diff --git a/src/thirdparty/blake2/blake2-config.h b/src/thirdparty/blake2/blake2-config.h
new file mode 100644
index 0000000..a524aa9
--- /dev/null
+++ b/src/thirdparty/blake2/blake2-config.h
@@ -0,0 +1,72 @@
+/*
+   BLAKE2 reference source code package - optimized C implementations
+
+   Copyright 2012, Samuel Neves <sneves@dei.uc.pt>.  You may use this under the
+   terms of the CC0, the OpenSSL Licence, or the Apache Public License 2.0, at
+   your option.  The terms of these licenses can be found at:
+
+   - CC0 1.0 Universal : http://creativecommons.org/publicdomain/zero/1.0
+   - OpenSSL license   : https://www.openssl.org/source/license.html
+   - Apache 2.0        : http://www.apache.org/licenses/LICENSE-2.0
+
+   More information about the BLAKE2 hash function can be found at
+   https://blake2.net.
+*/
+#ifndef BLAKE2_CONFIG_H
+#define BLAKE2_CONFIG_H
+
+/* These don't work everywhere */
+#if defined(__SSE2__) || defined(__x86_64__) || defined(__amd64__)
+#define HAVE_SSE2
+#endif
+
+#if defined(__SSSE3__)
+#define HAVE_SSSE3
+#endif
+
+#if defined(__SSE4_1__)
+#define HAVE_SSE41
+#endif
+
+#if defined(__AVX__)
+#define HAVE_AVX
+#endif
+
+#if defined(__XOP__)
+#define HAVE_XOP
+#endif
+
+
+#ifdef HAVE_AVX2
+#ifndef HAVE_AVX
+#define HAVE_AVX
+#endif
+#endif
+
+#ifdef HAVE_XOP
+#ifndef HAVE_AVX
+#define HAVE_AVX
+#endif
+#endif
+
+#ifdef HAVE_AVX
+#ifndef HAVE_SSE41
+#define HAVE_SSE41
+#endif
+#endif
+
+#ifdef HAVE_SSE41
+#ifndef HAVE_SSSE3
+#define HAVE_SSSE3
+#endif
+#endif
+
+#ifdef HAVE_SSSE3
+#define HAVE_SSE2
+#endif
+
+#if !defined(HAVE_SSE2)
+#error "This code requires at least SSE2."
+#endif
+
+#endif
diff --git a/src/thirdparty/blake2/blake2-impl.h b/src/thirdparty/blake2/blake2-impl.h
new file mode 100644
index 0000000..c1df82e
--- /dev/null
+++ b/src/thirdparty/blake2/blake2-impl.h
@@ -0,0 +1,160 @@
+/*
+   BLAKE2 reference source code package - reference C implementations
+
+   Copyright 2012, Samuel Neves <sneves@dei.uc.pt>.  You may use this under the
+   terms of the CC0, the OpenSSL Licence, or the Apache Public License 2.0, at
+   your option.  The terms of these licenses can be found at:
+
+   - CC0 1.0 Universal : http://creativecommons.org/publicdomain/zero/1.0
+   - OpenSSL license   : https://www.openssl.org/source/license.html
+   - Apache 2.0        : http://www.apache.org/licenses/LICENSE-2.0
+
+   More information about the BLAKE2 hash function can be found at
+   https://blake2.net.
