diff --git a/algo/scrypt.c b/algo/scrypt.c
index f81a2a9..b554971 100644
--- a/algo/scrypt.c
+++ b/algo/scrypt.c
@@ -408,1189 +408,6 @@ void scrypt_core(uint32_t *X, uint32_t *V, int N);
 void scrypt_core_3way(uint32_t *X, uint32_t *V, int N);
 #endif
 
-#elif defined(__aarch64__)
-#include <arm_neon.h>
-
-#undef HAVE_SHA256_4WAY
-#define SCRYPT_MAX_WAYS 3
-#define HAVE_SCRYPT_3WAY 1
-#define scrypt_best_throughput() 3
-
-static inline void xor_salsa8(uint32_t B[16], const uint32_t Bx[16])
-{
-	uint32_t x00,x01,x02,x03,x04,x05,x06,x07,x08,x09,x10,x11,x12,x13,x14,x15;
-	int i;
-
-	x00 = (B[ 0] ^= Bx[ 0]);
-	x01 = (B[ 1] ^= Bx[ 1]);
-	x02 = (B[ 2] ^= Bx[ 2]);
-	x03 = (B[ 3] ^= Bx[ 3]);
-	x04 = (B[ 4] ^= Bx[ 4]);
-	x05 = (B[ 5] ^= Bx[ 5]);
-	x06 = (B[ 6] ^= Bx[ 6]);
-	x07 = (B[ 7] ^= Bx[ 7]);
-	x08 = (B[ 8] ^= Bx[ 8]);
-	x09 = (B[ 9] ^= Bx[ 9]);
-	x10 = (B[10] ^= Bx[10]);
-	x11 = (B[11] ^= Bx[11]);
-	x12 = (B[12] ^= Bx[12]);
-	x13 = (B[13] ^= Bx[13]);
-	x14 = (B[14] ^= Bx[14]);
-	x15 = (B[15] ^= Bx[15]);
-	for (i = 0; i < 8; i += 2) {
-#define R(a, b) (((a) << (b)) | ((a) >> (32 - (b))))
-		/* Operate on columns. */
-		x04 ^= R(x00+x12, 7);	x09 ^= R(x05+x01, 7);
-		x14 ^= R(x10+x06, 7);	x03 ^= R(x15+x11, 7);
-		
-		x08 ^= R(x04+x00, 9);	x13 ^= R(x09+x05, 9);
-		x02 ^= R(x14+x10, 9);	x07 ^= R(x03+x15, 9);
-		
-		x12 ^= R(x08+x04,13);	x01 ^= R(x13+x09,13);
-		x06 ^= R(x02+x14,13);	x11 ^= R(x07+x03,13);
-		
-		x00 ^= R(x12+x08,18);	x05 ^= R(x01+x13,18);
-		x10 ^= R(x06+x02,18);	x15 ^= R(x11+x07,18);
-		
-		/* Operate on rows. */
-		x01 ^= R(x00+x03, 7);	x06 ^= R(x05+x04, 7);
-		x11 ^= R(x10+x09, 7);	x12 ^= R(x15+x14, 7);
-		
-		x02 ^= R(x01+x00, 9);	x07 ^= R(x06+x05, 9);
-		x08 ^= R(x11+x10, 9);	x13 ^= R(x12+x15, 9);
-		
-		x03 ^= R(x02+x01,13);	x04 ^= R(x07+x06,13);
-		x09 ^= R(x08+x11,13);	x14 ^= R(x13+x12,13);
-		
-		x00 ^= R(x03+x02,18);	x05 ^= R(x04+x07,18);
-		x10 ^= R(x09+x08,18);	x15 ^= R(x14+x13,18);
-#undef R
-	}
-	B[ 0] += x00;
-	B[ 1] += x01;
-	B[ 2] += x02;
-	B[ 3] += x03;
-	B[ 4] += x04;
-	B[ 5] += x05;
-	B[ 6] += x06;
-	B[ 7] += x07;
-	B[ 8] += x08;
-	B[ 9] += x09;
-	B[10] += x10;
-	B[11] += x11;
-	B[12] += x12;
-	B[13] += x13;
-	B[14] += x14;
-	B[15] += x15;
-}
-
-
-static inline void xor_salsa8_prefetch(uint32_t B[16], const uint32_t Bx[16], uint32_t* V, uint32_t N)
-{
-	uint32_t x00,x01,x02,x03,x04,x05,x06,x07,x08,x09,x10,x11,x12,x13,x14,x15;
-	int i;
-
-	x00 = (B[ 0] ^= Bx[ 0]);
-	x01 = (B[ 1] ^= Bx[ 1]);
-	x02 = (B[ 2] ^= Bx[ 2]);
-	x03 = (B[ 3] ^= Bx[ 3]);
-	x04 = (B[ 4] ^= Bx[ 4]);
-	x05 = (B[ 5] ^= Bx[ 5]);
-	x06 = (B[ 6] ^= Bx[ 6]);
-	x07 = (B[ 7] ^= Bx[ 7]);
-	x08 = (B[ 8] ^= Bx[ 8]);
-	x09 = (B[ 9] ^= Bx[ 9]);
-	x10 = (B[10] ^= Bx[10]);
-	x11 = (B[11] ^= Bx[11]);
-	x12 = (B[12] ^= Bx[12]);
-	x13 = (B[13] ^= Bx[13]);
-	x14 = (B[14] ^= Bx[14]);
-	x15 = (B[15] ^= Bx[15]);
-	for (i = 0; i < 8; i += 2) {
-#define R(a, b) (((a) << (b)) | ((a) >> (32 - (b))))
-		/* Operate on columns. */
-		x04 ^= R(x00+x12, 7);	x09 ^= R(x05+x01, 7);
-		x14 ^= R(x10+x06, 7);	x03 ^= R(x15+x11, 7);
-		
-		x08 ^= R(x04+x00, 9);	x13 ^= R(x09+x05, 9);
-		x02 ^= R(x14+x10, 9);	x07 ^= R(x03+x15, 9);
-		
-		x12 ^= R(x08+x04,13);	x01 ^= R(x13+x09,13);
-		x06 ^= R(x02+x14,13);	x11 ^= R(x07+x03,13);
-		
-		x00 ^= R(x12+x08,18);	x05 ^= R(x01+x13,18);
-		x10 ^= R(x06+x02,18);	x15 ^= R(x11+x07,18);
-		
-		/* Operate on rows. */
-		x01 ^= R(x00+x03, 7);	x06 ^= R(x05+x04, 7);
-		x11 ^= R(x10+x09, 7);	x12 ^= R(x15+x14, 7);
-		
-		x02 ^= R(x01+x00, 9);	x07 ^= R(x06+x05, 9);
-		x08 ^= R(x11+x10, 9);	x13 ^= R(x12+x15, 9);
-		
-		x03 ^= R(x02+x01,13);	x04 ^= R(x07+x06,13);
-		x09 ^= R(x08+x11,13);	x14 ^= R(x13+x12,13);
-		
-		x00 ^= R(x03+x02,18);	x05 ^= R(x04+x07,18);
-		x10 ^= R(x09+x08,18);	x15 ^= R(x14+x13,18);
-#undef R
-	}
-	B[ 0] += x00;
-	uint32_t one = 32 * (B[0] & (N - 1));
-	__builtin_prefetch(&V[one + 0]);
-	__builtin_prefetch(&V[one + 8]);
-	__builtin_prefetch(&V[one + 16]);
-	__builtin_prefetch(&V[one + 24]);
-	asm("":::"memory");
-	B[ 1] += x01;
-	B[ 2] += x02;
-	B[ 3] += x03;
-	B[ 4] += x04;
-	B[ 5] += x05;
-	B[ 6] += x06;
-	B[ 7] += x07;
-	B[ 8] += x08;
-	B[ 9] += x09;
-	B[10] += x10;
-	B[11] += x11;
-	B[12] += x12;
-	B[13] += x13;
-	B[14] += x14;
-	B[15] += x15;
-}
-
-static inline void scrypt_core(uint32_t *X, uint32_t *V, int N)
-{
-	int i;
-
-	for (i = 0; i < N; i++) {
-		memcpy(&V[i * 32], X, 128);
-		xor_salsa8(&X[0], &X[16]);
-		xor_salsa8(&X[16], &X[0]);
-	}
-	for (i = 0; i < N; i++) {
-		uint32_t j = 32 * (X[16] & (N - 1));
-		for (uint8_t k = 0; k < 32; k++)
-			X[k] ^= V[j + k];
-		xor_salsa8(&X[0], &X[16]);
-		xor_salsa8_prefetch(&X[16], &X[0], V, N);
-	}
-}
-
-
-static inline void scrypt_shuffle(uint32_t B[16])
-{
-	uint32_t x0 = 	B[0]; 
-	uint32_t x1 = 	B[1]; 
-	uint32_t x2 = 	B[2];
-	uint32_t x3 = 	B[3]; 
-	uint32_t x4 = 	B[4];
-	uint32_t x5 = 	B[5]; 
-	uint32_t x6 = 	B[6]; 
-	uint32_t x7 = 	B[7]; 
-	uint32_t x8 = 	B[8]; 
-	uint32_t x9 = 	B[9]; 
-	uint32_t x10 = B[10]; 
-	uint32_t x11 = B[11]; 
-	uint32_t x12 = B[12];
-	uint32_t x13 = B[13]; 
-	uint32_t x14 = B[14]; 
-	uint32_t x15 = B[15]; 
-
-	B[0] = x0;  B[1] = x5;  B[2] = x10;  B[3] = x15;
-	B[4] = x12; B[5] = x1;  B[6] = x6;   B[7] = x11;
-	B[8] = x8;  B[9] = x13; B[10] = x2;  B[11] = x7;
-	B[12] = x4; B[13] = x9; B[14] = x14; B[15] = x3;
-}
-
-static inline void scrypt_core_3way(uint32_t B[32 * 3], uint32_t *V, uint32_t N)
-{
-	uint32_t* W = V;
-
-	scrypt_shuffle(&B[0  + 0]);
-	scrypt_shuffle(&B[16 + 0]);
-	scrypt_shuffle(&B[0 + 32]);
-	scrypt_shuffle(&B[16 + 32]);
-	scrypt_shuffle(&B[0 + 64]);
-	scrypt_shuffle(&B[16 + 64]);
-
-	uint32x4x4_t q_a, q_b, q_c, q_tmp;
-	uint32x4x4_t ba_a, bb_a, bc_a, ba_b, bb_b, bc_b;
-
-	ba_a.val[0] = vld1q_u32(&B[( 0) / 4]);
-	ba_a.val[1] = vld1q_u32(&B[(16) / 4]);
-	ba_a.val[2] = vld1q_u32(&B[(32) / 4]);
-	ba_a.val[3] = vld1q_u32(&B[(48) / 4]);
-	ba_b.val[0] = vld1q_u32(&B[(0 + 64 + 0) / 4]);
-	ba_b.val[1] = vld1q_u32(&B[(0 + 64 + 16) / 4]);
-	ba_b.val[2] = vld1q_u32(&B[(0 + 64 + 32) / 4]);
-	ba_b.val[3] = vld1q_u32(&B[(0 + 64 + 48) / 4]);
-
-	bb_a.val[0] = vld1q_u32(&B[(128 +  0) / 4]);
-	bb_a.val[1] = vld1q_u32(&B[(128 + 16) / 4]);
-	bb_a.val[2] = vld1q_u32(&B[(128 + 32) / 4]);
-	bb_a.val[3] = vld1q_u32(&B[(128 + 48) / 4]);
