	%ymm11, 1*256+3*32(%rbx)
+	vmovdqa	%ymm12, 2*256+0*32(%rbx)
+	vmovdqa	%ymm13, 2*256+1*32(%rbx)
+	vmovdqa	%ymm14, 2*256+2*32(%rbx)
+	vmovdqa	%ymm15, 2*256+3*32(%rbx)
+	
+	salsa8_core_6way_avx2
+	vpaddd	0*256+0*32(%rbx), %ymm0, %ymm0
+	vpaddd	0*256+1*32(%rbx), %ymm1, %ymm1
+	vpaddd	0*256+2*32(%rbx), %ymm2, %ymm2
+	vpaddd	0*256+3*32(%rbx), %ymm3, %ymm3
+	vpaddd	1*256+0*32(%rbx), %ymm8, %ymm8
+	vpaddd	1*256+1*32(%rbx), %ymm9, %ymm9
+	vpaddd	1*256+2*32(%rbx), %ymm10, %ymm10
+	vpaddd	1*256+3*32(%rbx), %ymm11, %ymm11
+	vpaddd	2*256+0*32(%rbx), %ymm12, %ymm12
+	vpaddd	2*256+1*32(%rbx), %ymm13, %ymm13
+	vpaddd	2*256+2*32(%rbx), %ymm14, %ymm14
+	vpaddd	2*256+3*32(%rbx), %ymm15, %ymm15
+	vmovdqa	%ymm0, 0*256+0*32(%rsp)
+	vmovdqa	%ymm1, 0*256+1*32(%rsp)
+	vmovdqa	%ymm2, 0*256+2*32(%rsp)
+	vmovdqa	%ymm3, 0*256+3*32(%rsp)
+	vmovdqa	%ymm8, 1*256+0*32(%rsp)
+	vmovdqa	%ymm9, 1*256+1*32(%rsp)
+	vmovdqa	%ymm10, 1*256+2*32(%rsp)
+	vmovdqa	%ymm11, 1*256+3*32(%rsp)
+	vmovdqa	%ymm12, 2*256+0*32(%rsp)
+	vmovdqa	%ymm13, 2*256+1*32(%rsp)
+	vmovdqa	%ymm14, 2*256+2*32(%rsp)
+	vmovdqa	%ymm15, 2*256+3*32(%rsp)
+	
+	vpxor	0*256+4*32(%rbx), %ymm0, %ymm0
+	vpxor	0*256+5*32(%rbx), %ymm1, %ymm1
+	vpxor	0*256+6*32(%rbx), %ymm2, %ymm2
+	vpxor	0*256+7*32(%rbx), %ymm3, %ymm3
+	vpxor	1*256+4*32(%rbx), %ymm8, %ymm8
+	vpxor	1*256+5*32(%rbx), %ymm9, %ymm9
+	vpxor	1*256+6*32(%rbx), %ymm10, %ymm10
+	vpxor	1*256+7*32(%rbx), %ymm11, %ymm11
+	vpxor	2*256+4*32(%rbx), %ymm12, %ymm12
+	vpxor	2*256+5*32(%rbx), %ymm13, %ymm13
+	vpxor	2*256+6*32(%rbx), %ymm14, %ymm14
+	vpxor	2*256+7*32(%rbx), %ymm15, %ymm15
+	vmovdqa	%ymm0, 0*256+4*32(%rsp)
+	vmovdqa	%ymm1, 0*256+5*32(%rsp)
+	vmovdqa	%ymm2, 0*256+6*32(%rsp)
+	vmovdqa	%ymm3, 0*256+7*32(%rsp)
+	vmovdqa	%ymm8, 1*256+4*32(%rsp)
+	vmovdqa	%ymm9, 1*256+5*32(%rsp)
+	vmovdqa	%ymm10, 1*256+6*32(%rsp)
+	vmovdqa	%ymm11, 1*256+7*32(%rsp)
+	vmovdqa	%ymm12, 2*256+4*32(%rsp)
+	vmovdqa	%ymm13, 2*256+5*32(%rsp)
+	vmovdqa	%ymm14, 2*256+6*32(%rsp)
+	vmovdqa	%ymm15, 2*256+7*32(%rsp)
+	salsa8_core_6way_avx2
+	vpaddd	0*256+4*32(%rsp), %ymm0, %ymm0
+	vpaddd	0*256+5*32(%rsp), %ymm1, %ymm1
+	vpaddd	0*256+6*32(%rsp), %ymm2, %ymm2
+	vpaddd	0*256+7*32(%rsp), %ymm3, %ymm3
+	vpaddd	1*256+4*32(%rsp), %ymm8, %ymm8
+	vpaddd	1*256+5*32(%rsp), %ymm9, %ymm9
+	vpaddd	1*256+6*32(%rsp), %ymm10, %ymm10
+	vpaddd	1*256+7*32(%rsp), %ymm11, %ymm11
+	vpaddd	2*256+4*32(%rsp), %ymm12, %ymm12
+	vpaddd	2*256+5*32(%rsp), %ymm13, %ymm13
+	vpaddd	2*256+6*32(%rsp), %ymm14, %ymm14
+	vpaddd	2*256+7*32(%rsp), %ymm15, %ymm15
+	
+	addq	$6*128, %rbx
+	cmpq	%rax, %rbx
+	jne scrypt_core_6way_avx2_loop1
+	
+	vmovdqa	%ymm0, 0*256+4*32(%rsp)
+	vmovdqa	%ymm1, 0*256+5*32(%rsp)
+	vmovdqa	%ymm2, 0*256+6*32(%rsp)
+	vmovdqa	%ymm3, 0*256+7*32(%rsp)
+	vmovdqa	%ymm8, 1*256+4*32(%rsp)
+	vmovdqa	%ymm9, 1*256+5*32(%rsp)
+	vmovdqa	%ymm10, 1*256+6*32(%rsp)
+	vmovdqa	%ymm11, 1*256+7*32(%rsp)
+	vmovdqa	%ymm12, 2*256+4*32(%rsp)
+	vmovdqa	%ymm13, 2*256+5*32(%rsp)
+	vmovdqa	%ymm14, 2*256+6*32(%rsp)
+	vmovdqa	%ymm15, 2*256+7*32(%rsp)
+	
+	movq	$1024, %rcx
+scrypt_core_6way_avx2_loop2:
+	vmovd	%xmm0, %ebp
+	vmovd	%xmm8, %ebx
+	vmovd	%xmm12, %eax
+	vextracti128	$1, %ymm0, %xmm4
+	vextracti128	$1, %ymm8, %xmm5
+	vextracti128	$1, %ymm12, %xmm6
+	vmovd	%xmm4, %r8d
+	vmovd	%xmm5, %r9d
+	vmovd	%xmm6, %r10d
+	vpxor	0*256+0*32(%rsp), %ymm0, %ymm0
+	vpxor	0*256+1*32(%rsp), %ymm1, %ymm1
+	vpxor	0*256+2*32(%rsp), %ymm2, %ymm2
+	vpxor	0*256+3*32(%rsp), %ymm3, %ymm3
+	vpxor	1*256+0*32(%rsp), %ymm8, %ymm8
+	vpxor	1*256+1*32(%rsp), %ymm9, %ymm9
+	vpxor	1*256+2*32(%rsp), %ymm10, %ymm10
+	vpxor	1*256+3*32(%rsp), %ymm11, %ymm11
+	vpxor	2*256+0*32(%rsp), %ymm12, %ymm12
+	vpxor	2*256+1*32(%rsp), %ymm13, %ymm13
+	vpxor	2*256+2*32(%rsp), %ymm14, %ymm14
+	vpxor	2*256+3*32(%rsp), %ymm15, %ymm15
+	andl	$1023, %ebp
+	leaq	0(%rbp, %rbp, 2), %rbp
+	shll	$8, %ebp
+	andl	$1023, %ebx
+	leaq	1(%rbx, %rbx, 2), %rbx
+	shll	$8, %ebx
+	andl	$1023, %eax
+	leaq	2(%rax, %rax, 2), %rax
+	shll	$8, %eax
+	andl	$1023, %r8d
+	leaq	0(%r8, %r8, 2), %r8
+	shll	$8, %r8d
+	andl	$1023, %r9d
+	leaq	1(%r9, %r9, 2), %r9
+	shll	$8, %r9d
+	andl	$1023, %r10d
+	leaq	2(%r10, %r10, 2), %r10
+	shll	$8, %r10d
+	vmovdqa	0*32(%rsi, %rbp), %xmm4
+	vinserti128	$1, 0*32+16(%rsi, %r8), %ymm4, %ymm4
+	vmovdqa	1*32(%rsi, %rbp), %xmm5
+	vinserti128	$1, 1*32+16(%rsi, %r8), %ymm5, %ymm5
+	vmovdqa	2*32(%rsi, %rbp), %xmm6
+	vinserti128	$1, 2*32+16(%rsi, %r8), %ymm6, %ymm6
+	vmovdqa	3*32(%rsi, %rbp), %xmm7
+	vinserti128	$1, 3*32+16(%rsi, %r8), %ymm7, %ymm7
+	vpxor	%ymm4, %ymm0, %ymm0
+	vpxor	%ymm5, %ymm1, %ymm1
+	vpxor	%ymm6, %ymm2, %ymm2
+	vpxor	%ymm7, %ymm3, %ymm3
+	vmovdqa	0*32(%rsi, %rbx), %xmm4
+	vinserti128	$1, 0*32+16(%rsi, %r9), %ymm4, %ymm4
+	vmovdqa	1*32(%rsi, %rbx), %xmm5
+	vinserti128	$1, 1*32+16(%rsi, %r9), %ymm5, %ymm5
+	vmovdqa	2*32(%rsi, %rbx), %xmm6
+	vinserti128	$1, 2*32+16(%rsi, %r9), %ymm6, %ymm6
+	vmovdqa	3*32(%rsi, %rbx), %xmm7
+	vinserti128	$1, 3*32+16(%rsi, %r9), %ymm7, %ymm7
+	vpxor	%ymm4, %ymm8, %ymm8
+	vpxor	%ymm5, %ymm9, %ymm9
+	vpxor	%ymm6, %ymm10, %ymm10
+	vpxor	%ymm7, %ymm11, %ymm11
+	vmovdqa	0*32(%rsi, %rax), %xmm4
+	vinserti128	$1, 0*32+16(%rsi, %r10), %ymm4, %ymm4
+	vmovdqa	1*32(%rsi, %rax), %xmm5
+	vinserti128	$1, 1*32+16(%rsi, %r10), %ymm5, %ymm5
+	vmovdqa	2*32(%rsi, %rax), %xmm6
+	vinserti128	$1, 2*32+16(%rsi, %r10), %ymm6, %ymm6
+	vmovdqa	3*32(%rsi, %rax), %xmm7
+	vinserti128	$1, 3*32+16(%rsi, %r10), %ymm7, %ymm7
+	vpxor	%ymm4, %ymm12, %ymm12
+	vpxor	%ymm5, %ymm13, %ymm13
+	vpxor	%ymm6, %ymm14, %ymm14
+	vpxor	%ymm7, %ymm15, %ymm15
+	
+	vmovdqa	%ymm0, 0*256+0*32(%rsp)
+	vmovdqa	%ymm1, 0*256+1*32(%rsp)
+	vmovdqa	%ymm2, 0*256+2*32(%rsp)
+	vmovdqa	%ymm3, 0*256+3*32(%rsp)
+	vmovdqa	%ymm8, 1*256+0*32(%rsp)
+	vmovdqa	%ymm9, 1*256+1*32(%rsp)
+	vmovdqa	%ymm10, 1*256+2*32(%rsp)
+	vmovdqa	%ymm11, 1*256+3*32(%rsp)
+	vmovdqa	%ymm12, 2*256+0*32(%rsp)
+	vmovdqa	%ymm13, 2*256+1*32(%rsp)
+	vmovdqa	%ymm14, 2*256+2*32(%rsp)
+	vmovdqa	%ymm15, 2*256+3*32(%rsp)
+	salsa8_core_6way_avx2
+	vpaddd	0*256+0*32(%rsp), %ymm0, %ymm0
+	vpaddd	0*256+1*32(%rsp), %ymm1, %ymm1
+	vpaddd	0*256+2*32(%rsp), %ymm2, %ymm2
+	vpaddd	0*256+3*32(%rsp), %ymm3, %ymm3
+	vpaddd	1*256+0*32(%rsp), %ymm8, %ymm8
+	vpaddd	1*256+1*32(%rsp), %ymm9, %ymm9
+	vpaddd	1*256+2*32(%rsp), %ymm10, %ymm10
+	vpaddd	1*256+3*32(%rsp), %ymm11, %ymm11
+	vpaddd	2*256+0*32(%rsp), %ymm12, %ymm12
+	vpaddd	2*256+1*32(%rsp), %ymm13, %ymm13
+	vpaddd	2*256+2*32(%rsp), %ymm14, %ymm14
+	vpaddd	2*256+3*32(%rsp), %ymm15, %ymm15
+	vmovdqa	%ymm0, 0*256+0*32(%rsp)
+	vmovdqa	%ymm1, 0*256+1*32(%rsp)
+	vmovdqa	%ymm2, 0*256+2*32(%rsp)
+	vmovdqa	%ymm3, 0*256+3*32(%rsp)
+	vmovdqa	%ymm8, 1*256+0*32(%rsp)
+	vmovdqa	%ymm9, 1*256+1*32(%rsp)
+	vmovdqa	%ymm10, 1*256+2*32(%rsp)
+	vmovdqa	%ymm11, 1*256+3*32(%rsp)
+	vmovdqa	%ymm12, 2*256+0*32(%rsp)
+	vmovdqa	%ymm13, 2*256+1*32(%rsp)
+	vmovdqa	%ymm14, 2*256+2*32(%rsp)
+	vmovdqa	%ymm15, 2*256+3*32(%rsp)
+	
+	vmovdqa	4*32(%rsi, %rbp), %xmm4
+	vinserti128	$1, 4*32+16(%rsi, %r8), %ymm4, %ymm4
+	vmovdqa	5*32(%rsi, %rbp), %xmm5
+	vinserti128	$1, 5*32+16(%rsi, %r8), %ymm5, %ymm5
+	vmovdqa	6*32(%rsi, %rbp), %xmm6
+	vinserti128	$1, 6*32+16(%rsi, %r8), %ymm6, %ymm6
+	vmovdqa	7*32(%rsi, %rbp), %xmm7
+	vinserti128	$1, 7*32+16(%rsi, %r8), %ymm7, %ymm7
+	vpxor	%ymm4, %ymm0, %ymm0
+	vpxor	%ymm5, %ymm1, %ymm1
+	vpxor	%ymm6, %ymm2, %ymm2
+	vpxor	%ymm7, %ymm3, %ymm3
+	vmovdqa	4*32(%rsi, %rbx), %xmm4
+	vinserti128	$1, 4*32+16(%rsi, %r9), %ymm4, %ymm4
+	vmovdqa	5*32(%rsi, %rbx), %xmm5
+	vinserti128	$1, 5*32+16(%rsi, %r9), %ymm5, %ymm5
+	vmovdqa	6*32(%rsi, %rbx), %xmm6
+	vinserti128	$1, 6*32+16(%rsi, %r9), %ymm6, %ymm6
+	vmovdqa	7*32(%rsi, %rbx), %xmm7
+	vinserti128	$1, 7*32+16(%rsi, %r9), %ymm7, %ymm7
+	vpxor	%ymm4, %ymm8, %ymm8
+	vpxor	%ymm5, %ymm9, %ymm9
+	vpxor	%ymm6, %ymm10, %ymm10
+	vpxor	%ymm7, %ymm11, %ymm11
+	vmovdqa	4*32(%rsi, %rax), %xmm4
+	vinserti128	$1, 4*32+16(%rsi, %r10), %ymm4, %ymm4
+	vmovdqa	5*32(%rsi, %rax), %xmm5
+	vinserti128	$1, 5*32+16(%rsi, %r10), %ymm5, %ymm5
+	vmovdqa	6*32(%rsi, %rax), %xmm6
+	vinserti128	$1, 6*32+16(%rsi, %r10), %ymm6, %ymm6
+	vmovdqa	7*32(%rsi, %rax), %xmm7
+	vinserti128	$1, 7*32+16(%rsi, %r10), %ymm7, %ymm7
+	vpxor	%ymm4, %ymm12, %ymm12
+	vpxor	%ymm5, %ymm13, %ymm13
+	vpxor	%ymm6, %ymm14, %ymm14
+	vpxor	%ymm7, %ymm15, %ymm15
+	vpxor	0*256+4*32(%rsp), %ymm0, %ymm0
+	vpxor	0*256+5*32(%rsp), %ymm1, %ymm1
+	vpxor	0*256+6*32(%rsp), %ymm2, %ymm2
+	vpxor	0*256+7*32(%rsp), %ymm3, %ymm3
+	vpxor	1*256+4*32(%rsp), %ymm8, %ymm8
+	vpxor	1*256+5*32(%rsp), %ymm9, %ymm9
+	vpxor	1*256+6*32(%rsp), %ymm10, %ymm10
+	vpxor	1*256+7*32(%rsp), %ymm11, %ymm11
+	vpxor	2*256+4*32(%rsp), %ymm12, %ymm12
+	vpxor	2*256+5*32(%rsp), %ymm13, %ymm13
+	vpxor	2*256+6*32(%rsp), %ymm14, %ymm14
+	vpxor	2*256+7*32(%rsp), %ymm15, %ymm15
+	vmovdqa	%ymm0, 0*256+4*32(%rsp)
+	vmovdqa	%ymm1, 0*256+5*32(%rsp)
+	vmovdqa	%ymm2, 0*256+6*32(%rsp)
+	vmovdqa	%ymm3, 0*256+7*32(%rsp)
+	vmovdqa	%ymm8, 1*256+4*32(%rsp)
+	vmovdqa	%ymm9, 1*256+5*32(%rsp)
+	vmovdqa	%ymm10, 1*256+6*32(%rsp)
+	vmovdqa	%ymm11, 1*256+7*32(%rsp)
+	vmovdqa	%ymm12, 2*256+4*32(%rsp)
+	vmovdqa	%ymm13, 2*256+5*32(%rsp)
+	vmovdqa	%ymm14, 2*256+6*32(%rsp)
+	vmovdqa	%ymm15, 2*256+7*32(%rsp)
+	salsa8_core_6way_avx2
+	vpaddd	0*256+4*32(%rsp), %ymm0, %ymm0
+	vpaddd	0*256+5*32(%rsp), %ymm1, %ymm1
+	vpaddd	0*256+6*32(%rsp), %ymm2, %ymm2
+	vpaddd	0*256+7*32(%rsp), %ymm3, %ymm3
+	vpaddd	1*256+4*32(%rsp), %ymm8, %ymm8
+	vpaddd	1*256+5*32(%rsp), %ymm9, %ymm9
+	vpaddd	1*256+6*32(%rsp), %ymm10, %ymm10
+	vpaddd	1*256+7*32(%rsp), %ymm11, %ymm11
+	vpaddd	2*256+4*32(%rsp), %ymm12, %ymm12
+	vpaddd	2*256+5*32(%rsp), %ymm13, %ymm13
+	vpaddd	2*256+6*32(%rsp), %ymm14, %ymm14
+	vpaddd	2*256+7*32(%rsp), %ymm15, %ymm15
+	vmovdqa	%ymm0, 0*256+4*32(%rsp)
+	vmovdqa	%ymm1, 0*256+5*32(%rsp)
+	vmovdqa	%ymm2, 0*256+6*32(%rsp)
+	vmovdqa	%ymm3, 0*256+7*32(%rsp)
+	vmovdqa	%ymm8, 1*256+4*32(%rsp)
+	vmovdqa	%ymm9, 1*256+5*32(%rsp)
+	vmovdqa	%ymm10, 1*256+6*32(%rsp)
+	vmovdqa	%ymm11, 1*256+7*32(%rsp)
+	vmovdqa	%ymm12, 2*256+4*32(%rsp)
+	vmovdqa	%ymm13, 2*256+5*32(%rsp)
+	vmovdqa	%ymm14, 2*256+6*32(%rsp)
+	vmovdqa	%ymm15, 2*256+7*32(%rsp)
+	
+	subq	$1, %rcx
+	ja scrypt_core_6way_avx2_loop2
+	
+	scrypt_shuffle_unpack2 %rsp, 0*128, %rdi, 0*256+0
+	scrypt_shuffle_unpack2 %rsp, 1*128, %rdi, 0*256+64
+	scrypt_shuffle_unpack2 %rsp, 2*128, %rdi, 1*256+0
+	scrypt_shuffle_unpack2 %rsp, 3*128, %rdi, 1*256+64
+	scrypt_shuffle_unpack2 %rsp, 4*128, %rdi, 2*256+0
+	scrypt_shuffle_unpack2 %rsp, 5*128, %rdi, 2*256+64
+	
+	scrypt_core_6way_cleanup
+	ret
+
+#endif /* USE_AVX2 */
+
 #endif
diff --git a/scrypt.c b/scrypt.c
index bc4f512ed..06fd76a19 100644
--- a/scrypt.c
+++ b/scrypt.c
@@ -1,5 +1,5 @@
 /*
- * Copyright 2009 Colin Percival, 2011 ArtForz, 2011-2012 pooler
+ * Copyright 2009 Colin Percival, 2011 ArtForz, 2011-2013 pooler
  * All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
@@ -256,6 +256,128 @@ static inline void PBKDF2_SHA256_128_32_4way(uint32_t *tstate,
 #endif /* HAVE_SHA256_4WAY */
 