+*/
+#ifndef BLAKE2_IMPL_H
+#define BLAKE2_IMPL_H
+
+#include <stdint.h>
+#include <string.h>
+
+#if !defined(__cplusplus) && (!defined(__STDC_VERSION__) || __STDC_VERSION__ < 199901L)
+  #if   defined(_MSC_VER)
+    #define BLAKE2_INLINE __inline
+  #elif defined(__GNUC__)
+    #define BLAKE2_INLINE __inline__
+  #else
+    #define BLAKE2_INLINE
+  #endif
+#else
+  #define BLAKE2_INLINE inline
+#endif
+
+static BLAKE2_INLINE uint32_t load32( const void *src )
+{
+#if defined(NATIVE_LITTLE_ENDIAN)
+  uint32_t w;
+  memcpy(&w, src, sizeof w);
+  return w;
+#else
+  const uint8_t *p = ( const uint8_t * )src;
+  return (( uint32_t )( p[0] ) <<  0) |
+         (( uint32_t )( p[1] ) <<  8) |
+         (( uint32_t )( p[2] ) << 16) |
+         (( uint32_t )( p[3] ) << 24) ;
+#endif
+}
+
+static BLAKE2_INLINE uint64_t load64( const void *src )
+{
+#if defined(NATIVE_LITTLE_ENDIAN)
+  uint64_t w;
+  memcpy(&w, src, sizeof w);
+  return w;
+#else
+  const uint8_t *p = ( const uint8_t * )src;
+  return (( uint64_t )( p[0] ) <<  0) |
+         (( uint64_t )( p[1] ) <<  8) |
+         (( uint64_t )( p[2] ) << 16) |
+         (( uint64_t )( p[3] ) << 24) |
+         (( uint64_t )( p[4] ) << 32) |
+         (( uint64_t )( p[5] ) << 40) |
+         (( uint64_t )( p[6] ) << 48) |
+         (( uint64_t )( p[7] ) << 56) ;
+#endif
+}
+
+static BLAKE2_INLINE uint16_t load16( const void *src )
+{
+#if defined(NATIVE_LITTLE_ENDIAN)
+  uint16_t w;
+  memcpy(&w, src, sizeof w);
+  return w;
+#else
+  const uint8_t *p = ( const uint8_t * )src;
+  return ( uint16_t )((( uint32_t )( p[0] ) <<  0) |
+                      (( uint32_t )( p[1] ) <<  8));
+#endif
+}
+
+static BLAKE2_INLINE void store16( void *dst, uint16_t w )
+{
+#if defined(NATIVE_LITTLE_ENDIAN)
+  memcpy(dst, &w, sizeof w);
+#else
+  uint8_t *p = ( uint8_t * )dst;
+  *p++ = ( uint8_t )w; w >>= 8;
+  *p++ = ( uint8_t )w;
+#endif
+}
+
+static BLAKE2_INLINE void store32( void *dst, uint32_t w )
+{
+#if defined(NATIVE_LITTLE_ENDIAN)
+  memcpy(dst, &w, sizeof w);
+#else
+  uint8_t *p = ( uint8_t * )dst;
+  p[0] = (uint8_t)(w >>  0);
+  p[1] = (uint8_t)(w >>  8);
+  p[2] = (uint8_t)(w >> 16);
+  p[3] = (uint8_t)(w >> 24);
+#endif
+}
+
+static BLAKE2_INLINE void store64( void *dst, uint64_t w )
+{
+#if defined(NATIVE_LITTLE_ENDIAN)
+  memcpy(dst, &w, sizeof w);
+#else
+  uint8_t *p = ( uint8_t * )dst;
+  p[0] = (uint8_t)(w >>  0);
+  p[1] = (uint8_t)(w >>  8);
+  p[2] = (uint8_t)(w >> 16);
+  p[3] = (uint8_t)(w >> 24);
+  p[4] = (uint8_t)(w >> 32);
+  p[5] = (uint8_t)(w >> 40);
+  p[6] = (uint8_t)(w >> 48);
+  p[7] = (uint8_t)(w >> 56);
+#endif
+}
+
+static BLAKE2_INLINE uint64_t load48( const void *src )
+{
+  const uint8_t *p = ( const uint8_t * )src;
+  return (( uint64_t )( p[0] ) <<  0) |
+         (( uint64_t )( p[1] ) <<  8) |