-	bb_b.val[0] = vld1q_u32(&B[(128 + 64 + 0) / 4]);
-	bb_b.val[1] = vld1q_u32(&B[(128 + 64 + 16) / 4]);
-	bb_b.val[2] = vld1q_u32(&B[(128 + 64 + 32) / 4]);
-	bb_b.val[3] = vld1q_u32(&B[(128 + 64 + 48) / 4]);
-	
-	bc_a.val[0] = vld1q_u32(&B[(256 + 0) / 4]);
-	bc_a.val[1] = vld1q_u32(&B[(256 + 16) / 4]);
-	bc_a.val[2] = vld1q_u32(&B[(256 + 32) / 4]);
-	bc_a.val[3] = vld1q_u32(&B[(256 + 48) / 4]);
-	bc_b.val[0] = vld1q_u32(&B[(256 + 64 + 0) / 4]);
-	bc_b.val[1] = vld1q_u32(&B[(256 + 64 + 16) / 4]);
-	bc_b.val[2] = vld1q_u32(&B[(256 + 64 + 32) / 4]);
-	bc_b.val[3] = vld1q_u32(&B[(256 + 64 + 48) / 4]);
-
-	// prep
-
-	vst1q_u32(&V[( 0) / 4], ba_a.val[0]);
-	vst1q_u32(&V[(16) / 4], ba_a.val[1]);
-	vst1q_u32(&V[(32) / 4], ba_a.val[2]);
-	vst1q_u32(&V[(48) / 4], ba_a.val[3]);
-	vst1q_u32(&V[(64) / 4],  ba_b.val[0]);
-	vst1q_u32(&V[(80) / 4],  ba_b.val[1]);
-	vst1q_u32(&V[(96) / 4],  ba_b.val[2]);
-	vst1q_u32(&V[(112) / 4], ba_b.val[3]);
-
-	vst1q_u32(&V[(128 +  0) / 4], bb_a.val[0]);
-	vst1q_u32(&V[(128 + 16) / 4], bb_a.val[1]);
-	vst1q_u32(&V[(128 + 32) / 4], bb_a.val[2]);
-	vst1q_u32(&V[(128 + 48) / 4], bb_a.val[3]);
-	vst1q_u32(&V[(128 + 64) / 4],  bb_b.val[0]);
-	vst1q_u32(&V[(128 + 80) / 4],  bb_b.val[1]);
-	vst1q_u32(&V[(128 + 96) / 4],  bb_b.val[2]);
-	vst1q_u32(&V[(128 + 112) / 4], bb_b.val[3]);
-
-	vst1q_u32(&V[(256 +  0) / 4], bc_a.val[0]);
-	vst1q_u32(&V[(256 + 16) / 4], bc_a.val[1]);
-	vst1q_u32(&V[(256 + 32) / 4], bc_a.val[2]);
-	vst1q_u32(&V[(256 + 48) / 4], bc_a.val[3]);
-	vst1q_u32(&V[(256 + 64) / 4], bc_b.val[0]);
-	vst1q_u32(&V[(256 + 80) / 4], bc_b.val[1]);
-	vst1q_u32(&V[(256 + 96) / 4], bc_b.val[2]);
-	vst1q_u32(&V[(256 + 112) / 4],bc_b.val[3]);
-
-	V += 96;
-
-	for (int n = 0; n < N; n++)
-	{
-		// loop 1 part a
-		q_a.val[0] = veorq_u32(ba_b.val[0], ba_a.val[0]);
-		q_a.val[1] = veorq_u32(ba_b.val[1], ba_a.val[1]);
-		q_a.val[2] = veorq_u32(ba_b.val[2], ba_a.val[2]);
-		q_a.val[3] = veorq_u32(ba_b.val[3], ba_a.val[3]);
-
-		q_b.val[0] = veorq_u32(bb_b.val[0], bb_a.val[0]);
-		q_b.val[1] = veorq_u32(bb_b.val[1], bb_a.val[1]);
-		q_b.val[2] = veorq_u32(bb_b.val[2], bb_a.val[2]);
-		q_b.val[3] = veorq_u32(bb_b.val[3], bb_a.val[3]);
-
-		q_c.val[0] = veorq_u32(bc_b.val[0], bc_a.val[0]);
-		q_c.val[1] = veorq_u32(bc_b.val[1], bc_a.val[1]);
-		q_c.val[2] = veorq_u32(bc_b.val[2], bc_a.val[2]);
-		q_c.val[3] = veorq_u32(bc_b.val[3], bc_a.val[3]);
-
-		ba_a = q_a;
-		bb_a = q_b;
-		bc_a = q_c;
-
-		for (int i = 0; i < 4; i ++)
-		{
-			q_tmp.val[0] = vaddq_u32(q_a.val[0], q_a.val[1]);  	
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 7);	
-			q_tmp.val[2] = vaddq_u32(q_b.val[0], q_b.val[1]);  	
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 25);
-			q_a.val[3] = veorq_u32(q_tmp.val[1], q_a.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 7);
-			q_tmp.val[3] = vaddq_u32(q_c.val[0], q_c.val[1]); 
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 25);
-			q_b.val[3] = veorq_u32(q_tmp.val[1], q_b.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 7); 				
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 25);				
-			q_c.val[3] = veorq_u32(q_tmp.val[1], q_c.val[3]);
-
-			q_tmp.val[0] = vaddq_u32(q_a.val[3], q_a.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 9);
-			q_tmp.val[2] = vaddq_u32(q_b.val[3], q_b.val[0]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 23);
-			q_a.val[2] = veorq_u32(q_tmp.val[1], q_a.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 9);
-			q_tmp.val[3] = vaddq_u32(q_c.val[3], q_c.val[0]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 23);
-			q_b.val[2] = veorq_u32(q_tmp.val[1], q_b.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 9);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 23);
-			q_c.val[2] = veorq_u32(q_tmp.val[1], q_c.val[2]);
-
-			q_tmp.val[0] = vaddq_u32(q_a.val[2], q_a.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 13);
-			q_tmp.val[2] = vaddq_u32(q_b.val[2], q_b.val[3]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 19);
-			q_a.val[1] = veorq_u32(q_tmp.val[1], q_a.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 13);
-			q_tmp.val[3] = vaddq_u32(q_c.val[2], q_c.val[3]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 19);
-			q_b.val[1] = veorq_u32(q_tmp.val[1], q_b.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 13);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 19);
-			q_c.val[1] = veorq_u32(q_tmp.val[1], q_c.val[1]);
-			
-			q_a.val[3] = vextq_u32(q_a.val[3], q_a.val[3], 3);
-			q_b.val[3] = vextq_u32(q_b.val[3], q_b.val[3], 3);
-			q_c.val[3] = vextq_u32(q_c.val[3], q_c.val[3], 3);
-			
-			q_tmp.val[0] = vaddq_u32(q_a.val[1], q_a.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 18);
-			q_tmp.val[2] = vaddq_u32(q_b.val[1], q_b.val[2]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 14);
-			q_a.val[0] = veorq_u32(q_tmp.val[1], q_a.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 18);
-			q_tmp.val[3] = vaddq_u32(q_c.val[1], q_c.val[2]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 14);
-			q_b.val[0] = veorq_u32(q_tmp.val[1], q_b.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 18);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 14);
-			q_c.val[0] = veorq_u32(q_tmp.val[1], q_c.val[0]);
-			
-			q_a.val[2] = vextq_u32(q_a.val[2], q_a.val[2], 2);
-			q_b.val[2] = vextq_u32(q_b.val[2], q_b.val[2], 2);
-			q_c.val[2] = vextq_u32(q_c.val[2], q_c.val[2], 2);
-			
-			q_tmp.val[0] = vaddq_u32(q_a.val[0], q_a.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 7);
-			q_tmp.val[2] = vaddq_u32(q_b.val[0], q_b.val[3]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 25);
-			q_a.val[1] = vextq_u32(q_a.val[1], q_a.val[1], 1);
-			q_a.val[1] = veorq_u32(q_tmp.val[1], q_a.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 7);
-			q_tmp.val[3] = vaddq_u32(q_c.val[0], q_c.val[3]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 25);
-			q_b.val[1] = vextq_u32(q_b.val[1], q_b.val[1], 1);
-			q_b.val[1] = veorq_u32(q_tmp.val[1], q_b.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 7);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 25);
-			q_c.val[1] = vextq_u32(q_c.val[1], q_c.val[1], 1);