 
+#ifdef HAVE_SHA256_8WAY
+
+static const uint32_t finalblk_8way[8 * 16] __attribute__((aligned(32))) = {
+	0x00000001, 0x00000001, 0x00000001, 0x00000001, 0x00000001, 0x00000001, 0x00000001, 0x00000001,
+	0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000,
+	0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
+	0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
+	0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
+	0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
+	0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
+	0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
+	0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
+	0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
+	0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
+	0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
+	0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
+	0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
+	0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
+	0x00000620, 0x00000620, 0x00000620, 0x00000620, 0x00000620, 0x00000620, 0x00000620, 0x00000620
+};
+
+static inline void HMAC_SHA256_80_init_8way(const uint32_t *key,
+	uint32_t *tstate, uint32_t *ostate)
+{
+	uint32_t ihash[8 * 8] __attribute__((aligned(32)));
+	uint32_t pad[8 * 16] __attribute__((aligned(32)));
+	int i;
+	
+	/* tstate is assumed to contain the midstate of key */
+	memcpy(pad, key + 8 * 16, 8 * 16);
+	for (i = 0; i < 8; i++)
+		pad[8 * 4 + i] = 0x80000000;
+	memset(pad + 8 * 5, 0x00, 8 * 40);
+	for (i = 0; i < 8; i++)
+		pad[8 * 15 + i] = 0x00000280;
+	sha256_transform_8way(tstate, pad, 0);
+	memcpy(ihash, tstate, 8 * 32);
+	
+	sha256_init_8way(ostate);
+	for (i = 0; i < 8 * 8; i++)
+		pad[i] = ihash[i] ^ 0x5c5c5c5c;
+	for (; i < 8 * 16; i++)
+		pad[i] = 0x5c5c5c5c;
+	sha256_transform_8way(ostate, pad, 0);
+	
+	sha256_init_8way(tstate);
+	for (i = 0; i < 8 * 8; i++)
+		pad[i] = ihash[i] ^ 0x36363636;
+	for (; i < 8 * 16; i++)
+		pad[i] = 0x36363636;
+	sha256_transform_8way(tstate, pad, 0);
+}
+
+static inline void PBKDF2_SHA256_80_128_8way(const uint32_t *tstate,
+	const uint32_t *ostate, const uint32_t *salt, uint32_t *output)
+{
+	uint32_t istate[8 * 8] __attribute__((aligned(32)));
+	uint32_t ostate2[8 * 8] __attribute__((aligned(32)));
+	uint32_t ibuf[8 * 16] __attribute__((aligned(32)));
+	uint32_t obuf[8 * 16] __attribute__((aligned(32)));
+	int i, j;
+	
+	memcpy(istate, tstate, 8 * 32);
+	sha256_transform_8way(istate, salt, 0);
+	
+	memcpy(ibuf, salt + 8 * 16, 8 * 16);
+	for (i = 0; i < 8; i++)
+		ibuf[8 * 5 + i] = 0x80000000;
+	memset(ibuf + 8 * 6, 0x00, 8 * 36);
+	for (i = 0; i < 8; i++)
+		ibuf[8 * 15 + i] = 0x000004a0;
+	
+	for (i = 0; i < 8; i++)
+		obuf[8 * 8 + i] = 0x80000000;
+	memset(obuf + 8 * 9, 0x00, 8 * 24);
+	for (i = 0; i < 8; i++)
+		obuf[8 * 15 + i] = 0x00000300;
+	
+	for (i = 0; i < 4; i++) {
+		memcpy(obuf, istate, 8 * 32);
+		ibuf[8 * 4 + 0] = i + 1;
+		ibuf[8 * 4 + 1] = i + 1;
+		ibuf[8 * 4 + 2] = i + 1;
+		ibuf[8 * 4 + 3] = i + 1;
+		ibuf[8 * 4 + 4] = i + 1;
+		ibuf[8 * 4 + 5] = i + 1;
+		ibuf[8 * 4 + 6] = i + 1;
+		ibuf[8 * 4 + 7] = i + 1;
+		sha256_transform_8way(obuf, ibuf, 0);
+		
+		memcpy(ostate2, ostate, 8 * 32);
+		sha256_transform_8way(ostate2, obuf, 0);
+		for (j = 0; j < 8 * 8; j++)
+			output[8 * 8 * i + j] = swab32(ostate2[j]);
+	}
+}
+
+static inline void PBKDF2_SHA256_128_32_8way(uint32_t *tstate,
+	uint32_t *ostate, const uint32_t *salt, uint32_t *output)
+{
+	uint32_t buf[8 * 16] __attribute__((aligned(32)));
+	int i;
+	
+	sha256_transform_8way(tstate, salt, 1);
+	sha256_transform_8way(tstate, salt + 8 * 16, 1);
+	sha256_transform_8way(tstate, finalblk_8way, 0);
+	
+	memcpy(buf, tstate, 8 * 32);
+	for (i = 0; i < 8; i++)
+		buf[8 * 8 + i] = 0x80000000;
+	memset(buf + 8 * 9, 0x00, 8 * 24);
+	for (i = 0; i < 8; i++)
+		buf[8 * 15 + i] = 0x00000300;
+	sha256_transform_8way(ostate, buf, 0);
+	
+	for (i = 0; i < 8 * 8; i++)
+		output[i] = swab32(ostate[i]);
+}
+
+#endif /* HAVE_SHA256_8WAY */
+
+
 #if defined(__x86_64__)
 