+         (( uint64_t )( p[2] ) << 16) |
+         (( uint64_t )( p[3] ) << 24) |
+         (( uint64_t )( p[4] ) << 32) |
+         (( uint64_t )( p[5] ) << 40) ;
+}
+
+static BLAKE2_INLINE void store48( void *dst, uint64_t w )
+{
+  uint8_t *p = ( uint8_t * )dst;
+  p[0] = (uint8_t)(w >>  0);
+  p[1] = (uint8_t)(w >>  8);
+  p[2] = (uint8_t)(w >> 16);
+  p[3] = (uint8_t)(w >> 24);
+  p[4] = (uint8_t)(w >> 32);
+  p[5] = (uint8_t)(w >> 40);
+}
+
+static BLAKE2_INLINE uint32_t rotr32( const uint32_t w, const unsigned c )
+{
+  return ( w >> c ) | ( w << ( 32 - c ) );
+}
+
+static BLAKE2_INLINE uint64_t rotr64( const uint64_t w, const unsigned c )
+{
+  return ( w >> c ) | ( w << ( 64 - c ) );
+}
+
+/* prevents compiler optimizing out memset() */
+static BLAKE2_INLINE void secure_zero_memory(void *v, size_t n)
+{
+  static void *(*const volatile memset_v)(void *, int, size_t) = &memset;
+  memset_v(v, 0, n);
+}
+
+#endif
diff --git a/src/thirdparty/blake2/blake2.h b/src/thirdparty/blake2/blake2.h
new file mode 100644
index 0000000..ca39030
--- /dev/null
+++ b/src/thirdparty/blake2/blake2.h
@@ -0,0 +1,195 @@
+/*
+   BLAKE2 reference source code package - reference C implementations
+
+   Copyright 2012, Samuel Neves <sneves@dei.uc.pt>.  You may use this under the
+   terms of the CC0, the OpenSSL Licence, or the Apache Public License 2.0, at
+   your option.  The terms of these licenses can be found at:
+
+   - CC0 1.0 Universal : http://creativecommons.org/publicdomain/zero/1.0
+   - OpenSSL license   : https://www.openssl.org/source/license.html
+   - Apache 2.0        : http://www.apache.org/licenses/LICENSE-2.0
+
+   More information about the BLAKE2 hash function can be found at
+   https://blake2.net.
+*/
+#ifndef BLAKE2_H
+#define BLAKE2_H
+
+#include <stddef.h>
+#include <stdint.h>
+
+#if defined(_MSC_VER)
+#define BLAKE2_PACKED(x) __pragma(pack(push, 1)) x __pragma(pack(pop))
+#else
+#define BLAKE2_PACKED(x) x __attribute__((packed))
+#endif
+
+#if defined(__cplusplus)
+extern "C" {
+#endif
+
+  enum blake2s_constant
+  {
+    BLAKE2S_BLOCKBYTES = 64,
+    BLAKE2S_OUTBYTES   = 32,
+    BLAKE2S_KEYBYTES   = 32,
+    BLAKE2S_SALTBYTES  = 8,
+    BLAKE2S_PERSONALBYTES = 8
+  };
+
+  enum blake2b_constant
+  {
+    BLAKE2B_BLOCKBYTES = 128,
+    BLAKE2B_OUTBYTES   = 64,
+    BLAKE2B_KEYBYTES   = 64,
+    BLAKE2B_SALTBYTES  = 16,
+    BLAKE2B_PERSONALBYTES = 16
+  };
+
+  typedef struct blake2s_state__
+  {
+    uint32_t h[8];
+    uint32_t t[2];
+    uint32_t f[2];
+    uint8_t  buf[BLAKE2S_BLOCKBYTES];
+    size_t   buflen;
+    size_t   outlen;
+    uint8_t  last_node;
+  } blake2s_state;
+
+  typedef struct blake2b_state__
+  {
+    uint64_t h[8];
+    uint64_t t[2];
+    uint64_t f[2];
+    uint8_t  buf[BLAKE2B_BLOCKBYTES];
+    size_t   buflen;