-			q_c.val[1] = veorq_u32(q_tmp.val[1], q_c.val[1]);
-
-			q_tmp.val[0] = vaddq_u32(q_a.val[1], q_a.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 9);
-			q_tmp.val[2] = vaddq_u32(q_b.val[1], q_b.val[0]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 23);
-			q_a.val[2] = veorq_u32(q_tmp.val[1], q_a.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 9);
-			q_tmp.val[3] = vaddq_u32(q_c.val[1], q_c.val[0]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 23);
-			q_b.val[2] = veorq_u32(q_tmp.val[1], q_b.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 9);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 23);
-			q_c.val[2] = veorq_u32(q_tmp.val[1], q_c.val[2]);
-
-			q_tmp.val[0] = vaddq_u32(q_a.val[2], q_a.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 13);
-			q_tmp.val[2] = vaddq_u32(q_b.val[2], q_b.val[1]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 19);
-			q_a.val[3] = veorq_u32(q_tmp.val[1], q_a.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 13);
-			q_tmp.val[3] = vaddq_u32(q_c.val[2], q_c.val[1]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 19);
-			q_b.val[3] = veorq_u32(q_tmp.val[1], q_b.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 13);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 19);
-			q_c.val[3] = veorq_u32(q_tmp.val[1], q_c.val[3]);
-			q_a.val[1] = vextq_u32(q_a.val[1], q_a.val[1], 3);
-			q_b.val[1] = vextq_u32(q_b.val[1], q_b.val[1], 3);
-			q_c.val[1] = vextq_u32(q_c.val[1], q_c.val[1], 3);
-
-			q_tmp.val[0] = vaddq_u32(q_a.val[3], q_a.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 18);
-			q_tmp.val[2] = vaddq_u32(q_b.val[3], q_b.val[2]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 14);
-			q_a.val[2] = vextq_u32(q_a.val[2], q_a.val[2], 2);
-			q_b.val[2] = vextq_u32(q_b.val[2], q_b.val[2], 2);
-			q_a.val[0] = veorq_u32(q_tmp.val[1], q_a.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 18);
-			q_tmp.val[3] = vaddq_u32(q_c.val[3], q_c.val[2]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 14);
-			q_c.val[2] = vextq_u32(q_c.val[2], q_c.val[2], 2);
-			q_b.val[3] = vextq_u32(q_b.val[3], q_b.val[3], 1);
-			q_b.val[0] = veorq_u32(q_tmp.val[1], q_b.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 18);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 14);
-			q_a.val[3] = vextq_u32(q_a.val[3], q_a.val[3], 1);
-			q_c.val[3] = vextq_u32(q_c.val[3], q_c.val[3], 1);
-			q_c.val[0] = veorq_u32(q_tmp.val[1], q_c.val[0]);
-		}
-		ba_a.val[0] = vaddq_u32(ba_a.val[0], q_a.val[0]);
-		ba_a.val[1] = vaddq_u32(ba_a.val[1], q_a.val[1]);
-		ba_a.val[2] = vaddq_u32(ba_a.val[2], q_a.val[2]);
-		ba_a.val[3] = vaddq_u32(ba_a.val[3], q_a.val[3]);
-
-		q_a = ba_a;
-
-		bb_a.val[0] = vaddq_u32(bb_a.val[0], q_b.val[0]);
-		bb_a.val[1] = vaddq_u32(bb_a.val[1], q_b.val[1]);
-		bb_a.val[2] = vaddq_u32(bb_a.val[2], q_b.val[2]);
-		bb_a.val[3] = vaddq_u32(bb_a.val[3], q_b.val[3]);
-
-		q_b = bb_a;
-
-		bc_a.val[0] = vaddq_u32(bc_a.val[0], q_c.val[0]);
-		bc_a.val[1] = vaddq_u32(bc_a.val[1], q_c.val[1]);
-		bc_a.val[2] = vaddq_u32(bc_a.val[2], q_c.val[2]);
-		bc_a.val[3] = vaddq_u32(bc_a.val[3], q_c.val[3]);
-
-		q_c = bc_a;
-		
-		for (int i = 0; i < 4; i++)
-		{
-			vst1q_u32(&V[      (i * 4) ], ba_a.val[i]);
-			vst1q_u32(&V[(32 + (i * 4))], bb_a.val[i]);
-			vst1q_u32(&V[(64 + (i * 4))], bc_a.val[i]);
-		}
-
-		// loop 1 part b
-
-		q_a.val[0] = veorq_u32(ba_b.val[0], q_a.val[0]);
-		q_a.val[1] = veorq_u32(ba_b.val[1], q_a.val[1]);
-		q_a.val[2] = veorq_u32(ba_b.val[2], q_a.val[2]);
-		q_a.val[3] = veorq_u32(ba_b.val[3], q_a.val[3]);
-		ba_b = q_a;
-
-		q_b.val[0] = veorq_u32(bb_b.val[0], q_b.val[0]);
-		q_b.val[1] = veorq_u32(bb_b.val[1], q_b.val[1]);
-		q_b.val[2] = veorq_u32(bb_b.val[2], q_b.val[2]);
-		q_b.val[3] = veorq_u32(bb_b.val[3], q_b.val[3]);
-		bb_b = q_b;
-
-		q_c.val[0] = veorq_u32(bc_b.val[0], q_c.val[0]);
-		q_c.val[1] = veorq_u32(bc_b.val[1], q_c.val[1]);
-		q_c.val[2] = veorq_u32(bc_b.val[2], q_c.val[2]);
-		q_c.val[3] = veorq_u32(bc_b.val[3], q_c.val[3]);
-		bc_b = q_c;
-
-
-		for (int i = 0; i < 4; i ++)
-		{
-			q_tmp.val[0] = vaddq_u32(q_a.val[0], q_a.val[1]);  	
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 7);	
-			q_tmp.val[2] = vaddq_u32(q_b.val[0], q_b.val[1]);  	
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 25);
-			q_a.val[3] = veorq_u32(q_tmp.val[1], q_a.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 7);
-			q_tmp.val[3] = vaddq_u32(q_c.val[0], q_c.val[1]); 
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 25);
-			q_b.val[3] = veorq_u32(q_tmp.val[1], q_b.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 7); 				
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 25);				
-			q_c.val[3] = veorq_u32(q_tmp.val[1], q_c.val[3]);
-
-			q_tmp.val[0] = vaddq_u32(q_a.val[3], q_a.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 9);
-			q_tmp.val[2] = vaddq_u32(q_b.val[3], q_b.val[0]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 23);
-			q_a.val[2] = veorq_u32(q_tmp.val[1], q_a.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 9);
-			q_tmp.val[3] = vaddq_u32(q_c.val[3], q_c.val[0]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 23);
-			q_b.val[2] = veorq_u32(q_tmp.val[1], q_b.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 9);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 23);
-			q_c.val[2] = veorq_u32(q_tmp.val[1], q_c.val[2]);
-
-			q_tmp.val[0] = vaddq_u32(q_a.val[2], q_a.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 13);
-			q_tmp.val[2] = vaddq_u32(q_b.val[2], q_b.val[3]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 19);
-			q_a.val[1] = veorq_u32(q_tmp.val[1], q_a.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 13);
-			q_tmp.val[3] = vaddq_u32(q_c.val[2], q_c.val[3]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 19);
-			q_b.val[1] = veorq_u32(q_tmp.val[1], q_b.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 13);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 19);
-			q_c.val[1] = veorq_u32(q_tmp.val[1], q_c.val[1]);
-			
-			q_a.val[3] = vextq_u32(q_a.val[3], q_a.val[3], 3);
-			q_b.val[3] = vextq_u32(q_b.val[3], q_b.val[3], 3);
-			q_c.val[3] = vextq_u32(q_c.val[3], q_c.val[3], 3);