 #define SCRYPT_MAX_WAYS 12
@@ -263,6 +385,12 @@ static inline void PBKDF2_SHA256_128_32_4way(uint32_t *tstate,
 int scrypt_best_throughput();
 void scrypt_core(uint32_t *X, uint32_t *V);
 void scrypt_core_3way(uint32_t *X, uint32_t *V);
+#if defined(USE_AVX2)
+#undef SCRYPT_MAX_WAYS
+#define SCRYPT_MAX_WAYS 24
+#define HAVE_SCRYPT_6WAY 1
+void scrypt_core_6way(uint32_t *X, uint32_t *V);
+#endif
 
 #elif defined(__i386__)
 
@@ -410,47 +538,32 @@ static void scrypt_1024_1_1_256_4way(const uint32_t *input,
 	uint32_t W[4 * 32] __attribute__((aligned(128)));
 	uint32_t X[4 * 32] __attribute__((aligned(128)));
 	uint32_t *V;
-	int i;
+	int i, k;
 	
 	V = (uint32_t *)(((uintptr_t)(scratchpad) + 63) & ~ (uintptr_t)(63));
 
-	for (i = 0; i < 20; i++) {
-		W[4 * i + 0] = input[0 * 20 + i];
-		W[4 * i + 1] = input[1 * 20 + i];
-		W[4 * i + 2] = input[2 * 20 + i];
-		W[4 * i + 3] = input[3 * 20 + i];
-	}
-	for (i = 0; i < 8; i++) {
-		tstate[4 * i + 0] = midstate[i];
-		tstate[4 * i + 1] = midstate[i];
-		tstate[4 * i + 2] = midstate[i];
-		tstate[4 * i + 3] = midstate[i];
-	}
+	for (i = 0; i < 20; i++)
+		for (k = 0; k < 4; k++)
+			W[4 * i + k] = input[k * 20 + i];
+	for (i = 0; i < 8; i++)
+		for (k = 0; k < 4; k++)
+			tstate[4 * i + k] = midstate[i];
 	HMAC_SHA256_80_init_4way(W, tstate, ostate);
 	PBKDF2_SHA256_80_128_4way(tstate, ostate, W, W);
-	for (i = 0; i < 32; i++) {
-		X[0 * 32 + i] = W[4 * i + 0];
-		X[1 * 32 + i] = W[4 * i + 1];
-		X[2 * 32 + i] = W[4 * i + 2];
-		X[3 * 32 + i] = W[4 * i + 3];
-	}
+	for (i = 0; i < 32; i++)
+		for (k = 0; k < 4; k++)
+			X[k * 32 + i] = W[4 * i + k];
 	scrypt_core(X + 0 * 32, V);
 	scrypt_core(X + 1 * 32, V);
 	scrypt_core(X + 2 * 32, V);
 	scrypt_core(X + 3 * 32, V);
-	for (i = 0; i < 32; i++) {
-		W[4 * i + 0] = X[0 * 32 + i];
-		W[4 * i + 1] = X[1 * 32 + i];
-		W[4 * i + 2] = X[2 * 32 + i];
-		W[4 * i + 3] = X[3 * 32 + i];
-	}
+	for (i = 0; i < 32; i++)
+		for (k = 0; k < 4; k++)
+			W[4 * i + k] = X[k * 32 + i];
 	PBKDF2_SHA256_128_32_4way(tstate, ostate, W, W);
-	for (i = 0; i < 8; i++) {
-		output[0 * 8 + i] = W[4 * i + 0];
-		output[1 * 8 + i] = W[4 * i + 1];
-		output[2 * 8 + i] = W[4 * i + 2];
-		output[3 * 8 + i] = W[4 * i + 3];
-	}
+	for (i = 0; i < 8; i++)
+		for (k = 0; k < 4; k++)
+			output[k * 8 + i] = W[4 * i + k];
 }
 #endif /* HAVE_SHA256_4WAY */
 
@@ -491,68 +604,97 @@ static void scrypt_1024_1_1_256_12way(const uint32_t *input,
 	uint32_t W[12 * 32] __attribute__((aligned(128)));
 	uint32_t X[12 * 32] __attribute__((aligned(128)));
 	uint32_t *V;
-	int i, j;
+	int i, j, k;
 	
 	V = (uint32_t *)(((uintptr_t)(scratchpad) + 63) & ~ (uintptr_t)(63));
 
-	for (j = 0; j < 3; j++) {
-		for (i = 0; i < 20; i++) {
-			W[128 * j + 4 * i + 0] = input[80 * j + 0 * 20 + i];
-			W[128 * j + 4 * i + 1] = input[80 * j + 1 * 20 + i];
-			W[128 * j + 4 * i + 2] = input[80 * j + 2 * 20 + i];
-			W[128 * j + 4 * i + 3] = input[80 * j + 3 * 20 + i];
-		}
-	}
-	for (j = 0; j < 3; j++) {
-		for (i = 0; i < 8; i++) {
-			tstate[32 * j + 4 * i + 0] = midstate[i];
-			tstate[32 * j + 4 * i + 1] = midstate[i];
-			tstate[32 * j + 4 * i + 2] = midstate[i];
-			tstate[32 * j + 4 * i + 3] = midstate[i];
-		}
-	}
+	for (j = 0; j < 3; j++)
+		for (i = 0; i < 20; i++)
+			for (k = 0; k < 4; k++)
+				W[128 * j + 4 * i + k] = input[80 * j + k * 20 + i];
+	for (j = 0; j < 3; j++)
+		for (i = 0; i < 8; i++)
+			for (k = 0; k < 4; k++)
+				tstate[32 * j + 4 * i + k] = midstate[i];
 	HMAC_SHA256_80_init_4way(W +   0, tstate +  0, ostate +  0);
 	HMAC_SHA256_80_init_4way(W + 128, tstate + 32, ostate + 32);
 	HMAC_SHA256_80_init_4way(W + 256, tstate + 64, ostate + 64);
 	PBKDF2_SHA256_80_128_4way(tstate +  0, ostate +  0, W +   0, W +   0);
 	PBKDF2_SHA256_80_128_4way(tstate + 32, ostate + 32, W + 128, W + 128);
 	PBKDF2_SHA256_80_128_4way(tstate + 64, ostate + 64, W + 256, W + 256);
-	for (j = 0; j < 3; j++) {
-		for (i = 0; i < 32; i++) {
-			X[128 * j + 0 * 32 + i] = W[128 * j + 4 * i + 0];
-			X[128 * j + 1 * 32 + i] = W[128 * j + 4 * i + 1];
-			X[128 * j + 2 * 32 + i] = W[128 * j + 4 * i + 2];
-			X[128 * j + 3 * 32 + i] = W[128 * j + 4 * i + 3];
-		}
-	}
+	for (j = 0; j < 3; j++)
+		for (i = 0; i < 32; i++)
+			for (k = 0; k < 4; k++)
+				X[128 * j + k * 32 + i] = W[128 * j + 4 * i + k];
 	scrypt_core_3way(X + 0 * 96, V);
 	scrypt_core_3way(X + 1 * 96, V);
 	scrypt_core_3way(X + 2 * 96, V);
 	scrypt_core_3way(X + 3 * 96, V);
-	for (j = 0; j < 3; j++) {
-		for (i = 0; i < 32; i++) {
-			W[128 * j + 4 * i + 0] = X[128 * j + 0 * 32 + i];
-			W[128 * j + 4 * i + 1] = X[128 * j + 1 * 32 + i];
-			W[128 * j + 4 * i + 2] = X[128 * j + 2 * 32 + i];
-			W[128 * j + 4 * i + 3] = X[128 * j + 3 * 32 + i];
-		}
-	}
+	for (j = 0; j < 3; j++)
+		for (i = 0; i < 32; i++)
+			for (k = 0; k < 4; k++)
+				W[128 * j + 4 * i + k] = X[128 * j + k * 32 + i];
 	PBKDF2_SHA256_128_32_4way(tstate +  0, ostate +  0, W +   0, W +   0);
 	PBKDF2_SHA256_128_32_4way(tstate + 32, ostate + 32, W + 128, W + 128);
 	PBKDF2_SHA256_128_32_4way(tstate + 64, ostate + 64, W + 256, W + 256);
-	for (j = 0; j < 3; j++) {
-		for (i = 0; i < 8; i++) {
-			output[32 * j + 0 * 8 + i] = W[128 * j + 4 * i + 0];
-			output[32 * j + 1 * 8 + i] = W[128 * j + 4 * i + 1];
-			output[32 * j + 2 * 8 + i] = W[128 * j + 4 * i + 2];
-			output[32 * j + 3 * 8 + i] = W[128 * j + 4 * i + 3];
-		}
-	}
+	for (j = 0; j < 3; j++)
+		for (i = 0; i < 8; i++)
+			for (k = 0; k < 4; k++)
+				output[32 * j + k * 8 + i] = W[128 * j + 4 * i + k];
 }
 #endif /* HAVE_SHA256_4WAY */
 
 #endif /* HAVE_SCRYPT_3WAY */
 
+#ifdef HAVE_SCRYPT_6WAY
+static void scrypt_1024_1_1_256_24way(const uint32_t *input,
+	uint32_t *output, uint32_t *midstate, unsigned char *scratchpad)
+{
+	uint32_t tstate[24 * 8] __attribute__((aligned(128)));
+	uint32_t ostate[24 * 8] __attribute__((aligned(128)));
+	uint32_t W[24 * 32] __attribute__((aligned(128)));
+	uint32_t X[24 * 32] __attribute__((aligned(128)));
+	uint32_t *V;
+	int i, j, k;
+	
+	V = (uint32_t *)(((uintptr_t)(scratchpad) + 63) & ~ (uintptr_t)(63));
+	
+	for (j = 0; j < 3; j++) 
+		for (i = 0; i < 20; i++)
+			for (k = 0; k < 8; k++)
+				W[8 * 32 * j + 8 * i + k] = input[8 * 20 * j + k * 20 + i];
+	for (j = 0; j < 3; j++)
+		for (i = 0; i < 8; i++)
+			for (k = 0; k < 8; k++)
+				tstate[8 * 8 * j + 8 * i + k] = midstate[i];
+	HMAC_SHA256_80_init_8way(W +   0, tstate +   0, ostate +   0);
+	HMAC_SHA256_80_init_8way(W + 256, tstate +  64, ostate +  64);
+	HMAC_SHA256_80_init_8way(W + 512, tstate + 128, ostate + 128);
+	PBKDF2_SHA256_80_128_8way(tstate +   0, ostate +   0, W +   0, W +   0);
+	PBKDF2_SHA256_80_128_8way(tstate +  64, ostate +  64, W + 256, W + 256);
+	PBKDF2_SHA256_80_128_8way(tstate + 128, ostate + 128, W + 512, W + 512);
+	for (j = 0; j < 3; j++)
+		for (i = 0; i < 32; i++)
+			for (k = 0; k < 8; k++)
+				X[8 * 32 * j + k * 32 + i] = W[8 * 32 * j + 8 * i + k];
+	scrypt_core_6way(X + 0 * 32, V);
+	scrypt_core_6way(X + 6 * 32, V);
+	scrypt_core_6way(X + 12 * 32, V);
+	scrypt_core_6way(X + 18 * 32, V);
+	for (j = 0; j < 3; j++)
+		for (i = 0; i < 32; i++)
+			for (k = 0; k < 8; k++)
+				W[8 * 32 * j + 8 * i + k] = X[8 * 32 * j + k * 32 + i];
+	PBKDF2_SHA256_128_32_8way(tstate +   0, ostate +   0, W +   0, W +   0);
+	PBKDF2_SHA256_128_32_8way(tstate +  64, ostate +  64, W + 256, W + 256);
+	PBKDF2_SHA256_128_32_8way(tstate + 128, ostate + 128, W + 512, W + 512);
+	for (j = 0; j < 3; j++)
+		for (i = 0; i < 8; i++)
+			for (k = 0; k < 8; k++)
+				output[8 * 8 * j + k * 8 + i] = W[8 * 32 * j + 8 * i + k];
+}
+#endif /* HAVE_SCRYPT_6WAY */
+
 int scanhash_scrypt(int thr_id, uint32_t *pdata,
 	unsigned char *scratchbuf, const uint32_t *ptarget,
 	uint32_t max_nonce, unsigned long *hashes_done)
@@ -589,6 +731,11 @@ int scanhash_scrypt(int thr_id, uint32_t *pdata,
 			scrypt_1024_1_1_256_12way(data, hash, midstate, scratchbuf);
 		else
 #endif
+#if defined(HAVE_SCRYPT_6WAY)
+		if (throughput == 24)
+			scrypt_1024_1_1_256_24way(data, hash, midstate, scratchbuf);
+		else
+#endif
 #if defined(HAVE_SCRYPT_3WAY)
 		if (throughput == 3)
 			scrypt_1024_1_1_256_3way(data, hash, midstate, scratchbuf);
diff --git a/sha2-x64.S b/sha2-x64.S
index 1fe87a652..a1581dd9d 100644
--- a/sha2-x64.S
+++ b/sha2-x64.S
@@ -1,5 +1,5 @@
 /*
- * Copyright 2012 pooler@litecoinpool.org
+ * Copyright 2012-2013 pooler@litecoinpool.org
  *
  * This program is free software; you can redistribute it and/or modify it
  * under the terms of the GNU General Public License as published by the Free
@@ -107,6 +107,102 @@ sha256d_4preext2_30:
 	.long 0x00400022, 0x00400022, 0x00400022, 0x00400022
 