+    size_t   outlen;
+    uint8_t  last_node;
+  } blake2b_state;
+
+  typedef struct blake2sp_state__
+  {
+    blake2s_state S[8][1];
+    blake2s_state R[1];
+    uint8_t       buf[8 * BLAKE2S_BLOCKBYTES];
+    size_t        buflen;
+    size_t        outlen;
+  } blake2sp_state;
+
+  typedef struct blake2bp_state__
+  {
+    blake2b_state S[4][1];
+    blake2b_state R[1];
+    uint8_t       buf[4 * BLAKE2B_BLOCKBYTES];
+    size_t        buflen;
+    size_t        outlen;
+  } blake2bp_state;
+
+
+  BLAKE2_PACKED(struct blake2s_param__
+  {
+    uint8_t  digest_length; /* 1 */
+    uint8_t  key_length;    /* 2 */
+    uint8_t  fanout;        /* 3 */
+    uint8_t  depth;         /* 4 */
+    uint32_t leaf_length;   /* 8 */
+    uint32_t node_offset;  /* 12 */
+    uint16_t xof_length;    /* 14 */
+    uint8_t  node_depth;    /* 15 */
+    uint8_t  inner_length;  /* 16 */
+    /* uint8_t  reserved[0]; */
+    uint8_t  salt[BLAKE2S_SALTBYTES]; /* 24 */
+    uint8_t  personal[BLAKE2S_PERSONALBYTES];  /* 32 */
+  });
+
+  typedef struct blake2s_param__ blake2s_param;
+
+  BLAKE2_PACKED(struct blake2b_param__
+  {
+    uint8_t  digest_length; /* 1 */
+    uint8_t  key_length;    /* 2 */
+    uint8_t  fanout;        /* 3 */
+    uint8_t  depth;         /* 4 */
+    uint32_t leaf_length;   /* 8 */
+    uint32_t node_offset;   /* 12 */
+    uint32_t xof_length;    /* 16 */
+    uint8_t  node_depth;    /* 17 */
+    uint8_t  inner_length;  /* 18 */
+    uint8_t  reserved[14];  /* 32 */
+    uint8_t  salt[BLAKE2B_SALTBYTES]; /* 48 */
+    uint8_t  personal[BLAKE2B_PERSONALBYTES];  /* 64 */
+  });
+
+  typedef struct blake2b_param__ blake2b_param;
+
+  typedef struct blake2xs_state__
+  {
+    blake2s_state S[1];
+    blake2s_param P[1];
+  } blake2xs_state;
+
+  typedef struct blake2xb_state__
+  {
+    blake2b_state S[1];
+    blake2b_param P[1];
+  } blake2xb_state;
+
+  /* Padded structs result in a compile-time error */
+  enum {
+    BLAKE2_DUMMY_1 = 1/(int)(sizeof(blake2s_param) == BLAKE2S_OUTBYTES),
+    BLAKE2_DUMMY_2 = 1/(int)(sizeof(blake2b_param) == BLAKE2B_OUTBYTES)
+  };
+
+  /* Streaming API */
+  int blake2s_init( blake2s_state *S, size_t outlen );
+  int blake2s_init_key( blake2s_state *S, size_t outlen, const void *key, size_t keylen );
+  int blake2s_init_param( blake2s_state *S, const blake2s_param *P );
+  int blake2s_update( blake2s_state *S, const void *in, size_t inlen );
+  int blake2s_final( blake2s_state *S, void *out, size_t outlen );
+
+  int blake2b_init( blake2b_state *S, size_t outlen );
+  int blake2b_init_key( blake2b_state *S, size_t outlen, const void *key, size_t keylen );
+  int blake2b_init_param( blake2b_state *S, const blake2b_param *P );
+  int blake2b_update( blake2b_state *S, const void *in, size_t inlen );