-			
-			q_tmp.val[0] = vaddq_u32(q_a.val[1], q_a.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 18);
-			q_tmp.val[2] = vaddq_u32(q_b.val[1], q_b.val[2]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 14);
-			q_a.val[0] = veorq_u32(q_tmp.val[1], q_a.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 18);
-			q_tmp.val[3] = vaddq_u32(q_c.val[1], q_c.val[2]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 14);
-			q_b.val[0] = veorq_u32(q_tmp.val[1], q_b.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 18);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 14);
-			q_c.val[0] = veorq_u32(q_tmp.val[1], q_c.val[0]);
-			
-			q_a.val[2] = vextq_u32(q_a.val[2], q_a.val[2], 2);
-			q_b.val[2] = vextq_u32(q_b.val[2], q_b.val[2], 2);
-			q_c.val[2] = vextq_u32(q_c.val[2], q_c.val[2], 2);
-			
-			q_tmp.val[0] = vaddq_u32(q_a.val[0], q_a.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 7);
-			q_tmp.val[2] = vaddq_u32(q_b.val[0], q_b.val[3]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 25);
-			q_a.val[1] = vextq_u32(q_a.val[1], q_a.val[1], 1);
-			q_a.val[1] = veorq_u32(q_tmp.val[1], q_a.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 7);
-			q_tmp.val[3] = vaddq_u32(q_c.val[0], q_c.val[3]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 25);
-			q_b.val[1] = vextq_u32(q_b.val[1], q_b.val[1], 1);
-			q_b.val[1] = veorq_u32(q_tmp.val[1], q_b.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 7);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 25);
-			q_c.val[1] = vextq_u32(q_c.val[1], q_c.val[1], 1);
-			q_c.val[1] = veorq_u32(q_tmp.val[1], q_c.val[1]);
-
-			q_tmp.val[0] = vaddq_u32(q_a.val[1], q_a.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 9);
-			q_tmp.val[2] = vaddq_u32(q_b.val[1], q_b.val[0]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 23);
-			q_a.val[2] = veorq_u32(q_tmp.val[1], q_a.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 9);
-			q_tmp.val[3] = vaddq_u32(q_c.val[1], q_c.val[0]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 23);
-			q_b.val[2] = veorq_u32(q_tmp.val[1], q_b.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 9);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 23);
-			q_c.val[2] = veorq_u32(q_tmp.val[1], q_c.val[2]);
-
-			q_tmp.val[0] = vaddq_u32(q_a.val[2], q_a.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 13);
-			q_tmp.val[2] = vaddq_u32(q_b.val[2], q_b.val[1]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 19);
-			q_a.val[3] = veorq_u32(q_tmp.val[1], q_a.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 13);
-			q_tmp.val[3] = vaddq_u32(q_c.val[2], q_c.val[1]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 19);
-			q_b.val[3] = veorq_u32(q_tmp.val[1], q_b.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 13);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 19);
-			q_c.val[3] = veorq_u32(q_tmp.val[1], q_c.val[3]);
-			q_a.val[1] = vextq_u32(q_a.val[1], q_a.val[1], 3);
-			q_b.val[1] = vextq_u32(q_b.val[1], q_b.val[1], 3);
-			q_c.val[1] = vextq_u32(q_c.val[1], q_c.val[1], 3);
-
-			q_tmp.val[0] = vaddq_u32(q_a.val[3], q_a.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 18);
-			q_tmp.val[2] = vaddq_u32(q_b.val[3], q_b.val[2]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 14);
-			q_a.val[2] = vextq_u32(q_a.val[2], q_a.val[2], 2);
-			q_b.val[2] = vextq_u32(q_b.val[2], q_b.val[2], 2);
-			q_a.val[0] = veorq_u32(q_tmp.val[1], q_a.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 18);
-			q_tmp.val[3] = vaddq_u32(q_c.val[3], q_c.val[2]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 14);
-			q_c.val[2] = vextq_u32(q_c.val[2], q_c.val[2], 2);
-			q_b.val[3] = vextq_u32(q_b.val[3], q_b.val[3], 1);
-			q_b.val[0] = veorq_u32(q_tmp.val[1], q_b.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 18);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 14);
-			q_a.val[3] = vextq_u32(q_a.val[3], q_a.val[3], 1);
-			q_c.val[3] = vextq_u32(q_c.val[3], q_c.val[3], 1);
-			q_c.val[0] = veorq_u32(q_tmp.val[1], q_c.val[0]);
-		}
-
-		ba_b.val[0] = vaddq_u32(q_a.val[0], ba_b.val[0]);
-		ba_b.val[1] = vaddq_u32(q_a.val[1], ba_b.val[1]);
-		ba_b.val[2] = vaddq_u32(q_a.val[2], ba_b.val[2]);
-		ba_b.val[3] = vaddq_u32(q_a.val[3], ba_b.val[3]);
-		bb_b.val[0] = vaddq_u32(q_b.val[0], bb_b.val[0]);
-		bb_b.val[1] = vaddq_u32(q_b.val[1], bb_b.val[1]);
-		bb_b.val[2] = vaddq_u32(q_b.val[2], bb_b.val[2]);
-		bb_b.val[3] = vaddq_u32(q_b.val[3], bb_b.val[3]);
-		bc_b.val[0] = vaddq_u32(q_c.val[0], bc_b.val[0]);
-		bc_b.val[1] = vaddq_u32(q_c.val[1], bc_b.val[1]);
-		bc_b.val[2] = vaddq_u32(q_c.val[2], bc_b.val[2]);
-		bc_b.val[3] = vaddq_u32(q_c.val[3], bc_b.val[3]);
-		for (int i = 0; i < 4; i++)
-		{
-			vst1q_u32(&V[(     16 + (i * 4))], ba_b.val[i]);
-			vst1q_u32(&V[(32 + 16 + (i * 4))], bb_b.val[i]);
-			vst1q_u32(&V[(64 + 16 + (i * 4))], bc_b.val[i]);
-		}
-		V += 96;
-	}
-	V = W;
-
-    // loop 2
-
-	uint32x4x4_t x;
-
-	uint32_t one =   32 * (3 * (ba_b.val[0][0] & (N - 1)) + 0);
-	uint32_t two =   32 * (3 * (bb_b.val[0][0] & (N - 1)) + 1);
-	uint32_t three = 32 * (3 * (bc_b.val[0][0] & (N - 1)) + 2);
-	x.val[0] = vld1q_u32(&W[one +  0]);
-	x.val[1] = vld1q_u32(&W[one +  4]);
-	x.val[2] = vld1q_u32(&W[one +  8]);
-	x.val[3] = vld1q_u32(&W[one + 12]);
-
-	for (int n = 0; n < N; n++)
-	{
-		// loop 2 part a
-
-		ba_a.val[0] = veorq_u32(ba_a.val[0], x.val[0]);
-			x.val[0] = vld1q_u32(&W[one + 16 +  0]);
-		ba_a.val[1] = veorq_u32(ba_a.val[1], x.val[1]);
-			x.val[1] = vld1q_u32(&W[one + 16 +  4]);
-		ba_a.val[2] = veorq_u32(ba_a.val[2], x.val[2]);
-			x.val[2] = vld1q_u32(&W[one + 16 +  8]);
-		ba_a.val[3] = veorq_u32(ba_a.val[3], x.val[3]);
-
-			ba_b.val[0] = veorq_u32(ba_b.val[0], x.val[0]);
-			ba_b.val[1] = veorq_u32(ba_b.val[1], x.val[1]);
-			x.val[3] = vld1q_u32(&W[one + 16 + 12]);
-			ba_b.val[2] = veorq_u32(ba_b.val[2], x.val[2]);
-			ba_b.val[3] = veorq_u32(ba_b.val[3], x.val[3]);
-		x.val[0] = vld1q_u32(&W[two +  0]);
-				q_a.val[0] = veorq_u32(ba_b.val[0], ba_a.val[0]);
-				q_a.val[1] = veorq_u32(ba_b.val[1], ba_a.val[1]);
-		x.val[1] = vld1q_u32(&W[two +  4]);
-				q_a.val[2] = veorq_u32(ba_b.val[2], ba_a.val[2]);