 
+#ifdef USE_AVX2
+
+	.data
+	.p2align 7
+sha256_8h:
+	.long 0x6a09e667, 0x6a09e667, 0x6a09e667, 0x6a09e667, 0x6a09e667, 0x6a09e667, 0x6a09e667, 0x6a09e667
+	.long 0xbb67ae85, 0xbb67ae85, 0xbb67ae85, 0xbb67ae85, 0xbb67ae85, 0xbb67ae85, 0xbb67ae85, 0xbb67ae85
+	.long 0x3c6ef372, 0x3c6ef372, 0x3c6ef372, 0x3c6ef372, 0x3c6ef372, 0x3c6ef372, 0x3c6ef372, 0x3c6ef372
+	.long 0xa54ff53a, 0xa54ff53a, 0xa54ff53a, 0xa54ff53a, 0xa54ff53a, 0xa54ff53a, 0xa54ff53a, 0xa54ff53a
+	.long 0x510e527f, 0x510e527f, 0x510e527f, 0x510e527f, 0x510e527f, 0x510e527f, 0x510e527f, 0x510e527f
+	.long 0x9b05688c, 0x9b05688c, 0x9b05688c, 0x9b05688c, 0x9b05688c, 0x9b05688c, 0x9b05688c, 0x9b05688c
+	.long 0x1f83d9ab, 0x1f83d9ab, 0x1f83d9ab, 0x1f83d9ab, 0x1f83d9ab, 0x1f83d9ab, 0x1f83d9ab, 0x1f83d9ab
+	.long 0x5be0cd19, 0x5be0cd19, 0x5be0cd19, 0x5be0cd19, 0x5be0cd19, 0x5be0cd19, 0x5be0cd19, 0x5be0cd19
+
+	.data
+	.p2align 7
+sha256_8k:
+	.long 0x428a2f98, 0x428a2f98, 0x428a2f98, 0x428a2f98, 0x428a2f98, 0x428a2f98, 0x428a2f98, 0x428a2f98
+	.long 0x71374491, 0x71374491, 0x71374491, 0x71374491, 0x71374491, 0x71374491, 0x71374491, 0x71374491
+	.long 0xb5c0fbcf, 0xb5c0fbcf, 0xb5c0fbcf, 0xb5c0fbcf, 0xb5c0fbcf, 0xb5c0fbcf, 0xb5c0fbcf, 0xb5c0fbcf
+	.long 0xe9b5dba5, 0xe9b5dba5, 0xe9b5dba5, 0xe9b5dba5, 0xe9b5dba5, 0xe9b5dba5, 0xe9b5dba5, 0xe9b5dba5
+	.long 0x3956c25b, 0x3956c25b, 0x3956c25b, 0x3956c25b, 0x3956c25b, 0x3956c25b, 0x3956c25b, 0x3956c25b
+	.long 0x59f111f1, 0x59f111f1, 0x59f111f1, 0x59f111f1, 0x59f111f1, 0x59f111f1, 0x59f111f1, 0x59f111f1
+	.long 0x923f82a4, 0x923f82a4, 0x923f82a4, 0x923f82a4, 0x923f82a4, 0x923f82a4, 0x923f82a4, 0x923f82a4
+	.long 0xab1c5ed5, 0xab1c5ed5, 0xab1c5ed5, 0xab1c5ed5, 0xab1c5ed5, 0xab1c5ed5, 0xab1c5ed5, 0xab1c5ed5
+	.long 0xd807aa98, 0xd807aa98, 0xd807aa98, 0xd807aa98, 0xd807aa98, 0xd807aa98, 0xd807aa98, 0xd807aa98
+	.long 0x12835b01, 0x12835b01, 0x12835b01, 0x12835b01, 0x12835b01, 0x12835b01, 0x12835b01, 0x12835b01
+	.long 0x243185be, 0x243185be, 0x243185be, 0x243185be, 0x243185be, 0x243185be, 0x243185be, 0x243185be
+	.long 0x550c7dc3, 0x550c7dc3, 0x550c7dc3, 0x550c7dc3, 0x550c7dc3, 0x550c7dc3, 0x550c7dc3, 0x550c7dc3
+	.long 0x72be5d74, 0x72be5d74, 0x72be5d74, 0x72be5d74, 0x72be5d74, 0x72be5d74, 0x72be5d74, 0x72be5d74
+	.long 0x80deb1fe, 0x80deb1fe, 0x80deb1fe, 0x80deb1fe, 0x80deb1fe, 0x80deb1fe, 0x80deb1fe, 0x80deb1fe
+	.long 0x9bdc06a7, 0x9bdc06a7, 0x9bdc06a7, 0x9bdc06a7, 0x9bdc06a7, 0x9bdc06a7, 0x9bdc06a7, 0x9bdc06a7
+	.long 0xc19bf174, 0xc19bf174, 0xc19bf174, 0xc19bf174, 0xc19bf174, 0xc19bf174, 0xc19bf174, 0xc19bf174
+	.long 0xe49b69c1, 0xe49b69c1, 0xe49b69c1, 0xe49b69c1, 0xe49b69c1, 0xe49b69c1, 0xe49b69c1, 0xe49b69c1
+	.long 0xefbe4786, 0xefbe4786, 0xefbe4786, 0xefbe4786, 0xefbe4786, 0xefbe4786, 0xefbe4786, 0xefbe4786
+	.long 0x0fc19dc6, 0x0fc19dc6, 0x0fc19dc6, 0x0fc19dc6, 0x0fc19dc6, 0x0fc19dc6, 0x0fc19dc6, 0x0fc19dc6
+	.long 0x240ca1cc, 0x240ca1cc, 0x240ca1cc, 0x240ca1cc, 0x240ca1cc, 0x240ca1cc, 0x240ca1cc, 0x240ca1cc
+	.long 0x2de92c6f, 0x2de92c6f, 0x2de92c6f, 0x2de92c6f, 0x2de92c6f, 0x2de92c6f, 0x2de92c6f, 0x2de92c6f
+	.long 0x4a7484aa, 0x4a7484aa, 0x4a7484aa, 0x4a7484aa, 0x4a7484aa, 0x4a7484aa, 0x4a7484aa, 0x4a7484aa
+	.long 0x5cb0a9dc, 0x5cb0a9dc, 0x5cb0a9dc, 0x5cb0a9dc, 0x5cb0a9dc, 0x5cb0a9dc, 0x5cb0a9dc, 0x5cb0a9dc
+	.long 0x76f988da, 0x76f988da, 0x76f988da, 0x76f988da, 0x76f988da, 0x76f988da, 0x76f988da, 0x76f988da
+	.long 0x983e5152, 0x983e5152, 0x983e5152, 0x983e5152, 0x983e5152, 0x983e5152, 0x983e5152, 0x983e5152
+	.long 0xa831c66d, 0xa831c66d, 0xa831c66d, 0xa831c66d, 0xa831c66d, 0xa831c66d, 0xa831c66d, 0xa831c66d
+	.long 0xb00327c8, 0xb00327c8, 0xb00327c8, 0xb00327c8, 0xb00327c8, 0xb00327c8, 0xb00327c8, 0xb00327c8
+	.long 0xbf597fc7, 0xbf597fc7, 0xbf597fc7, 0xbf597fc7, 0xbf597fc7, 0xbf597fc7, 0xbf597fc7, 0xbf597fc7
+	.long 0xc6e00bf3, 0xc6e00bf3, 0xc6e00bf3, 0xc6e00bf3, 0xc6e00bf3, 0xc6e00bf3, 0xc6e00bf3, 0xc6e00bf3
+	.long 0xd5a79147, 0xd5a79147, 0xd5a79147, 0xd5a79147, 0xd5a79147, 0xd5a79147, 0xd5a79147, 0xd5a79147
+	.long 0x06ca6351, 0x06ca6351, 0x06ca6351, 0x06ca6351, 0x06ca6351, 0x06ca6351, 0x06ca6351, 0x06ca6351
+	.long 0x14292967, 0x14292967, 0x14292967, 0x14292967, 0x14292967, 0x14292967, 0x14292967, 0x14292967
+	.long 0x27b70a85, 0x27b70a85, 0x27b70a85, 0x27b70a85, 0x27b70a85, 0x27b70a85, 0x27b70a85, 0x27b70a85
+	.long 0x2e1b2138, 0x2e1b2138, 0x2e1b2138, 0x2e1b2138, 0x2e1b2138, 0x2e1b2138, 0x2e1b2138, 0x2e1b2138
+	.long 0x4d2c6dfc, 0x4d2c6dfc, 0x4d2c6dfc, 0x4d2c6dfc, 0x4d2c6dfc, 0x4d2c6dfc, 0x4d2c6dfc, 0x4d2c6dfc
+	.long 0x53380d13, 0x53380d13, 0x53380d13, 0x53380d13, 0x53380d13, 0x53380d13, 0x53380d13, 0x53380d13
+	.long 0x650a7354, 0x650a7354, 0x650a7354, 0x650a7354, 0x650a7354, 0x650a7354, 0x650a7354, 0x650a7354
+	.long 0x766a0abb, 0x766a0abb, 0x766a0abb, 0x766a0abb, 0x766a0abb, 0x766a0abb, 0x766a0abb, 0x766a0abb
+	.long 0x81c2c92e, 0x81c2c92e, 0x81c2c92e, 0x81c2c92e, 0x81c2c92e, 0x81c2c92e, 0x81c2c92e, 0x81c2c92e
+	.long 0x92722c85, 0x92722c85, 0x92722c85, 0x92722c85, 0x92722c85, 0x92722c85, 0x92722c85, 0x92722c85
+	.long 0xa2bfe8a1, 0xa2bfe8a1, 0xa2bfe8a1, 0xa2bfe8a1, 0xa2bfe8a1, 0xa2bfe8a1, 0xa2bfe8a1, 0xa2bfe8a1
+	.long 0xa81a664b, 0xa81a664b, 0xa81a664b, 0xa81a664b, 0xa81a664b, 0xa81a664b, 0xa81a664b, 0xa81a664b
+	.long 0xc24b8b70, 0xc24b8b70, 0xc24b8b70, 0xc24b8b70, 0xc24b8b70, 0xc24b8b70, 0xc24b8b70, 0xc24b8b70
+	.long 0xc76c51a3, 0xc76c51a3, 0xc76c51a3, 0xc76c51a3, 0xc76c51a3, 0xc76c51a3, 0xc76c51a3, 0xc76c51a3
+	.long 0xd192e819, 0xd192e819, 0xd192e819, 0xd192e819, 0xd192e819, 0xd192e819, 0xd192e819, 0xd192e819
+	.long 0xd6990624, 0xd6990624, 0xd6990624, 0xd6990624, 0xd6990624, 0xd6990624, 0xd6990624, 0xd6990624
+	.long 0xf40e3585, 0xf40e3585, 0xf40e3585, 0xf40e3585, 0xf40e3585, 0xf40e3585, 0xf40e3585, 0xf40e3585
+	.long 0x106aa070, 0x106aa070, 0x106aa070, 0x106aa070, 0x106aa070, 0x106aa070, 0x106aa070, 0x106aa070
+	.long 0x19a4c116, 0x19a4c116, 0x19a4c116, 0x19a4c116, 0x19a4c116, 0x19a4c116, 0x19a4c116, 0x19a4c116
+	.long 0x1e376c08, 0x1e376c08, 0x1e376c08, 0x1e376c08, 0x1e376c08, 0x1e376c08, 0x1e376c08, 0x1e376c08
+	.long 0x2748774c, 0x2748774c, 0x2748774c, 0x2748774c, 0x2748774c, 0x2748774c, 0x2748774c, 0x2748774c
+	.long 0x34b0bcb5, 0x34b0bcb5, 0x34b0bcb5, 0x34b0bcb5, 0x34b0bcb5, 0x34b0bcb5, 0x34b0bcb5, 0x34b0bcb5
+	.long 0x391c0cb3, 0x391c0cb3, 0x391c0cb3, 0x391c0cb3, 0x391c0cb3, 0x391c0cb3, 0x391c0cb3, 0x391c0cb3
+	.long 0x4ed8aa4a, 0x4ed8aa4a, 0x4ed8aa4a, 0x4ed8aa4a, 0x4ed8aa4a, 0x4ed8aa4a, 0x4ed8aa4a, 0x4ed8aa4a
+	.long 0x5b9cca4f, 0x5b9cca4f, 0x5b9cca4f, 0x5b9cca4f, 0x5b9cca4f, 0x5b9cca4f, 0x5b9cca4f, 0x5b9cca4f
+	.long 0x682e6ff3, 0x682e6ff3, 0x682e6ff3, 0x682e6ff3, 0x682e6ff3, 0x682e6ff3, 0x682e6ff3, 0x682e6ff3
+	.long 0x748f82ee, 0x748f82ee, 0x748f82ee, 0x748f82ee, 0x748f82ee, 0x748f82ee, 0x748f82ee, 0x748f82ee
+	.long 0x78a5636f, 0x78a5636f, 0x78a5636f, 0x78a5636f, 0x78a5636f, 0x78a5636f, 0x78a5636f, 0x78a5636f
+	.long 0x84c87814, 0x84c87814, 0x84c87814, 0x84c87814, 0x84c87814, 0x84c87814, 0x84c87814, 0x84c87814
+	.long 0x8cc70208, 0x8cc70208, 0x8cc70208, 0x8cc70208, 0x8cc70208, 0x8cc70208, 0x8cc70208, 0x8cc70208
+	.long 0x90befffa, 0x90befffa, 0x90befffa, 0x90befffa, 0x90befffa, 0x90befffa, 0x90befffa, 0x90befffa
+	.long 0xa4506ceb, 0xa4506ceb, 0xa4506ceb, 0xa4506ceb, 0xa4506ceb, 0xa4506ceb, 0xa4506ceb, 0xa4506ceb
+	.long 0xbef9a3f7, 0xbef9a3f7, 0xbef9a3f7, 0xbef9a3f7, 0xbef9a3f7, 0xbef9a3f7, 0xbef9a3f7, 0xbef9a3f7
+	.long 0xc67178f2, 0xc67178f2, 0xc67178f2, 0xc67178f2, 0xc67178f2, 0xc67178f2, 0xc67178f2, 0xc67178f2
+
+	.data
+	.p2align 6
+sha256d_8preext2_17:
+	.long 0x00a00000, 0x00a00000, 0x00a00000, 0x00a00000, 0x00a00000, 0x00a00000, 0x00a00000, 0x00a00000
+sha256d_8preext2_23:
+	.long 0x11002000, 0x11002000, 0x11002000, 0x11002000, 0x11002000, 0x11002000, 0x11002000, 0x11002000
+sha256d_8preext2_24:
+	.long 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000, 0x80000000
+sha256d_8preext2_30:
+	.long 0x00400022, 0x00400022, 0x00400022, 0x00400022, 0x00400022, 0x00400022, 0x00400022, 0x00400022
+
+#endif /* USE_AVX2 */
+
+
 	.text
 	.p2align 6
 	.globl sha256_init_4way
@@ -139,6 +235,40 @@ _sha256_init_4way:
 	ret
 