+  int blake2b_final( blake2b_state *S, void *out, size_t outlen );
+
+  int blake2sp_init( blake2sp_state *S, size_t outlen );
+  int blake2sp_init_key( blake2sp_state *S, size_t outlen, const void *key, size_t keylen );
+  int blake2sp_update( blake2sp_state *S, const void *in, size_t inlen );
+  int blake2sp_final( blake2sp_state *S, void *out, size_t outlen );
+
+  int blake2bp_init( blake2bp_state *S, size_t outlen );
+  int blake2bp_init_key( blake2bp_state *S, size_t outlen, const void *key, size_t keylen );
+  int blake2bp_update( blake2bp_state *S, const void *in, size_t inlen );
+  int blake2bp_final( blake2bp_state *S, void *out, size_t outlen );
+
+  /* Variable output length API */
+  int blake2xs_init( blake2xs_state *S, const size_t outlen );
+  int blake2xs_init_key( blake2xs_state *S, const size_t outlen, const void *key, size_t keylen );
+  int blake2xs_update( blake2xs_state *S, const void *in, size_t inlen );
+  int blake2xs_final(blake2xs_state *S, void *out, size_t outlen);
+
+  int blake2xb_init( blake2xb_state *S, const size_t outlen );
+  int blake2xb_init_key( blake2xb_state *S, const size_t outlen, const void *key, size_t keylen );
+  int blake2xb_update( blake2xb_state *S, const void *in, size_t inlen );
+  int blake2xb_final(blake2xb_state *S, void *out, size_t outlen);
+
+  /* Simple API */
+  int blake2s( void *out, size_t outlen, const void *in, size_t inlen, const void *key, size_t keylen );
+  int blake2b( void *out, size_t outlen, const void *in, size_t inlen, const void *key, size_t keylen );
+
+  int blake2sp( void *out, size_t outlen, const void *in, size_t inlen, const void *key, size_t keylen );
+  int blake2bp( void *out, size_t outlen, const void *in, size_t inlen, const void *key, size_t keylen );
+
+  int blake2xs( void *out, size_t outlen, const void *in, size_t inlen, const void *key, size_t keylen );
+  int blake2xb( void *out, size_t outlen, const void *in, size_t inlen, const void *key, size_t keylen );
+
+  /* This is simply an alias for blake2b */
+  int blake2( void *out, size_t outlen, const void *in, size_t inlen, const void *key, size_t keylen );
+
+#if defined(__cplusplus)
+}
+#endif
+
+#endif
diff --git a/src/thirdparty/blake2/blake2b-load-sse2.h b/src/thirdparty/blake2/blake2b-load-sse2.h
new file mode 100644
index 0000000..23a8d40
--- /dev/null
+++ b/src/thirdparty/blake2/blake2b-load-sse2.h
@@ -0,0 +1,68 @@
+/*
+   BLAKE2 reference source code package - optimized C implementations
+
+   Copyright 2012, Samuel Neves <sneves@dei.uc.pt>.  You may use this under the
+   terms of the CC0, the OpenSSL Licence, or the Apache Public License 2.0, at
+   your option.  The terms of these licenses can be found at:
+
+   - CC0 1.0 Universal : http://creativecommons.org/publicdomain/zero/1.0
+   - OpenSSL license   : https://www.openssl.org/source/license.html