-				q_a.val[3] = veorq_u32(ba_b.val[3], ba_a.val[3]);
-		x.val[2] = vld1q_u32(&W[two +  8]);
-		ba_a = q_a;
-
-		x.val[3] = vld1q_u32(&W[two + 12]);
-
-		bb_a.val[0] = veorq_u32(bb_a.val[0], x.val[0]);
-			x.val[0] = vld1q_u32(&W[two + 16 +  0]);
-		bb_a.val[1] = veorq_u32(bb_a.val[1], x.val[1]);
-			x.val[1] = vld1q_u32(&W[two + 16 +  4]);
-		bb_a.val[2] = veorq_u32(bb_a.val[2], x.val[2]);
-			x.val[2] = vld1q_u32(&W[two + 16 +  8]);
-		bb_a.val[3] = veorq_u32(bb_a.val[3], x.val[3]);
-			bb_b.val[0] = veorq_u32(bb_b.val[0], x.val[0]);
-			x.val[3] = vld1q_u32(&W[two + 16 + 12]);
-			bb_b.val[1] = veorq_u32(bb_b.val[1], x.val[1]);
-		x.val[0] = vld1q_u32(&W[three +  0]);
-			bb_b.val[2] = veorq_u32(bb_b.val[2], x.val[2]);
-			bb_b.val[3] = veorq_u32(bb_b.val[3], x.val[3]);
-		x.val[1] = vld1q_u32(&W[three +  4]);
-				q_b.val[0] = veorq_u32(bb_b.val[0], bb_a.val[0]);
-				q_b.val[1] = veorq_u32(bb_b.val[1], bb_a.val[1]);
-		x.val[2] = vld1q_u32(&W[three +  8]);
-				q_b.val[2] = veorq_u32(bb_b.val[2], bb_a.val[2]);
-				q_b.val[3] = veorq_u32(bb_b.val[3], bb_a.val[3]);
-		x.val[3] = vld1q_u32(&W[three + 12]);
-		bb_a = q_b;
-
-		bc_a.val[0] = veorq_u32(bc_a.val[0], x.val[0]);
-			x.val[0] = vld1q_u32(&W[three + 16 +  0]);
-		bc_a.val[1] = veorq_u32(bc_a.val[1], x.val[1]);
-			x.val[1] = vld1q_u32(&W[three + 16 +  4]);
-		bc_a.val[2] = veorq_u32(bc_a.val[2], x.val[2]);
-			x.val[2] = vld1q_u32(&W[three + 16 +  8]);
-		bc_a.val[3] = veorq_u32(bc_a.val[3], x.val[3]);
-			bc_b.val[0] = veorq_u32(bc_b.val[0], x.val[0]);
-			x.val[3] = vld1q_u32(&W[three + 16 + 12]);
-			bc_b.val[1] = veorq_u32(bc_b.val[1], x.val[1]);
-			bc_b.val[2] = veorq_u32(bc_b.val[2], x.val[2]);
-			bc_b.val[3] = veorq_u32(bc_b.val[3], x.val[3]);
-				q_c.val[0] = veorq_u32(bc_b.val[0], bc_a.val[0]);
-				q_c.val[1] = veorq_u32(bc_b.val[1], bc_a.val[1]);
-				q_c.val[2] = veorq_u32(bc_b.val[2], bc_a.val[2]);
-				q_c.val[3] = veorq_u32(bc_b.val[3], bc_a.val[3]);
-		bc_a = q_c;
-
-		for (int i = 0; i < 4; i++)
-		{
-			q_tmp.val[0] = vaddq_u32(q_a.val[0], q_a.val[1]);  	
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 7);	
-			q_tmp.val[2] = vaddq_u32(q_b.val[0], q_b.val[1]);  	
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 25);
-			q_a.val[3] = veorq_u32(q_tmp.val[1], q_a.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 7);
-			q_tmp.val[3] = vaddq_u32(q_c.val[0], q_c.val[1]); 
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 25);
-			q_b.val[3] = veorq_u32(q_tmp.val[1], q_b.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 7); 				
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 25);				
-			q_c.val[3] = veorq_u32(q_tmp.val[1], q_c.val[3]);
-
-			q_tmp.val[0] = vaddq_u32(q_a.val[3], q_a.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 9);
-			q_tmp.val[2] = vaddq_u32(q_b.val[3], q_b.val[0]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 23);
-			q_a.val[2] = veorq_u32(q_tmp.val[1], q_a.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 9);
-			q_tmp.val[3] = vaddq_u32(q_c.val[3], q_c.val[0]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 23);
-			q_b.val[2] = veorq_u32(q_tmp.val[1], q_b.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 9);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 23);
-			q_c.val[2] = veorq_u32(q_tmp.val[1], q_c.val[2]);
-
-			q_tmp.val[0] = vaddq_u32(q_a.val[2], q_a.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 13);
-			q_tmp.val[2] = vaddq_u32(q_b.val[2], q_b.val[3]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 19);
-			q_a.val[1] = veorq_u32(q_tmp.val[1], q_a.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 13);
-			q_tmp.val[3] = vaddq_u32(q_c.val[2], q_c.val[3]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 19);
-			q_b.val[1] = veorq_u32(q_tmp.val[1], q_b.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 13);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 19);
-			q_c.val[1] = veorq_u32(q_tmp.val[1], q_c.val[1]);
-			
-			q_a.val[3] = vextq_u32(q_a.val[3], q_a.val[3], 3);
-			q_b.val[3] = vextq_u32(q_b.val[3], q_b.val[3], 3);
-			q_c.val[3] = vextq_u32(q_c.val[3], q_c.val[3], 3);
-			
-			q_tmp.val[0] = vaddq_u32(q_a.val[1], q_a.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 18);
-			q_tmp.val[2] = vaddq_u32(q_b.val[1], q_b.val[2]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 14);
-			q_a.val[0] = veorq_u32(q_tmp.val[1], q_a.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 18);
-			q_tmp.val[3] = vaddq_u32(q_c.val[1], q_c.val[2]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 14);
-			q_b.val[0] = veorq_u32(q_tmp.val[1], q_b.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 18);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 14);
-			q_c.val[0] = veorq_u32(q_tmp.val[1], q_c.val[0]);
-			
-			q_a.val[2] = vextq_u32(q_a.val[2], q_a.val[2], 2);
-			q_b.val[2] = vextq_u32(q_b.val[2], q_b.val[2], 2);
-			q_c.val[2] = vextq_u32(q_c.val[2], q_c.val[2], 2);
-			
-			q_tmp.val[0] = vaddq_u32(q_a.val[0], q_a.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 7);
-			q_tmp.val[2] = vaddq_u32(q_b.val[0], q_b.val[3]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 25);
-			q_a.val[1] = vextq_u32(q_a.val[1], q_a.val[1], 1);
-			q_a.val[1] = veorq_u32(q_tmp.val[1], q_a.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 7);
-			q_tmp.val[3] = vaddq_u32(q_c.val[0], q_c.val[3]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 25);
-			q_b.val[1] = vextq_u32(q_b.val[1], q_b.val[1], 1);
-			q_b.val[1] = veorq_u32(q_tmp.val[1], q_b.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 7);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 25);
-			q_c.val[1] = vextq_u32(q_c.val[1], q_c.val[1], 1);
-			q_c.val[1] = veorq_u32(q_tmp.val[1], q_c.val[1]);
-
-			q_tmp.val[0] = vaddq_u32(q_a.val[1], q_a.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 9);
-			q_tmp.val[2] = vaddq_u32(q_b.val[1], q_b.val[0]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 23);
-			q_a.val[2] = veorq_u32(q_tmp.val[1], q_a.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 9);
-			q_tmp.val[3] = vaddq_u32(q_c.val[1], q_c.val[0]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 23);