 
+#ifdef USE_AVX2
+	.text
+	.p2align 6
+	.globl sha256_init_8way
+	.globl _sha256_init_8way
+sha256_init_8way:
+_sha256_init_8way:
+#if defined(WIN64)
+	pushq	%rdi
+	movq	%rcx, %rdi
+#endif
+	vpbroadcastd	sha256_4h+0(%rip), %ymm0
+	vpbroadcastd	sha256_4h+16(%rip), %ymm1
+	vpbroadcastd	sha256_4h+32(%rip), %ymm2
+	vpbroadcastd	sha256_4h+48(%rip), %ymm3
+	vmovdqu	%ymm0, 0*32(%rdi)
+	vmovdqu	%ymm1, 1*32(%rdi)
+	vmovdqu	%ymm2, 2*32(%rdi)
+	vmovdqu	%ymm3, 3*32(%rdi)
+	vpbroadcastd	sha256_4h+64(%rip), %ymm0
+	vpbroadcastd	sha256_4h+80(%rip), %ymm1
+	vpbroadcastd	sha256_4h+96(%rip), %ymm2
+	vpbroadcastd	sha256_4h+112(%rip), %ymm3
+	vmovdqu	%ymm0, 4*32(%rdi)
+	vmovdqu	%ymm1, 5*32(%rdi)
+	vmovdqu	%ymm2, 6*32(%rdi)
+	vmovdqu	%ymm3, 7*32(%rdi)
+#if defined(WIN64)
+	popq	%rdi
+#endif
+	ret
+#endif /* USE_AVX2 */
+
+
 .macro sha256_sse2_extend_round i
 	movdqa	(\i-15)*16(%rax), %xmm0
 	movdqa	%xmm0, %xmm2
@@ -441,6 +571,143 @@ _sha256_init_4way:
 #endif /* USE_AVX */
 
 
+#if defined(USE_AVX2)
+
+.macro sha256_avx2_extend_round i
+	vmovdqa	(\i-15)*32(%rax), %ymm0
+	vpslld	$14, %ymm0, %ymm2
+	vpsrld	$3, %ymm0, %ymm0
+	vpsrld	$4, %ymm0, %ymm1
+	vpxor	%ymm1, %ymm0, %ymm0
+	vpxor	%ymm2, %ymm0, %ymm0
+	vpsrld	$11, %ymm1, %ymm1
+	vpslld	$11, %ymm2, %ymm2
+	vpxor	%ymm1, %ymm0, %ymm0
+	vpxor	%ymm2, %ymm0, %ymm0
+	vpaddd	(\i-16)*32(%rax), %ymm0, %ymm0
+	vpaddd	(\i-7)*32(%rax), %ymm0, %ymm0
+
+	vpslld	$13, %ymm3, %ymm2
+	vpsrld	$10, %ymm3, %ymm3
+	vpsrld	$7, %ymm3, %ymm1
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpsrld	$2, %ymm1, %ymm1
+	vpslld	$2, %ymm2, %ymm2
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpaddd	%ymm0, %ymm3, %ymm3
+	vmovdqa	%ymm3, \i*32(%rax)
+.endm
+
+.macro sha256_avx2_extend_doubleround i
+	vmovdqa	(\i-15)*32(%rax), %ymm0
+	vmovdqa	(\i-14)*32(%rax), %ymm4
+	vpslld	$14, %ymm0, %ymm2
+	vpslld	$14, %ymm4, %ymm6
+	vpsrld	$3, %ymm0, %ymm8
+	vpsrld	$3, %ymm4, %ymm4
+	vpsrld	$7, %ymm0, %ymm1
+	vpsrld	$4, %ymm4, %ymm5
+	vpxor	%ymm1, %ymm8, %ymm8
+	vpxor	%ymm5, %ymm4, %ymm4
+	vpsrld	$11, %ymm1, %ymm1
+	vpsrld	$11, %ymm5, %ymm5
+	vpxor	%ymm2, %ymm8, %ymm8
+	vpxor	%ymm6, %ymm4, %ymm4
+	vpslld	$11, %ymm2, %ymm2
+	vpslld	$11, %ymm6, %ymm6
+	vpxor	%ymm1, %ymm8, %ymm8
+	vpxor	%ymm5, %ymm4, %ymm4
+	vpxor	%ymm2, %ymm8, %ymm8
+	vpxor	%ymm6, %ymm4, %ymm4
+
+	vpaddd	%ymm0, %ymm4, %ymm4
+	vpaddd	(\i-16)*32(%rax), %ymm8, %ymm0
+
+	vpslld	$13, %ymm3, %ymm2
+	vpslld	$13, %ymm7, %ymm6
+	vpsrld	$10, %ymm3, %ymm3
+	vpsrld	$10, %ymm7, %ymm7
+
+	vpaddd	(\i-7)*32(%rax), %ymm0, %ymm0
+	vpaddd	(\i-6)*32(%rax), %ymm4, %ymm4
+
+	vpsrld	$7, %ymm3, %ymm1
+	vpsrld	$7, %ymm7, %ymm5
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpsrld	$2, %ymm1, %ymm1
+	vpsrld	$2, %ymm5, %ymm5
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpslld	$2, %ymm2, %ymm2
+	vpslld	$2, %ymm6, %ymm6
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+
+	vpaddd	%ymm0, %ymm3, %ymm3
+	vpaddd	%ymm4, %ymm7, %ymm7
+	vmovdqa	%ymm3, \i*32(%rax)
+	vmovdqa	%ymm7, (\i+1)*32(%rax)
+.endm
+
+.macro sha256_avx2_main_round i, r0, r1, r2, r3, r4, r5, r6, r7
+	vpaddd	32*(\i)(%rax), \r0, %ymm6
+	vpaddd	32*(\i)(%rcx), %ymm6, %ymm6
+
+	vpandn	\r1, \r3, %ymm1
+	vpand	\r3, \r2, %ymm2
+	vpxor	%ymm2, %ymm1, %ymm1
+	vpaddd	%ymm1, %ymm6, %ymm6
+
+	vpslld	$7, \r3, %ymm1
+	vpsrld	$6, \r3, \r0
+	vpsrld	$5, \r0, %ymm2
+	vpxor	%ymm1, \r0, \r0
+	vpxor	%ymm2, \r0, \r0
+	vpslld	$14, %ymm1, %ymm1
+	vpsrld	$14, %ymm2, %ymm2
+	vpxor	%ymm1, \r0, \r0
+	vpxor	%ymm2, \r0, \r0
+	vpslld	$5, %ymm1, %ymm1
+	vpxor	%ymm1, \r0, \r0
+	vpaddd	\r0, %ymm6, %ymm6
+	vpaddd	%ymm6, \r4, \r0
+
+	vpand	\r6, \r5, %ymm2
+	vpand	\r7, \r5, \r4
+	vpand	\r7, \r6, %ymm1
+	vpxor	\r4, %ymm1, %ymm1
+	vpxor	%ymm2, %ymm1, %ymm1
+	vpaddd	%ymm1, %ymm6, %ymm6
+
+	vpslld	$10, \r7, %ymm2
+	vpsrld	$2, \r7, \r4
+	vpsrld	$11, \r4, %ymm1
+	vpxor	%ymm2, \r4, \r4
+	vpxor	%ymm1, \r4, \r4
+	vpslld	$9, %ymm2, %ymm2
+	vpsrld	$9, %ymm1, %ymm1
+	vpxor	%ymm2, \r4, \r4
+	vpxor	%ymm1, \r4, \r4
+	vpslld	$11, %ymm2, %ymm2
+	vpxor	%ymm2, \r4, \r4
+	vpaddd	%ymm6, \r4, \r4
+.endm
+
+.macro sha256_avx2_main_quadround i
+	sha256_avx2_main_round \i+0, %ymm10, %ymm9, %ymm8, %ymm0, %ymm3, %ymm4, %ymm5, %ymm7
+	sha256_avx2_main_round \i+1, %ymm9, %ymm8, %ymm0, %ymm10, %ymm4, %ymm5, %ymm7, %ymm3
+	sha256_avx2_main_round \i+2, %ymm8, %ymm0, %ymm10, %ymm9, %ymm5, %ymm7, %ymm3, %ymm4
+	sha256_avx2_main_round \i+3, %ymm0, %ymm10, %ymm9, %ymm8, %ymm7, %ymm3, %ymm4, %ymm5
+.endm
+
+#endif /* USE_AVX2 */
+
+
 #if defined(USE_XOP)
 