+   - Apache 2.0        : http://www.apache.org/licenses/LICENSE-2.0
+
+   More information about the BLAKE2 hash function can be found at
+   https://blake2.net.
+*/
+#ifndef BLAKE2B_LOAD_SSE2_H
+#define BLAKE2B_LOAD_SSE2_H
+
+#define LOAD_MSG_0_1(b0, b1) b0 = _mm_set_epi64x(m2, m0); b1 = _mm_set_epi64x(m6, m4)
+#define LOAD_MSG_0_2(b0, b1) b0 = _mm_set_epi64x(m3, m1); b1 = _mm_set_epi64x(m7, m5)
+#define LOAD_MSG_0_3(b0, b1) b0 = _mm_set_epi64x(m10, m8); b1 = _mm_set_epi64x(m14, m12)
+#define LOAD_MSG_0_4(b0, b1) b0 = _mm_set_epi64x(m11, m9); b1 = _mm_set_epi64x(m15, m13)
+#define LOAD_MSG_1_1(b0, b1) b0 = _mm_set_epi64x(m4, m14); b1 = _mm_set_epi64x(m13, m9)
+#define LOAD_MSG_1_2(b0, b1) b0 = _mm_set_epi64x(m8, m10); b1 = _mm_set_epi64x(m6, m15)
+#define LOAD_MSG_1_3(b0, b1) b0 = _mm_set_epi64x(m0, m1); b1 = _mm_set_epi64x(m5, m11)
+#define LOAD_MSG_1_4(b0, b1) b0 = _mm_set_epi64x(m2, m12); b1 = _mm_set_epi64x(m3, m7)
+#define LOAD_MSG_2_1(b0, b1) b0 = _mm_set_epi64x(m12, m11); b1 = _mm_set_epi64x(m15, m5)
+#define LOAD_MSG_2_2(b0, b1) b0 = _mm_set_epi64x(m0, m8); b1 = _mm_set_epi64x(m13, m2)
+#define LOAD_MSG_2_3(b0, b1) b0 = _mm_set_epi64x(m3, m10); b1 = _mm_set_epi64x(m9, m7)
+#define LOAD_MSG_2_4(b0, b1) b0 = _mm_set_epi64x(m6, m14); b1 = _mm_set_epi64x(m4, m1)
+#define LOAD_MSG_3_1(b0, b1) b0 = _mm_set_epi64x(m3, m7); b1 = _mm_set_epi64x(m11, m13)
+#define LOAD_MSG_3_2(b0, b1) b0 = _mm_set_epi64x(m1, m9); b1 = _mm_set_epi64x(m14, m12)
+#define LOAD_MSG_3_3(b0, b1) b0 = _mm_set_epi64x(m5, m2); b1 = _mm_set_epi64x(m15, m4)
+#define LOAD_MSG_3_4(b0, b1) b0 = _mm_set_epi64x(m10, m6); b1 = _mm_set_epi64x(m8, m0)
+#define LOAD_MSG_4_1(b0, b1) b0 = _mm_set_epi64x(m5, m9); b1 = _mm_set_epi64x(m10, m2)
+#define LOAD_MSG_4_2(b0, b1) b0 = _mm_set_epi64x(m7, m0); b1 = _mm_set_epi64x(m15, m4)
+#define LOAD_MSG_4_3(b0, b1) b0 = _mm_set_epi64x(m11, m14); b1 = _mm_set_epi64x(m3, m6)
+#define LOAD_MSG_4_4(b0, b1) b0 = _mm_set_epi64x(m12, m1); b1 = _mm_set_epi64x(m13, m8)
+#define LOAD_MSG_5_1(b0, b1) b0 = _mm_set_epi64x(m6, m2); b1 = _mm_set_epi64x(m8, m0)
+#define LOAD_MSG_5_2(b0, b1) b0 = _mm_set_epi64x(m10, m12); b1 = _mm_set_epi64x(m3, m11)
+#define LOAD_MSG_5_3(b0, b1) b0 = _mm_set_epi64x(m7, m4); b1 = _mm_set_epi64x(m1, m15)
+#define LOAD_MSG_5_4(b0, b1) b0 = _mm_set_epi64x(m5, m13); b1 = _mm_set_epi64x(m9, m14)
+#define LOAD_MSG_6_1(b0, b1) b0 = _mm_set_epi64x(m1, m12); b1 = _mm_set_epi64x(m4, m14)
+#define LOAD_MSG_6_2(b0, b1) b0 = _mm_set_epi64x(m15, m5); b1 = _mm_set_epi64x(m10, m13)
+#define LOAD_MSG_6_3(b0, b1) b0 = _mm_set_epi64x(m6, m0); b1 = _mm_set_epi64x(m8, m9)
+#define LOAD_MSG_6_4(b0, b1) b0 = _mm_set_epi64x(m3, m7); b1 = _mm_set_epi64x(m11, m2)
+#define LOAD_MSG_7_1(b0, b1) b0 = _mm_set_epi64x(m7, m13); b1 = _mm_set_epi64x(m3, m12)