-			q_b.val[2] = veorq_u32(q_tmp.val[1], q_b.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 9);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 23);
-			q_c.val[2] = veorq_u32(q_tmp.val[1], q_c.val[2]);
-
-			q_tmp.val[0] = vaddq_u32(q_a.val[2], q_a.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 13);
-			q_tmp.val[2] = vaddq_u32(q_b.val[2], q_b.val[1]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 19);
-			q_a.val[3] = veorq_u32(q_tmp.val[1], q_a.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 13);
-			q_tmp.val[3] = vaddq_u32(q_c.val[2], q_c.val[1]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 19);
-			q_b.val[3] = veorq_u32(q_tmp.val[1], q_b.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 13);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 19);
-			q_c.val[3] = veorq_u32(q_tmp.val[1], q_c.val[3]);
-			q_a.val[1] = vextq_u32(q_a.val[1], q_a.val[1], 3);
-			q_b.val[1] = vextq_u32(q_b.val[1], q_b.val[1], 3);
-			q_c.val[1] = vextq_u32(q_c.val[1], q_c.val[1], 3);
-
-			q_tmp.val[0] = vaddq_u32(q_a.val[3], q_a.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 18);
-			q_tmp.val[2] = vaddq_u32(q_b.val[3], q_b.val[2]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 14);
-			q_a.val[2] = vextq_u32(q_a.val[2], q_a.val[2], 2);
-			q_b.val[2] = vextq_u32(q_b.val[2], q_b.val[2], 2);
-			q_a.val[0] = veorq_u32(q_tmp.val[1], q_a.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 18);
-			q_tmp.val[3] = vaddq_u32(q_c.val[3], q_c.val[2]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 14);
-			q_c.val[2] = vextq_u32(q_c.val[2], q_c.val[2], 2);
-			q_b.val[3] = vextq_u32(q_b.val[3], q_b.val[3], 1);
-			q_b.val[0] = veorq_u32(q_tmp.val[1], q_b.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 18);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 14);
-			q_a.val[3] = vextq_u32(q_a.val[3], q_a.val[3], 1);
-			q_c.val[3] = vextq_u32(q_c.val[3], q_c.val[3], 1);
-			q_c.val[0] = veorq_u32(q_tmp.val[1], q_c.val[0]);
-		}
-		ba_a.val[0] = vaddq_u32(ba_a.val[0], q_a.val[0]);
-		ba_a.val[1] = vaddq_u32(ba_a.val[1], q_a.val[1]);
-		ba_a.val[2] = vaddq_u32(ba_a.val[2], q_a.val[2]);
-		ba_a.val[3] = vaddq_u32(ba_a.val[3], q_a.val[3]);
-
-		q_a = ba_a;
-
-		bb_a.val[0] = vaddq_u32(bb_a.val[0], q_b.val[0]);
-		bb_a.val[1] = vaddq_u32(bb_a.val[1], q_b.val[1]);
-		bb_a.val[2] = vaddq_u32(bb_a.val[2], q_b.val[2]);
-		bb_a.val[3] = vaddq_u32(bb_a.val[3], q_b.val[3]);
-		q_b = bb_a;
-
-		bc_a.val[0] = vaddq_u32(bc_a.val[0], q_c.val[0]);
-		bc_a.val[1] = vaddq_u32(bc_a.val[1], q_c.val[1]);
-		bc_a.val[2] = vaddq_u32(bc_a.val[2], q_c.val[2]);
-		bc_a.val[3] = vaddq_u32(bc_a.val[3], q_c.val[3]);
-		q_c = bc_a;
-
-		// loop 2 b
-
-		q_a.val[0] = veorq_u32(ba_b.val[0], q_a.val[0]);
-		q_a.val[1] = veorq_u32(ba_b.val[1], q_a.val[1]);
-		q_a.val[2] = veorq_u32(ba_b.val[2], q_a.val[2]);
-		q_a.val[3] = veorq_u32(ba_b.val[3], q_a.val[3]);
-		ba_b = q_a;
-
-		q_b.val[0] = veorq_u32(bb_b.val[0], q_b.val[0]);
-		q_b.val[1] = veorq_u32(bb_b.val[1], q_b.val[1]);
-		q_b.val[2] = veorq_u32(bb_b.val[2], q_b.val[2]);
-		q_b.val[3] = veorq_u32(bb_b.val[3], q_b.val[3]);
-		bb_b = q_b;
-
-		q_c.val[0] = veorq_u32(bc_b.val[0], q_c.val[0]);
-		q_c.val[1] = veorq_u32(bc_b.val[1], q_c.val[1]);
-		q_c.val[2] = veorq_u32(bc_b.val[2], q_c.val[2]);
-		q_c.val[3] = veorq_u32(bc_b.val[3], q_c.val[3]);
-		bc_b = q_c;
-
-
-		for (int i = 0; i < 3; i++)
-		{
-			q_tmp.val[0] = vaddq_u32(q_a.val[0], q_a.val[1]);  	
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 7);	
-			q_tmp.val[2] = vaddq_u32(q_b.val[0], q_b.val[1]);  	
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 25);
-			q_a.val[3] = veorq_u32(q_tmp.val[1], q_a.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 7);
-			q_tmp.val[3] = vaddq_u32(q_c.val[0], q_c.val[1]); 
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 25);
-			q_b.val[3] = veorq_u32(q_tmp.val[1], q_b.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 7); 				
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 25);				
-			q_c.val[3] = veorq_u32(q_tmp.val[1], q_c.val[3]);
-
-			q_tmp.val[0] = vaddq_u32(q_a.val[3], q_a.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 9);
-			q_tmp.val[2] = vaddq_u32(q_b.val[3], q_b.val[0]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 23);
-			q_a.val[2] = veorq_u32(q_tmp.val[1], q_a.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 9);
-			q_tmp.val[3] = vaddq_u32(q_c.val[3], q_c.val[0]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 23);
-			q_b.val[2] = veorq_u32(q_tmp.val[1], q_b.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 9);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 23);
-			q_c.val[2] = veorq_u32(q_tmp.val[1], q_c.val[2]);
-
-			q_tmp.val[0] = vaddq_u32(q_a.val[2], q_a.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 13);
-			q_tmp.val[2] = vaddq_u32(q_b.val[2], q_b.val[3]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 19);
-			q_a.val[1] = veorq_u32(q_tmp.val[1], q_a.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 13);
-			q_tmp.val[3] = vaddq_u32(q_c.val[2], q_c.val[3]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 19);
-			q_b.val[1] = veorq_u32(q_tmp.val[1], q_b.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 13);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 19);
-			q_c.val[1] = veorq_u32(q_tmp.val[1], q_c.val[1]);
-			
-			q_a.val[3] = vextq_u32(q_a.val[3], q_a.val[3], 3);
-			q_b.val[3] = vextq_u32(q_b.val[3], q_b.val[3], 3);
-			q_c.val[3] = vextq_u32(q_c.val[3], q_c.val[3], 3);
-			
-			q_tmp.val[0] = vaddq_u32(q_a.val[1], q_a.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 18);
-			q_tmp.val[2] = vaddq_u32(q_b.val[1], q_b.val[2]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 14);
-			q_a.val[0] = veorq_u32(q_tmp.val[1], q_a.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 18);
-			q_tmp.val[3] = vaddq_u32(q_c.val[1], q_c.val[2]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 14);
-			q_b.val[0] = veorq_u32(q_tmp.val[1], q_b.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 18);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 14);
-			q_c.val[0] = veorq_u32(q_tmp.val[1], q_c.val[0]);
-			