 .macro sha256_xop_extend_round i
@@ -955,6 +1222,201 @@ sha256_transform_4way_finish:
 	popq	%rdi
 #endif
 	ret
+
+
+#ifdef USE_AVX2
+
+	.text
+	.p2align 6
+sha256_transform_8way_core_avx2:
+	leaq	8*64(%rsp), %rax
+	vmovdqa	-2*32(%rax), %ymm3
+	vmovdqa	-1*32(%rax), %ymm7
+	sha256_avx2_extend_doubleround 0
+	sha256_avx2_extend_doubleround 2
+	sha256_avx2_extend_doubleround 4
+	sha256_avx2_extend_doubleround 6
+	sha256_avx2_extend_doubleround 8
+	sha256_avx2_extend_doubleround 10
+	sha256_avx2_extend_doubleround 12
+	sha256_avx2_extend_doubleround 14
+	sha256_avx2_extend_doubleround 16
+	sha256_avx2_extend_doubleround 18
+	sha256_avx2_extend_doubleround 20
+	sha256_avx2_extend_doubleround 22
+	sha256_avx2_extend_doubleround 24
+	sha256_avx2_extend_doubleround 26
+	sha256_avx2_extend_doubleround 28
+	sha256_avx2_extend_doubleround 30
+	sha256_avx2_extend_doubleround 32
+	sha256_avx2_extend_doubleround 34
+	sha256_avx2_extend_doubleround 36
+	sha256_avx2_extend_doubleround 38
+	sha256_avx2_extend_doubleround 40
+	sha256_avx2_extend_doubleround 42
+	sha256_avx2_extend_doubleround 44
+	sha256_avx2_extend_doubleround 46
+	vmovdqu	0*32(%rdi), %ymm7
+	vmovdqu	1*32(%rdi), %ymm5
+	vmovdqu	2*32(%rdi), %ymm4
+	vmovdqu	3*32(%rdi), %ymm3
+	vmovdqu	4*32(%rdi), %ymm0
+	vmovdqu	5*32(%rdi), %ymm8
+	vmovdqu	6*32(%rdi), %ymm9
+	vmovdqu	7*32(%rdi), %ymm10
+	movq	%rsp, %rax
+	leaq	sha256_8k(%rip), %rcx
+	sha256_avx2_main_quadround 0
+	sha256_avx2_main_quadround 4
+	sha256_avx2_main_quadround 8
+	sha256_avx2_main_quadround 12
+	sha256_avx2_main_quadround 16
+	sha256_avx2_main_quadround 20
+	sha256_avx2_main_quadround 24
+	sha256_avx2_main_quadround 28
+	sha256_avx2_main_quadround 32
+	sha256_avx2_main_quadround 36
+	sha256_avx2_main_quadround 40
+	sha256_avx2_main_quadround 44
+	sha256_avx2_main_quadround 48
+	sha256_avx2_main_quadround 52
+	sha256_avx2_main_quadround 56
+	sha256_avx2_main_quadround 60
+	jmp sha256_transform_8way_finish
+
+.macro p2bswap_avx2_rsi_rsp i
+	vmovdqu	\i*32(%rsi), %ymm0
+	vmovdqu	(\i+1)*32(%rsi), %ymm2
+	vpshuflw	$0xb1, %ymm0, %ymm0
+	vpshuflw	$0xb1, %ymm2, %ymm2
+	vpshufhw	$0xb1, %ymm0, %ymm0
+	vpshufhw	$0xb1, %ymm2, %ymm2
+	vpsrlw	$8, %ymm0, %ymm1
+	vpsrlw	$8, %ymm2, %ymm3
+	vpsllw	$8, %ymm0, %ymm0
+	vpsllw	$8, %ymm2, %ymm2
+	vpxor	%ymm1, %ymm0, %ymm0
+	vpxor	%ymm3, %ymm2, %ymm2
+	vmovdqa	%ymm0, \i*32(%rsp)
+	vmovdqa	%ymm2, (\i+1)*32(%rsp)
+.endm
+	
+	.text
+	.p2align 6
+	.globl sha256_transform_8way
+	.globl _sha256_transform_8way
+sha256_transform_8way:
+_sha256_transform_8way:
+#if defined(WIN64)
+	pushq	%rdi
+	subq	$96, %rsp
+	vmovdqa	%xmm6, 0(%rsp)
+	vmovdqa	%xmm7, 16(%rsp)
+	vmovdqa	%xmm8, 32(%rsp)
+	vmovdqa	%xmm9, 48(%rsp)
+	vmovdqa	%xmm10, 64(%rsp)
+	vmovdqa	%xmm11, 80(%rsp)
+	pushq	%rsi
+	movq	%rcx, %rdi
+	movq	%rdx, %rsi
+	movq	%r8, %rdx
+#endif
+	movq	%rsp, %r8
+	subq	$64*32, %rsp
+	andq	$-128, %rsp
+	
+	testq	%rdx, %rdx
+	jnz sha256_transform_8way_swap
+	
+	vmovdqu	0*32(%rsi), %ymm0
+	vmovdqu	1*32(%rsi), %ymm1
+	vmovdqu	2*32(%rsi), %ymm2
+	vmovdqu	3*32(%rsi), %ymm3
+	vmovdqu	4*32(%rsi), %ymm4
+	vmovdqu	5*32(%rsi), %ymm5
+	vmovdqu	6*32(%rsi), %ymm6
+	vmovdqu	7*32(%rsi), %ymm7
+	vmovdqa	%ymm0, 0*32(%rsp)
+	vmovdqa	%ymm1, 1*32(%rsp)
+	vmovdqa	%ymm2, 2*32(%rsp)
+	vmovdqa	%ymm3, 3*32(%rsp)
+	vmovdqa	%ymm4, 4*32(%rsp)
+	vmovdqa	%ymm5, 5*32(%rsp)
+	vmovdqa	%ymm6, 6*32(%rsp)
+	vmovdqa	%ymm7, 7*32(%rsp)
+	vmovdqu	8*32(%rsi), %ymm0
+	vmovdqu	9*32(%rsi), %ymm1
+	vmovdqu	10*32(%rsi), %ymm2
+	vmovdqu	11*32(%rsi), %ymm3
+	vmovdqu	12*32(%rsi), %ymm4
+	vmovdqu	13*32(%rsi), %ymm5
+	vmovdqu	14*32(%rsi), %ymm6
+	vmovdqu	15*32(%rsi), %ymm7
+	vmovdqa	%ymm0, 8*32(%rsp)
+	vmovdqa	%ymm1, 9*32(%rsp)
+	vmovdqa	%ymm2, 10*32(%rsp)
+	vmovdqa	%ymm3, 11*32(%rsp)
+	vmovdqa	%ymm4, 12*32(%rsp)
+	vmovdqa	%ymm5, 13*32(%rsp)
+	vmovdqa	%ymm6, 14*32(%rsp)
+	vmovdqa	%ymm7, 15*32(%rsp)
+	jmp sha256_transform_8way_core_avx2
+	
+	.p2align 6
+sha256_transform_8way_swap:
+	p2bswap_avx2_rsi_rsp 0
+	p2bswap_avx2_rsi_rsp 2
+	p2bswap_avx2_rsi_rsp 4
+	p2bswap_avx2_rsi_rsp 6
+	p2bswap_avx2_rsi_rsp 8
+	p2bswap_avx2_rsi_rsp 10
+	p2bswap_avx2_rsi_rsp 12
+	p2bswap_avx2_rsi_rsp 14
+	jmp sha256_transform_8way_core_avx2
+	
+	.p2align 6
+sha256_transform_8way_finish:
+	vmovdqu	0*32(%rdi), %ymm2
+	vmovdqu	1*32(%rdi), %ymm6
+	vmovdqu	2*32(%rdi), %ymm11
+	vmovdqu	3*32(%rdi), %ymm1
+	vpaddd	%ymm2, %ymm7, %ymm7
+	vpaddd	%ymm6, %ymm5, %ymm5
+	vpaddd	%ymm11, %ymm4, %ymm4
+	vpaddd	%ymm1, %ymm3, %ymm3
+	vmovdqu	4*32(%rdi), %ymm2
+	vmovdqu	5*32(%rdi), %ymm6
+	vmovdqu	6*32(%rdi), %ymm11
+	vmovdqu	7*32(%rdi), %ymm1
+	vpaddd	%ymm2, %ymm0, %ymm0
+	vpaddd	%ymm6, %ymm8, %ymm8
+	vpaddd	%ymm11, %ymm9, %ymm9
+	vpaddd	%ymm1, %ymm10, %ymm10
+	
+	vmovdqu	%ymm7, 0*32(%rdi)
+	vmovdqu	%ymm5, 1*32(%rdi)
+	vmovdqu	%ymm4, 2*32(%rdi)
+	vmovdqu	%ymm3, 3*32(%rdi)
+	vmovdqu	%ymm0, 4*32(%rdi)
+	vmovdqu	%ymm8, 5*32(%rdi)
+	vmovdqu	%ymm9, 6*32(%rdi)
+	vmovdqu	%ymm10, 7*32(%rdi)
+	
+	movq	%r8, %rsp
+#if defined(WIN64)
+	popq	%rsi
+	vmovdqa	0(%rsp), %xmm6
+	vmovdqa	16(%rsp), %xmm7
+	vmovdqa	32(%rsp), %xmm8
+	vmovdqa	48(%rsp), %xmm9
+	vmovdqa	64(%rsp), %xmm10
+	vmovdqa	80(%rsp), %xmm11
+	addq	$96, %rsp
+	popq	%rdi
+#endif
+	ret
+
+#endif /* USE_AVX2 */
 	