+#define LOAD_MSG_7_2(b0, b1) b0 = _mm_set_epi64x(m14, m11); b1 = _mm_set_epi64x(m9, m1)
+#define LOAD_MSG_7_3(b0, b1) b0 = _mm_set_epi64x(m15, m5); b1 = _mm_set_epi64x(m2, m8)
+#define LOAD_MSG_7_4(b0, b1) b0 = _mm_set_epi64x(m4, m0); b1 = _mm_set_epi64x(m10, m6)
+#define LOAD_MSG_8_1(b0, b1) b0 = _mm_set_epi64x(m14, m6); b1 = _mm_set_epi64x(m0, m11)
+#define LOAD_MSG_8_2(b0, b1) b0 = _mm_set_epi64x(m9, m15); b1 = _mm_set_epi64x(m8, m3)
+#define LOAD_MSG_8_3(b0, b1) b0 = _mm_set_epi64x(m13, m12); b1 = _mm_set_epi64x(m10, m1)
+#define LOAD_MSG_8_4(b0, b1) b0 = _mm_set_epi64x(m7, m2); b1 = _mm_set_epi64x(m5, m4)
+#define LOAD_MSG_9_1(b0, b1) b0 = _mm_set_epi64x(m8, m10); b1 = _mm_set_epi64x(m1, m7)
+#define LOAD_MSG_9_2(b0, b1) b0 = _mm_set_epi64x(m4, m2); b1 = _mm_set_epi64x(m5, m6)
+#define LOAD_MSG_9_3(b0, b1) b0 = _mm_set_epi64x(m9, m15); b1 = _mm_set_epi64x(m13, m3)
+#define LOAD_MSG_9_4(b0, b1) b0 = _mm_set_epi64x(m14, m11); b1 = _mm_set_epi64x(m0, m12)
+#define LOAD_MSG_10_1(b0, b1) b0 = _mm_set_epi64x(m2, m0); b1 = _mm_set_epi64x(m6, m4)
+#define LOAD_MSG_10_2(b0, b1) b0 = _mm_set_epi64x(m3, m1); b1 = _mm_set_epi64x(m7, m5)
+#define LOAD_MSG_10_3(b0, b1) b0 = _mm_set_epi64x(m10, m8); b1 = _mm_set_epi64x(m14, m12)
+#define LOAD_MSG_10_4(b0, b1) b0 = _mm_set_epi64x(m11, m9); b1 = _mm_set_epi64x(m15, m13)
+#define LOAD_MSG_11_1(b0, b1) b0 = _mm_set_epi64x(m4, m14); b1 = _mm_set_epi64x(m13, m9)
+#define LOAD_MSG_11_2(b0, b1) b0 = _mm_set_epi64x(m8, m10); b1 = _mm_set_epi64x(m6, m15)
+#define LOAD_MSG_11_3(b0, b1) b0 = _mm_set_epi64x(m0, m1); b1 = _mm_set_epi64x(m5, m11)
+#define LOAD_MSG_11_4(b0, b1) b0 = _mm_set_epi64x(m2, m12); b1 = _mm_set_epi64x(m3, m7)
+
+
+#endif
diff --git a/src/thirdparty/blake2/blake2b-load-sse41.h b/src/thirdparty/blake2/blake2b-load-sse41.h
new file mode 100644
index 0000000..0eca865
--- /dev/null
+++ b/src/thirdparty/blake2/blake2b-load-sse41.h
@@ -0,0 +1,402 @@
+/*
+   BLAKE2 reference source code package - optimized C implementations
+
+   Copyright 2012, Samuel Neves <sneves@dei.uc.pt>.  You may use this under the
+   terms of the CC0, the OpenSSL Licence, or the Apache Public License 2.0, at
+   your option.  The terms of these licenses can be found at:
+
+   - CC0 1.0 Universal : http://creativecommons.org/publicdomain/zero/1.0
+   - OpenSSL license   : https://www.openssl.org/source/license.html
+   - Apache 2.0        : http://www.apache.org/licenses/LICENSE-2.0
+
+   More information about the BLAKE2 hash function can be found at
+   https://blake2.net.
+*/
+#ifndef BLAKE2B_LOAD_SSE41_H
+#define BLAKE2B_LOAD_SSE41_H
+
+#define LOAD_MSG_0_1(b0, b1) \
+do \
+{ \
+b0 = _mm_unpacklo_epi64(m0, m1); \
+b1 = _mm_unpacklo_epi64(m2, m3); \
+} while(0)
+
+
+#define LOAD_MSG_0_2(b0, b1) \
+do \
+{ \