-			q_a.val[2] = vextq_u32(q_a.val[2], q_a.val[2], 2);
-			q_b.val[2] = vextq_u32(q_b.val[2], q_b.val[2], 2);
-			q_c.val[2] = vextq_u32(q_c.val[2], q_c.val[2], 2);
-			
-			q_tmp.val[0] = vaddq_u32(q_a.val[0], q_a.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 7);
-			q_tmp.val[2] = vaddq_u32(q_b.val[0], q_b.val[3]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 25);
-			q_a.val[1] = vextq_u32(q_a.val[1], q_a.val[1], 1);
-			q_a.val[1] = veorq_u32(q_tmp.val[1], q_a.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 7);
-			q_tmp.val[3] = vaddq_u32(q_c.val[0], q_c.val[3]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 25);
-			q_b.val[1] = vextq_u32(q_b.val[1], q_b.val[1], 1);
-			q_b.val[1] = veorq_u32(q_tmp.val[1], q_b.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 7);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 25);
-			q_c.val[1] = vextq_u32(q_c.val[1], q_c.val[1], 1);
-			q_c.val[1] = veorq_u32(q_tmp.val[1], q_c.val[1]);
-
-			q_tmp.val[0] = vaddq_u32(q_a.val[1], q_a.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 9);
-			q_tmp.val[2] = vaddq_u32(q_b.val[1], q_b.val[0]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 23);
-			q_a.val[2] = veorq_u32(q_tmp.val[1], q_a.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 9);
-			q_tmp.val[3] = vaddq_u32(q_c.val[1], q_c.val[0]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 23);
-			q_b.val[2] = veorq_u32(q_tmp.val[1], q_b.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 9);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 23);
-			q_c.val[2] = veorq_u32(q_tmp.val[1], q_c.val[2]);
-
-			q_tmp.val[0] = vaddq_u32(q_a.val[2], q_a.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 13);
-			q_tmp.val[2] = vaddq_u32(q_b.val[2], q_b.val[1]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 19);
-			q_a.val[3] = veorq_u32(q_tmp.val[1], q_a.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 13);
-			q_tmp.val[3] = vaddq_u32(q_c.val[2], q_c.val[1]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 19);
-			q_b.val[3] = veorq_u32(q_tmp.val[1], q_b.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 13);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 19);
-			q_c.val[3] = veorq_u32(q_tmp.val[1], q_c.val[3]);
-			q_a.val[1] = vextq_u32(q_a.val[1], q_a.val[1], 3);
-			q_b.val[1] = vextq_u32(q_b.val[1], q_b.val[1], 3);
-			q_c.val[1] = vextq_u32(q_c.val[1], q_c.val[1], 3);
-
-			q_tmp.val[0] = vaddq_u32(q_a.val[3], q_a.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 18);
-			q_tmp.val[2] = vaddq_u32(q_b.val[3], q_b.val[2]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 14);
-			q_a.val[2] = vextq_u32(q_a.val[2], q_a.val[2], 2);
-			q_b.val[2] = vextq_u32(q_b.val[2], q_b.val[2], 2);
-			q_a.val[0] = veorq_u32(q_tmp.val[1], q_a.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 18);
-			q_tmp.val[3] = vaddq_u32(q_c.val[3], q_c.val[2]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 14);
-			q_c.val[2] = vextq_u32(q_c.val[2], q_c.val[2], 2);
-			q_b.val[3] = vextq_u32(q_b.val[3], q_b.val[3], 1);
-			q_b.val[0] = veorq_u32(q_tmp.val[1], q_b.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 18);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 14);
-			q_a.val[3] = vextq_u32(q_a.val[3], q_a.val[3], 1);
-			q_c.val[3] = vextq_u32(q_c.val[3], q_c.val[3], 1);
-			q_c.val[0] = veorq_u32(q_tmp.val[1], q_c.val[0]);
-		}
-		{
-			//1
-			q_tmp.val[0] = vaddq_u32(q_a.val[0], q_a.val[1]);  	
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 7);	
-			q_tmp.val[2] = vaddq_u32(q_b.val[0], q_b.val[1]);  	
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 25);
-			q_a.val[3] = veorq_u32(q_tmp.val[1], q_a.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 7);
-			q_tmp.val[3] = vaddq_u32(q_c.val[0], q_c.val[1]); 
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 25);
-			q_b.val[3] = veorq_u32(q_tmp.val[1], q_b.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 7); 				
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 25);				
-			q_c.val[3] = veorq_u32(q_tmp.val[1], q_c.val[3]);
-			//2
-			q_tmp.val[0] = vaddq_u32(q_a.val[3], q_a.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 9);
-			q_tmp.val[2] = vaddq_u32(q_b.val[3], q_b.val[0]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 23);
-			q_a.val[2] = veorq_u32(q_tmp.val[1], q_a.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 9);
-			q_tmp.val[3] = vaddq_u32(q_c.val[3], q_c.val[0]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 23);
-			q_b.val[2] = veorq_u32(q_tmp.val[1], q_b.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 9);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 23);
-			q_c.val[2] = veorq_u32(q_tmp.val[1], q_c.val[2]);
-			//3
-			q_tmp.val[0] = vaddq_u32(q_a.val[2], q_a.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 13);
-			q_tmp.val[2] = vaddq_u32(q_b.val[2], q_b.val[3]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 19);
-			q_a.val[1] = veorq_u32(q_tmp.val[1], q_a.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 13);
-			q_tmp.val[3] = vaddq_u32(q_c.val[2], q_c.val[3]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 19);
-			q_b.val[1] = veorq_u32(q_tmp.val[1], q_b.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 13);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 19);
-			q_c.val[1] = veorq_u32(q_tmp.val[1], q_c.val[1]);
-			
-			q_a.val[3] = vextq_u32(q_a.val[3], q_a.val[3], 3);
-			q_b.val[3] = vextq_u32(q_b.val[3], q_b.val[3], 3);
-			q_c.val[3] = vextq_u32(q_c.val[3], q_c.val[3], 3);
-			//4
-			q_tmp.val[0] = vaddq_u32(q_a.val[1], q_a.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 18);
-			q_tmp.val[2] = vaddq_u32(q_b.val[1], q_b.val[2]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 14);
-			q_a.val[0] = veorq_u32(q_tmp.val[1], q_a.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 18);
-			q_tmp.val[3] = vaddq_u32(q_c.val[1], q_c.val[2]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 14);
-			q_b.val[0] = veorq_u32(q_tmp.val[1], q_b.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 18);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 14);
-			q_c.val[0] = veorq_u32(q_tmp.val[1], q_c.val[0]);
-			