 	
 	.data
@@ -2604,4 +3066,596 @@ sha256_use_4way_done:
 	movl	$1, %eax
 	ret
 
+
+#if defined(USE_AVX2)
+
+	.text
+	.p2align 6
+	.globl sha256d_ms_8way
+	.globl _sha256d_ms_8way
+sha256d_ms_8way:
+_sha256d_ms_8way:
+sha256d_ms_8way_avx2:
+#if defined(WIN64)
+	pushq	%rdi
+	subq	$80, %rsp
+	vmovdqa	%xmm6, 0(%rsp)
+	vmovdqa	%xmm7, 16(%rsp)
+	vmovdqa	%xmm8, 32(%rsp)
+	vmovdqa	%xmm9, 48(%rsp)
+	vmovdqa	%xmm10, 64(%rsp)
+	pushq	%rsi
+	movq	%rcx, %rdi
+	movq	%rdx, %rsi
+	movq	%r8, %rdx
+	movq	%r9, %rcx
+#endif
+	pushq	%rbp
+	movq	%rsp, %rbp
+	subq	$64*32, %rsp
+	andq	$-128, %rsp
+	
+	leaq	16*32(%rsi), %rax
+	
+sha256d_ms_8way_avx2_extend_loop1:
+	vmovdqa	3*32(%rsi), %ymm0
+	vmovdqa	2*32(%rax), %ymm3
+	vmovdqa	3*32(%rax), %ymm7
+	vmovdqa	%ymm3, 2*32(%rsp)
+	vmovdqa	%ymm7, 3*32(%rsp)
+	vpaddd	%ymm0, %ymm7, %ymm7
+	vpslld	$14, %ymm0, %ymm2
+	vpsrld	$3, %ymm0, %ymm0
+	vpsrld	$4, %ymm0, %ymm1
+	vpxor	%ymm1, %ymm0, %ymm0
+	vpxor	%ymm2, %ymm0, %ymm0
+	vpsrld	$11, %ymm1, %ymm1
+	vpslld	$11, %ymm2, %ymm2
+	vpxor	%ymm1, %ymm0, %ymm0
+	vpxor	%ymm2, %ymm0, %ymm0
+	vpaddd	%ymm0, %ymm3, %ymm3
+	vmovdqa	%ymm3, 2*32(%rax)
+	vmovdqa	%ymm7, 3*32(%rax)
+	
+	vmovdqa	4*32(%rax), %ymm0
+	vmovdqa	%ymm0, 4*32(%rsp)
+	vpslld	$13, %ymm3, %ymm2
+	vpslld	$13, %ymm7, %ymm6
+	vpsrld	$10, %ymm3, %ymm3
+	vpsrld	$10, %ymm7, %ymm7
+	vpsrld	$7, %ymm3, %ymm1
+	vpsrld	$7, %ymm7, %ymm5
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpsrld	$2, %ymm1, %ymm1
+	vpsrld	$2, %ymm5, %ymm5
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpslld	$2, %ymm2, %ymm2
+	vpslld	$2, %ymm6, %ymm6
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpaddd	%ymm0, %ymm3, %ymm3
+	vmovdqa	%ymm3, 4*32(%rax)
+	vmovdqa	%ymm7, 5*32(%rax)
+	
+	vmovdqa	6*32(%rax), %ymm0
+	vmovdqa	7*32(%rax), %ymm4
+	vmovdqa	%ymm0, 6*32(%rsp)
+	vmovdqa	%ymm4, 7*32(%rsp)
+	vpslld	$13, %ymm3, %ymm2
+	vpslld	$13, %ymm7, %ymm6
+	vpsrld	$10, %ymm3, %ymm3
+	vpsrld	$10, %ymm7, %ymm7
+	vpsrld	$7, %ymm3, %ymm1
+	vpsrld	$7, %ymm7, %ymm5
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpsrld	$2, %ymm1, %ymm1
+	vpsrld	$2, %ymm5, %ymm5
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpslld	$2, %ymm2, %ymm2
+	vpslld	$2, %ymm6, %ymm6
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpaddd	%ymm0, %ymm3, %ymm3
+	vpaddd	%ymm4, %ymm7, %ymm7
+	vmovdqa	%ymm3, 6*32(%rax)
+	vmovdqa	%ymm7, 7*32(%rax)
+	
+	vmovdqa	8*32(%rax), %ymm0
+	vmovdqa	2*32(%rax), %ymm4
+	vmovdqa	%ymm0, 8*32(%rsp)
+	vpslld	$13, %ymm3, %ymm2
+	vpslld	$13, %ymm7, %ymm6
+	vpsrld	$10, %ymm3, %ymm3
+	vpsrld	$10, %ymm7, %ymm7
+	vpsrld	$7, %ymm3, %ymm1
+	vpsrld	$7, %ymm7, %ymm5
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpsrld	$2, %ymm1, %ymm1
+	vpsrld	$2, %ymm5, %ymm5
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpslld	$2, %ymm2, %ymm2
+	vpslld	$2, %ymm6, %ymm6
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpaddd	%ymm0, %ymm3, %ymm3
+	vpaddd	%ymm4, %ymm7, %ymm7
+	vmovdqa	%ymm3, 8*32(%rax)
+	vmovdqa	%ymm7, 9*32(%rax)
+	
+	vpslld	$13, %ymm3, %ymm2
+	vpslld	$13, %ymm7, %ymm6
+	vpsrld	$10, %ymm3, %ymm3
+	vpsrld	$10, %ymm7, %ymm7
+	vpsrld	$7, %ymm3, %ymm1
+	vpsrld	$7, %ymm7, %ymm5
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpsrld	$2, %ymm1, %ymm1
+	vpsrld	$2, %ymm5, %ymm5
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpslld	$2, %ymm2, %ymm2
+	vpslld	$2, %ymm6, %ymm6
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpaddd	3*32(%rax), %ymm3, %ymm3
+	vpaddd	4*32(%rax), %ymm7, %ymm7
+	vmovdqa	%ymm3, 10*32(%rax)
+	vmovdqa	%ymm7, 11*32(%rax)
+	
+	vpslld	$13, %ymm3, %ymm2
+	vpslld	$13, %ymm7, %ymm6
+	vpsrld	$10, %ymm3, %ymm3
+	vpsrld	$10, %ymm7, %ymm7
+	vpsrld	$7, %ymm3, %ymm1
+	vpsrld	$7, %ymm7, %ymm5
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpsrld	$2, %ymm1, %ymm1
+	vpsrld	$2, %ymm5, %ymm5
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpslld	$2, %ymm2, %ymm2
+	vpslld	$2, %ymm6, %ymm6
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpaddd	5*32(%rax), %ymm3, %ymm3
+	vpaddd	6*32(%rax), %ymm7, %ymm7
+	vmovdqa	%ymm3, 12*32(%rax)
+	vmovdqa	%ymm7, 13*32(%rax)
+	
+	vmovdqa	14*32(%rax), %ymm0
+	vmovdqa	15*32(%rax), %ymm4
+	vmovdqa	%ymm0, 14*32(%rsp)
+	vmovdqa	%ymm4, 15*32(%rsp)
+	vpslld	$13, %ymm3, %ymm2
+	vpslld	$13, %ymm7, %ymm6
+	vpsrld	$10, %ymm3, %ymm3
+	vpsrld	$10, %ymm7, %ymm7
+	vpaddd	7*32(%rax), %ymm0, %ymm0
+	vpaddd	8*32(%rax), %ymm4, %ymm4
+	vpsrld	$7, %ymm3, %ymm1
+	vpsrld	$7, %ymm7, %ymm5
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpsrld	$2, %ymm1, %ymm1
+	vpsrld	$2, %ymm5, %ymm5
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpslld	$2, %ymm2, %ymm2
+	vpslld	$2, %ymm6, %ymm6
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpaddd	%ymm0, %ymm3, %ymm3
+	vpaddd	%ymm4, %ymm7, %ymm7
+	vmovdqa	%ymm3, 14*32(%rax)
+	vmovdqa	%ymm7, 15*32(%rax)
+	
+sha256d_ms_8way_avx2_extend_loop2:
+	sha256_avx2_extend_doubleround 16
+	sha256_avx2_extend_doubleround 18
+	sha256_avx2_extend_doubleround 20
+	sha256_avx2_extend_doubleround 22
+	sha256_avx2_extend_doubleround 24
+	sha256_avx2_extend_doubleround 26
+	sha256_avx2_extend_doubleround 28
+	sha256_avx2_extend_doubleround 30
+	sha256_avx2_extend_doubleround 32
+	sha256_avx2_extend_doubleround 34
+	sha256_avx2_extend_doubleround 36
+	sha256_avx2_extend_doubleround 38
+	sha256_avx2_extend_doubleround 40
+	sha256_avx2_extend_doubleround 42
+	jz sha256d_ms_8way_avx2_extend_coda2
+	sha256_avx2_extend_doubleround 44
+	sha256_avx2_extend_doubleround 46
+	
+	vmovdqa	0(%rcx), %ymm7
+	vmovdqa	32(%rcx), %ymm8
+	vmovdqa	64(%rcx), %ymm9
+	vmovdqa	96(%rcx), %ymm10
+	vmovdqa	128(%rcx), %ymm0
+	vmovdqa	160(%rcx), %ymm5
+	vmovdqa	192(%rcx), %ymm4
+	vmovdqa	224(%rcx), %ymm3
+	
+	movq	%rsi, %rax
+	leaq	sha256_8k(%rip), %rcx
+	jmp sha256d_ms_8way_avx2_main_loop1
+	
+sha256d_ms_8way_avx2_main_loop2:
+	sha256_avx2_main_round 0, %ymm10, %ymm9, %ymm8, %ymm0, %ymm3, %ymm4, %ymm5, %ymm7
+	sha256_avx2_main_round 1, %ymm9, %ymm8, %ymm0, %ymm10, %ymm4, %ymm5, %ymm7, %ymm3
+	sha256_avx2_main_round 2, %ymm8, %ymm0, %ymm10, %ymm9, %ymm5, %ymm7, %ymm3, %ymm4
+sha256d_ms_8way_avx2_main_loop1:
+	sha256_avx2_main_round 3, %ymm0, %ymm10, %ymm9, %ymm8, %ymm7, %ymm3, %ymm4, %ymm5
+	sha256_avx2_main_quadround 4
+	sha256_avx2_main_quadround 8
+	sha256_avx2_main_quadround 12
+	sha256_avx2_main_quadround 16
+	sha256_avx2_main_quadround 20
+	sha256_avx2_main_quadround 24
+	sha256_avx2_main_quadround 28
+	sha256_avx2_main_quadround 32
+	sha256_avx2_main_quadround 36
+	sha256_avx2_main_quadround 40
+	sha256_avx2_main_quadround 44
+	sha256_avx2_main_quadround 48
+	sha256_avx2_main_quadround 52
+	sha256_avx2_main_round 56, %ymm10, %ymm9, %ymm8, %ymm0, %ymm3, %ymm4, %ymm5, %ymm7
+	jz sha256d_ms_8way_avx2_finish
+	sha256_avx2_main_round 57, %ymm9, %ymm8, %ymm0, %ymm10, %ymm4, %ymm5, %ymm7, %ymm3
+	sha256_avx2_main_round 58, %ymm8, %ymm0, %ymm10, %ymm9, %ymm5, %ymm7, %ymm3, %ymm4
+	sha256_avx2_main_round 59, %ymm0, %ymm10, %ymm9, %ymm8, %ymm7, %ymm3, %ymm4, %ymm5
+	sha256_avx2_main_quadround 60
+	
+	vmovdqa	2*32(%rsp), %ymm1
+	vmovdqa	3*32(%rsp), %ymm2
+	vmovdqa	4*32(%rsp), %ymm6
+	vmovdqa	%ymm1, 18*32(%rsi)
+	vmovdqa	%ymm2, 19*32(%rsi)
+	vmovdqa	%ymm6, 20*32(%rsi)
+	vmovdqa	6*32(%rsp), %ymm1
+	vmovdqa	7*32(%rsp), %ymm2
+	vmovdqa	8*32(%rsp), %ymm6
+	vmovdqa	%ymm1, 22*32(%rsi)
+	vmovdqa	%ymm2, 23*32(%rsi)
+	vmovdqa	%ymm6, 24*32(%rsi)
+	vmovdqa	14*32(%rsp), %ymm1
+	vmovdqa	15*32(%rsp), %ymm2
+	vmovdqa	%ymm1, 30*32(%rsi)
+	vmovdqa	%ymm2, 31*32(%rsi)
+	
+	vpaddd	0(%rdx), %ymm7, %ymm7
+	vpaddd	32(%rdx), %ymm5, %ymm5
+	vpaddd	64(%rdx), %ymm4, %ymm4
+	vpaddd	96(%rdx), %ymm3, %ymm3
+	vpaddd	128(%rdx), %ymm0, %ymm0
+	vpaddd	160(%rdx), %ymm8, %ymm8
+	vpaddd	192(%rdx), %ymm9, %ymm9
+	vpaddd	224(%rdx), %ymm10, %ymm10
+	
+	vmovdqa	%ymm7, 0(%rsp)
+	vmovdqa	%ymm5, 32(%rsp)
+	vmovdqa	%ymm4, 64(%rsp)
+	vmovdqa	%ymm3, 96(%rsp)
+	vmovdqa	%ymm0, 128(%rsp)
+	vmovdqa	%ymm8, 160(%rsp)
+	vmovdqa	%ymm9, 192(%rsp)
+	vmovdqa	%ymm10, 224(%rsp)
+	
+	vpxor	%ymm0, %ymm0, %ymm0
+	movq	$0x8000000000000100, %rax
+	vmovd	%rax, %xmm1
+	vinserti128	$1, %xmm1, %ymm1, %ymm1
+	vpshufd	$0x55, %ymm1, %ymm2
+	vpshufd	$0x00, %ymm1, %ymm1
+	vmovdqa	%ymm2, 8*32(%rsp)
+	vmovdqa	%ymm0, 9*32(%rsp)
+	vmovdqa	%ymm0, 10*32(%rsp)
+	vmovdqa	%ymm0, 11*32(%rsp)
+	vmovdqa	%ymm0, 12*32(%rsp)
+	vmovdqa	%ymm0, 13*32(%rsp)
+	vmovdqa	%ymm0, 14*32(%rsp)
+	vmovdqa	%ymm1, 15*32(%rsp)
+	
+	leaq	16*32(%rsp), %rax
+	cmpq	%rax, %rax
+	
+	vmovdqa	-15*32(%rax), %ymm0
+	vmovdqa	-14*32(%rax), %ymm4
+	vpslld	$14, %ymm0, %ymm2
+	vpslld	$14, %ymm4, %ymm6
+	vpsrld	$3, %ymm0, %ymm8
+	vpsrld	$3, %ymm4, %ymm4
+	vpsrld	$7, %ymm0, %ymm1
+	vpsrld	$4, %ymm4, %ymm5
+	vpxor	%ymm1, %ymm8, %ymm8
+	vpxor	%ymm5, %ymm4, %ymm4
+	vpsrld	$11, %ymm1, %ymm1
+	vpsrld	$11, %ymm5, %ymm5
+	vpxor	%ymm2, %ymm8, %ymm8
+	vpxor	%ymm6, %ymm4, %ymm4
+	vpslld	$11, %ymm2, %ymm2
+	vpslld	$11, %ymm6, %ymm6
+	vpxor	%ymm1, %ymm8, %ymm8
+	vpxor	%ymm5, %ymm4, %ymm4
+	vpxor	%ymm2, %ymm8, %ymm8
+	vpxor	%ymm6, %ymm4, %ymm4
+	vpaddd	%ymm0, %ymm4, %ymm4
+	vpaddd	-16*32(%rax), %ymm8, %ymm3
+	vpaddd	sha256d_8preext2_17(%rip), %ymm4, %ymm7
+	vmovdqa	%ymm3, 0*32(%rax)
+	vmovdqa	%ymm7, 1*32(%rax)
+	
+	sha256_avx2_extend_doubleround 2
+	sha256_avx2_extend_doubleround 4
+	
+	vmovdqa	-9*32(%rax), %ymm0
+	vpslld	$14, %ymm0, %ymm2
+	vpsrld	$3, %ymm0, %ymm8
+	vpsrld	$7, %ymm0, %ymm1
+	vpxor	%ymm1, %ymm8, %ymm8
+	vpxor	%ymm2, %ymm8, %ymm8
+	vpsrld	$11, %ymm1, %ymm1
+	vpslld	$11, %ymm2, %ymm2
+	vpxor	%ymm1, %ymm8, %ymm8
+	vpxor	%ymm2, %ymm8, %ymm8
+	vpaddd	sha256d_8preext2_23(%rip), %ymm0, %ymm4
+	vpaddd	-10*32(%rax), %ymm8, %ymm0
+	vpslld	$13, %ymm3, %ymm2
+	vpslld	$13, %ymm7, %ymm6
+	vpsrld	$10, %ymm3, %ymm3
+	vpsrld	$10, %ymm7, %ymm7
+	vpaddd	-1*32(%rax), %ymm0, %ymm0
+	vpaddd	0*32(%rax), %ymm4, %ymm4
+	vpsrld	$7, %ymm3, %ymm1
+	vpsrld	$7, %ymm7, %ymm5
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpsrld	$2, %ymm1, %ymm1
+	vpsrld	$2, %ymm5, %ymm5
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpslld	$2, %ymm2, %ymm2
+	vpslld	$2, %ymm6, %ymm6
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpaddd	%ymm0, %ymm3, %ymm3
+	vpaddd	%ymm4, %ymm7, %ymm7
+	vmovdqa	%ymm3, 6*32(%rax)
+	vmovdqa	%ymm7, 7*32(%rax)
+	
+	vpslld	$13, %ymm3, %ymm2
+	vpslld	$13, %ymm7, %ymm6
+	vpsrld	$10, %ymm3, %ymm3
+	vpsrld	$10, %ymm7, %ymm7
+	vpsrld	$7, %ymm3, %ymm1
+	vpsrld	$7, %ymm7, %ymm5
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpsrld	$2, %ymm1, %ymm1
+	vpsrld	$2, %ymm5, %ymm5
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpslld	$2, %ymm2, %ymm2
+	vpslld	$2, %ymm6, %ymm6
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpaddd	sha256d_8preext2_24(%rip), %ymm3, %ymm3
+	vpaddd	1*32(%rax), %ymm3, %ymm3
+	vpaddd	2*32(%rax), %ymm7, %ymm7
+	vmovdqa	%ymm3, 8*32(%rax)
+	vmovdqa	%ymm7, 9*32(%rax)
+	
+	vpslld	$13, %ymm3, %ymm2
+	vpslld	$13, %ymm7, %ymm6
+	vpsrld	$10, %ymm3, %ymm3
+	vpsrld	$10, %ymm7, %ymm7
+	vpsrld	$7, %ymm3, %ymm1
+	vpsrld	$7, %ymm7, %ymm5
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpsrld	$2, %ymm1, %ymm1
+	vpsrld	$2, %ymm5, %ymm5
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpslld	$2, %ymm2, %ymm2
+	vpslld	$2, %ymm6, %ymm6
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpaddd	3*32(%rax), %ymm3, %ymm3
+	vpaddd	4*32(%rax), %ymm7, %ymm7
+	vmovdqa	%ymm3, 10*32(%rax)
+	vmovdqa	%ymm7, 11*32(%rax)
+	
+	vpslld	$13, %ymm3, %ymm2
+	vpslld	$13, %ymm7, %ymm6
+	vpsrld	$10, %ymm3, %ymm3
+	vpsrld	$10, %ymm7, %ymm7
+	vpsrld	$7, %ymm3, %ymm1
+	vpsrld	$7, %ymm7, %ymm5
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpsrld	$2, %ymm1, %ymm1
+	vpsrld	$2, %ymm5, %ymm5
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpslld	$2, %ymm2, %ymm2
+	vpslld	$2, %ymm6, %ymm6
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpaddd	5*32(%rax), %ymm3, %ymm3
+	vpaddd	6*32(%rax), %ymm7, %ymm7
+	vmovdqa	%ymm3, 12*32(%rax)
+	vmovdqa	%ymm7, 13*32(%rax)
+	
+	vmovdqa	sha256d_8preext2_30(%rip), %ymm0
+	vmovdqa	0*32(%rax), %ymm4
+	vpslld	$14, %ymm4, %ymm6
+	vpsrld	$3, %ymm4, %ymm4
+	vpsrld	$4, %ymm4, %ymm5
+	vpxor	%ymm5, %ymm4, %ymm4
+	vpxor	%ymm6, %ymm4, %ymm4
+	vpsrld	$11, %ymm5, %ymm5
+	vpslld	$11, %ymm6, %ymm6
+	vpxor	%ymm5, %ymm4, %ymm4
+	vpxor	%ymm6, %ymm4, %ymm4
+	vpaddd	-1*32(%rax), %ymm4, %ymm4
+	vpslld	$13, %ymm3, %ymm2
+	vpslld	$13, %ymm7, %ymm6
+	vpsrld	$10, %ymm3, %ymm3
+	vpsrld	$10, %ymm7, %ymm7
+	vpaddd	7*32(%rax), %ymm0, %ymm0
+	vpaddd	8*32(%rax), %ymm4, %ymm4
+	vpsrld	$7, %ymm3, %ymm1
+	vpsrld	$7, %ymm7, %ymm5
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpsrld	$2, %ymm1, %ymm1
+	vpsrld	$2, %ymm5, %ymm5
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpslld	$2, %ymm2, %ymm2
+	vpslld	$2, %ymm6, %ymm6
+	vpxor	%ymm1, %ymm3, %ymm3
+	vpxor	%ymm5, %ymm7, %ymm7
+	vpxor	%ymm2, %ymm3, %ymm3
+	vpxor	%ymm6, %ymm7, %ymm7
+	vpaddd	%ymm0, %ymm3, %ymm3
+	vpaddd	%ymm4, %ymm7, %ymm7
+	vmovdqa	%ymm3, 14*32(%rax)
+	vmovdqa	%ymm7, 15*32(%rax)
+	
+	jmp sha256d_ms_8way_avx2_extend_loop2
+	
+sha256d_ms_8way_avx2_extend_coda2:
+	sha256_avx2_extend_round 44
+	
+	vmovdqa	sha256_8h+0(%rip), %ymm7
+	vmovdqa	sha256_8h+32(%rip), %ymm5
+	vmovdqa	sha256_8h+64(%rip), %ymm4
+	vmovdqa	sha256_8h+96(%rip), %ymm3
+	vmovdqa	sha256_8h+128(%rip), %ymm0
+	vmovdqa	sha256_8h+160(%rip), %ymm8
+	vmovdqa	sha256_8h+192(%rip), %ymm9
+	vmovdqa	sha256_8h+224(%rip), %ymm10
+	
+	movq	%rsp, %rax
+	leaq	sha256_8k(%rip), %rcx
+	jmp sha256d_ms_8way_avx2_main_loop2
+
+.macro sha256_avx2_main_round_red i, r0, r1, r2, r3, r4
+	vpaddd	32*\i(%rax), \r0, %ymm6
+	vpaddd	32*\i(%rcx), %ymm6, %ymm6
+	vpandn	\r1, \r3, %ymm1
+	vpand	\r3, \r2, %ymm2
+	vpxor	%ymm2, %ymm1, %ymm1
+	vpaddd	%ymm1, %ymm6, %ymm6
+	vpslld	$7, \r3, %ymm1
+	vpsrld	$6, \r3, \r0
+	vpsrld	$5, \r0, %ymm2
+	vpxor	%ymm1, \r0, \r0
+	vpxor	%ymm2, \r0, \r0
+	vpslld	$14, %ymm1, %ymm1
+	vpsrld	$14, %ymm2, %ymm2
+	vpxor	%ymm1, \r0, \r0
+	vpxor	%ymm2, \r0, \r0
+	vpslld	$5, %ymm1, %ymm1
+	vpxor	%ymm1, \r0, \r0
+	vpaddd	\r0, %ymm6, %ymm6
+	vpaddd	%ymm6, \r4, \r0
+.endm
+
+sha256d_ms_8way_avx2_finish:
+	sha256_avx2_main_round_red 57, %ymm9, %ymm8, %ymm0, %ymm10, %ymm4
+	sha256_avx2_main_round_red 58, %ymm8, %ymm0, %ymm10, %ymm9, %ymm5
+	sha256_avx2_main_round_red 59, %ymm0, %ymm10, %ymm9, %ymm8, %ymm7
+	sha256_avx2_main_round_red 60, %ymm10, %ymm9, %ymm8, %ymm0, %ymm3
+	
+	vpaddd	sha256_8h+224(%rip), %ymm10, %ymm10
+	vmovdqa	%ymm10, 224(%rdi)
+	
+	movq	%rbp, %rsp
+	popq	%rbp
+#if defined(WIN64)
+	popq	%rsi
+	vmovdqa	0(%rsp), %xmm6
+	vmovdqa	16(%rsp), %xmm7
+	vmovdqa	32(%rsp), %xmm8
+	vmovdqa	48(%rsp), %xmm9
+	vmovdqa	64(%rsp), %xmm10
+	addq	$80, %rsp
+	popq	%rdi
+#endif
+	ret
+
+
+	.text
+	.p2align 6
+	.globl sha256_use_8way
+	.globl _sha256_use_8way
+sha256_use_8way:
+_sha256_use_8way:
+	pushq	%rbx
+	
+	/* Check for AVX and OSXSAVE support */
+	movl	$1, %eax
+	cpuid
+	andl	$0x18000000, %ecx
+	cmpl	$0x18000000, %ecx
+	jne sha256_use_8way_no
+	/* Check for AVX2 support */
+	movl	$7, %eax
+	xorl	%ecx, %ecx
+	cpuid
+	andl	$0x00000020, %ebx
+	cmpl	$0x00000020, %ebx
+	jne sha256_use_8way_no
+	/* Check for XMM and YMM state support */
+	xorl	%ecx, %ecx
+	xgetbv
+	andl	$0x00000006, %eax
+	cmpl	$0x00000006, %eax
+	jne sha256_use_8way_no
+	
+sha256_use_8way_yes:
+	movl	$1, %eax
+	jmp sha256_use_8way_done
+	
+sha256_use_8way_no:
+	xorl	%eax, %eax
+	
+sha256_use_8way_done:
+	popq	%rbx
+	ret
+
+#endif /* USE_AVX2 */
+
 #endif
diff --git a/sha2.c b/sha2.c
index 76b6ef348..817473586 100644
--- a/sha2.c
+++ b/sha2.c
@@ -522,6 +522,65 @@ static inline int scanhash_sha256d_4way(int thr_id, uint32_t *pdata,
 