-			q_a.val[2] = vextq_u32(q_a.val[2], q_a.val[2], 2);
-			q_b.val[2] = vextq_u32(q_b.val[2], q_b.val[2], 2);
-			q_c.val[2] = vextq_u32(q_c.val[2], q_c.val[2], 2);
-			//5
-			q_tmp.val[0] = vaddq_u32(q_a.val[0], q_a.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 7);
-			q_tmp.val[2] = vaddq_u32(q_b.val[0], q_b.val[3]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 25);
-			q_a.val[1] = vextq_u32(q_a.val[1], q_a.val[1], 1);
-			q_a.val[1] = veorq_u32(q_tmp.val[1], q_a.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 7);
-			q_tmp.val[3] = vaddq_u32(q_c.val[0], q_c.val[3]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 25);
-			q_b.val[1] = vextq_u32(q_b.val[1], q_b.val[1], 1);
-			q_b.val[1] = veorq_u32(q_tmp.val[1], q_b.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 7);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 25);
-			q_c.val[1] = vextq_u32(q_c.val[1], q_c.val[1], 1);
-			q_c.val[1] = veorq_u32(q_tmp.val[1], q_c.val[1]);
-			//6
-			q_tmp.val[0] = vaddq_u32(q_a.val[1], q_a.val[0]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 9);
-			q_tmp.val[2] = vaddq_u32(q_b.val[1], q_b.val[0]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 23);
-			q_a.val[2] = veorq_u32(q_tmp.val[1], q_a.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 9);
-			q_tmp.val[3] = vaddq_u32(q_c.val[1], q_c.val[0]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 23);
-			q_b.val[2] = veorq_u32(q_tmp.val[1], q_b.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 9);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 23);
-			q_c.val[2] = veorq_u32(q_tmp.val[1], q_c.val[2]);
-			//7
-			q_tmp.val[0] = vaddq_u32(q_a.val[2], q_a.val[1]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 13);
-			q_tmp.val[2] = vaddq_u32(q_b.val[2], q_b.val[1]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 19);
-			q_a.val[3] = veorq_u32(q_tmp.val[1], q_a.val[3]);
-				q_a.val[1] = vextq_u32(q_a.val[1], q_a.val[1], 3);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 13);
-			q_tmp.val[3] = vaddq_u32(q_c.val[2], q_c.val[1]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 19);
-			q_b.val[3] = veorq_u32(q_tmp.val[1], q_b.val[3]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 13);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 19);
-			q_c.val[3] = veorq_u32(q_tmp.val[1], q_c.val[3]);
-			q_b.val[1] = vextq_u32(q_b.val[1], q_b.val[1], 3);
-			q_c.val[1] = vextq_u32(q_c.val[1], q_c.val[1], 3);
-
-			//8
-			q_tmp.val[0] = vaddq_u32(q_a.val[3], q_a.val[2]);
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[0], 18);
-			q_tmp.val[2] = vaddq_u32(q_b.val[3], q_b.val[2]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[0], 14);
-			q_a.val[0] = veorq_u32(q_tmp.val[1], q_a.val[0]);
-				ba_b.val[0] = vaddq_u32(q_a.val[0], ba_b.val[0]);
-					one =	32 * (3 * (ba_b.val[0][0] & (N - 1)) + 0);
-					__builtin_prefetch(&W[one + 0]);
-					__builtin_prefetch(&W[one + 8]);
-					__builtin_prefetch(&W[one + 16]);
-					__builtin_prefetch(&W[one + 24]);
-			
-			q_a.val[2] = vextq_u32(q_a.val[2], q_a.val[2], 2);
-			q_b.val[2] = vextq_u32(q_b.val[2], q_b.val[2], 2);
-			
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[2], 18);
-			q_tmp.val[3] = vaddq_u32(q_c.val[3], q_c.val[2]);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[2], 14);
-			q_c.val[2] = vextq_u32(q_c.val[2], q_c.val[2], 2);
-			q_b.val[3] = vextq_u32(q_b.val[3], q_b.val[3], 1);
-			q_b.val[0] = veorq_u32(q_tmp.val[1], q_b.val[0]);
-				bb_b.val[0] = vaddq_u32(q_b.val[0], bb_b.val[0]);
-					two =	32 * (3 * (bb_b.val[0][0] & (N - 1)) + 1);
-					__builtin_prefetch(&W[two + 0]);
-					__builtin_prefetch(&W[two + 8]);
-					__builtin_prefetch(&W[two + 16]);
-					__builtin_prefetch(&W[two + 24]);
-
-			q_tmp.val[1] = vshlq_n_u32(q_tmp.val[3], 18);
-			q_tmp.val[1] = vsriq_n_u32(q_tmp.val[1], q_tmp.val[3], 14);
-			q_a.val[3] = vextq_u32(q_a.val[3], q_a.val[3], 1);
-			q_c.val[3] = vextq_u32(q_c.val[3], q_c.val[3], 1);
-			q_c.val[0] = veorq_u32(q_tmp.val[1], q_c.val[0]);
-				bc_b.val[0] = vaddq_u32(q_c.val[0], bc_b.val[0]);
-					three = 32 * (3 * (bc_b.val[0][0] & (N - 1)) + 2);
-					__builtin_prefetch(&W[three + 0]);
-					__builtin_prefetch(&W[three + 8]);
-					__builtin_prefetch(&W[three + 16]);
-					__builtin_prefetch(&W[three + 24]);
-		}
-
-		x.val[0] = vld1q_u32(&W[one +  0]);
-		ba_b.val[1] = vaddq_u32(q_a.val[1], ba_b.val[1]);
-		ba_b.val[2] = vaddq_u32(q_a.val[2], ba_b.val[2]);
-		ba_b.val[3] = vaddq_u32(q_a.val[3], ba_b.val[3]);
-		x.val[1] = vld1q_u32(&W[one +  4]);
-		bb_b.val[1] = vaddq_u32(q_b.val[1], bb_b.val[1]);
-		bb_b.val[2] = vaddq_u32(q_b.val[2], bb_b.val[2]);
-		bb_b.val[3] = vaddq_u32(q_b.val[3], bb_b.val[3]);
-		x.val[2] = vld1q_u32(&W[one +  8]);
-		bc_b.val[1] = vaddq_u32(q_c.val[1], bc_b.val[1]);
-		bc_b.val[2] = vaddq_u32(q_c.val[2], bc_b.val[2]);
-		bc_b.val[3] = vaddq_u32(q_c.val[3], bc_b.val[3]);
-		x.val[3] = vld1q_u32(&W[one + 12]);
-	}
-
-	vst1q_u32(&B[0],       ba_a.val[0]);
-	vst1q_u32(&B[4],       ba_a.val[1]);
-	vst1q_u32(&B[8],       ba_a.val[2]);
-	vst1q_u32(&B[12],      ba_a.val[3]);
-	vst1q_u32(&B[16 + 0],  ba_b.val[0]);
-	vst1q_u32(&B[16 + 4],  ba_b.val[1]);
-	vst1q_u32(&B[16 + 8],  ba_b.val[2]);
-	vst1q_u32(&B[16 + 12], ba_b.val[3]);
-
-	vst1q_u32(&B[32 + 0],  		bb_a.val[0]);
-	vst1q_u32(&B[32 + 4],  		bb_a.val[1]);
-	vst1q_u32(&B[32 + 8],  		bb_a.val[2]);
-	vst1q_u32(&B[32 + 12], 		bb_a.val[3]);
-	vst1q_u32(&B[32 + 16 + 0],  bb_b.val[0]);
-	vst1q_u32(&B[32 + 16 + 4],  bb_b.val[1]);
-	vst1q_u32(&B[32 + 16 + 8],  bb_b.val[2]);
-	vst1q_u32(&B[32 + 16 + 12], bb_b.val[3]);
-
-	vst1q_u32(&B[64 + 0],  		bc_a.val[0]);
-	vst1q_u32(&B[64 + 4],  		bc_a.val[1]);
-	vst1q_u32(&B[64 + 8],  		bc_a.val[2]);
-	vst1q_u32(&B[64 + 12], 		bc_a.val[3]);
-	vst1q_u32(&B[64 + 16 + 0],  bc_b.val[0]);
-	vst1q_u32(&B[64 + 16 + 4],  bc_b.val[1]);
-	vst1q_u32(&B[64 + 16 + 8],  bc_b.val[2]);
-	vst1q_u32(&B[64 + 16 + 12], bc_b.val[3]);
-
-	scrypt_shuffle(&B[0  + 0]);
-	scrypt_shuffle(&B[16 + 0]);
-	scrypt_shuffle(&B[0 + 32]);
-	scrypt_shuffle(&B[16 + 32]);
-	scrypt_shuffle(&B[0 + 64]);
-	scrypt_shuffle(&B[16 + 64]);
-}
 #else
 static inline void xor_salsa8(uint32_t B[16], const uint32_t Bx[16])
 {