 #endif /* HAVE_SHA256_4WAY */
 
+#ifdef HAVE_SHA256_8WAY
+
+void sha256d_ms_8way(uint32_t *hash,  uint32_t *data,
+	const uint32_t *midstate, const uint32_t *prehash);
+
+static inline int scanhash_sha256d_8way(int thr_id, uint32_t *pdata,
+	const uint32_t *ptarget, uint32_t max_nonce, unsigned long *hashes_done)
+{
+	uint32_t data[8 * 64] __attribute__((aligned(128)));
+	uint32_t hash[8 * 8] __attribute__((aligned(32)));
+	uint32_t midstate[8 * 8] __attribute__((aligned(32)));
+	uint32_t prehash[8 * 8] __attribute__((aligned(32)));
+	uint32_t n = pdata[19] - 1;
+	const uint32_t first_nonce = pdata[19];
+	const uint32_t Htarg = ptarget[7];
+	int i, j;
+	
+	memcpy(data, pdata + 16, 64);
+	sha256d_preextend(data);
+	for (i = 31; i >= 0; i--)
+		for (j = 0; j < 8; j++)
+			data[i * 8 + j] = data[i];
+	
+	sha256_init(midstate);
+	sha256_transform(midstate, pdata, 0);
+	memcpy(prehash, midstate, 32);
+	sha256d_prehash(prehash, pdata + 16);
+	for (i = 7; i >= 0; i--) {
+		for (j = 0; j < 8; j++) {
+			midstate[i * 8 + j] = midstate[i];
+			prehash[i * 8 + j] = prehash[i];
+		}
+	}
+	
+	do {
+		for (i = 0; i < 8; i++)
+			data[8 * 3 + i] = ++n;
+		
+		sha256d_ms_8way(hash, data, midstate, prehash);
+		
+		for (i = 0; i < 8; i++) {
+			if (swab32(hash[8 * 7 + i]) <= Htarg) {
+				pdata[19] = data[8 * 3 + i];
+				sha256d_80_swap(hash, pdata);
+				if (fulltest(hash, ptarget)) {
+					*hashes_done = n - first_nonce + 1;
+					return 1;
+				}
+			}
+		}
+	} while (n < max_nonce && !work_restart[thr_id].restart);
+	
+	*hashes_done = n - first_nonce + 1;
+	pdata[19] = n;
+	return 0;
+}
+
+#endif /* HAVE_SHA256_8WAY */
+
 int scanhash_sha256d(int thr_id, uint32_t *pdata, const uint32_t *ptarget,
 	uint32_t max_nonce, unsigned long *hashes_done)
 {
@@ -533,6 +592,11 @@ int scanhash_sha256d(int thr_id, uint32_t *pdata, const uint32_t *ptarget,
 	const uint32_t first_nonce = pdata[19];
 	const uint32_t Htarg = ptarget[7];
 	
+#ifdef HAVE_SHA256_8WAY
+	if (sha256_use_8way())
+		return scanhash_sha256d_8way(thr_id, pdata, ptarget,
+			max_nonce, hashes_done);
+#endif
 #ifdef HAVE_SHA256_4WAY
 	if (sha256_use_4way())
 		return scanhash_sha256d_4way(thr_id, pdata, ptarget,