+b0 = _mm_unpackhi_epi64(m0, m1); \
+b1 = _mm_unpackhi_epi64(m2, m3); \
+} while(0)
+
+
+#define LOAD_MSG_0_3(b0, b1) \
+do \
+{ \
+b0 = _mm_unpacklo_epi64(m4, m5); \
+b1 = _mm_unpacklo_epi64(m6, m7); \
+} while(0)
+
+
+#define LOAD_MSG_0_4(b0, b1) \
+do \
+{ \
+b0 = _mm_unpackhi_epi64(m4, m5); \
+b1 = _mm_unpackhi_epi64(m6, m7); \
+} while(0)
+
+
+#define LOAD_MSG_1_1(b0, b1) \
+do \
+{ \
+b0 = _mm_unpacklo_epi64(m7, m2); \
+b1 = _mm_unpackhi_epi64(m4, m6); \
+} while(0)
+
+
+#define LOAD_MSG_1_2(b0, b1) \
+do \
+{ \
+b0 = _mm_unpacklo_epi64(m5, m4); \
+b1 = _mm_alignr_epi8(m3, m7, 8); \
+} while(0)
+
+
+#define LOAD_MSG_1_3(b0, b1) \
+do \
+{ \
+b0 = _mm_shuffle_epi32(m0, _MM_SHUFFLE(1,0,3,2)); \
+b1 = _mm_unpackhi_epi64(m5, m2); \
+} while(0)
+
+
+#define LOAD_MSG_1_4(b0, b1) \
+do \
+{ \
+b0 = _mm_unpacklo_epi64(m6, m1); \
+b1 = _mm_unpackhi_epi64(m3, m1); \
+} while(0)
+
+
+#define LOAD_MSG_2_1(b0, b1) \
+do \
+{ \
+b0 = _mm_alignr_epi8(m6, m5, 8); \
+b1 = _mm_unpackhi_epi64(m2, m7); \
+} while(0)
+
+
+#define LOAD_MSG_2_2(b0, b1) \
+do \
+{ \
+b0 = _mm_unpacklo_epi64(m4, m0); \
+b1 = _mm_blend_epi16(m1, m6, 0xF0); \
+} while(0)
+
+
+#define LOAD_MSG_2_3(b0, b1) \
+do \
+{ \
+b0 = _mm_blend_epi16(m5, m1, 0xF0); \
+b1 = _mm_unpackhi_epi64(m3, m4); \
+} while(0)
+
+
+#define LOAD_MSG_2_4(b0, b1) \
+do \
+{ \
+b0 = _mm_unpacklo_epi64(m7, m3); \
+b1 = _mm_alignr_epi8(m2, m0, 8); \
+} while(0)
+
+
+#define LOAD_MSG_3_1(b0, b1) \
+do \
+{ \
+b0 = _mm_unpackhi_epi64(m3, m1); \
+b1 = _mm_unpackhi_epi64(m6, m5); \
+} while(0)
+
+
+#define LOAD_MSG_3_2(b0, b1) \
+do \
+{ \
+b0 = _mm_unpackhi_epi64(m4, m0); \
+b1 = _mm_unpacklo_epi64(m6, m7); \
+} while(0)
+
+
+#define LOAD_MSG_3_3(b0, b1) \
+do \
+{ \
+b0 = _mm_blend_epi16(m1, m2, 0xF0); \
+b1 = _mm_blend_epi16(m2, m7, 0xF0); \
+} while(0)
+
+
+#define LOAD_MSG_3_4(b0, b1) \
+do \
+{ \
+b0 = _mm_unpacklo_epi64(m3, m5); \
+b1 = _mm_unpacklo_epi64(m0, m4); \
+} while(0)
+
+
+#define LOAD_MSG_4_1(b0, b1) \
+do \
+{ \
+b0 = _mm_unpackhi_epi64(m4, m2); \
+b1 = _mm_unpacklo_epi64(m1, m5); \
+} while(0)
+
+
+#define LOAD_MSG_4_2(b0, b1) \
+do \
+{ \
+b0 = _mm_blend_epi16(m0, m3, 0xF0); \
+b1 = _mm_blend_epi16(m2, m7, 0xF0); \
+} while(0)
+
+
+#define LOAD_MSG_4_3(b0, b1) \
+do \
+{ \
+b0 = _mm_blend_epi16(m7, m5, 0xF0); \
+b1 = _mm_blend_epi16(m3, m1, 0xF0); \
+} while(0)
+
+
+#define LOAD_MSG_4_4(b0, b1) \
+do \
+{ \
+b0 = _mm_alignr_epi8(m6, m0, 8); \
+b1 = _mm_blend_epi16(m4, m6, 0xF0); \
+} while(0)
+
+
+#define LOAD_MSG_5_1(b0, b1) \
+do \
+{ \
+b0 = _mm_unpacklo_epi64(m1, m3); \
+b1 = _mm_unpacklo_epi64(m0, m4); \
+} while(0)
+
+
+#define LOAD_MSG_5_2(b0, b1) \
+do \
+{ \
+b0 = _mm_unpacklo_epi64(m6, m5); \
+b1 = _mm_unpackhi_epi64(m5, m1); \
+} while(0)
+
+
+#define LOAD_MSG_5_3(b0, b1) \
+do \
+{ \
+b0 = _mm_blend_epi16(m2, m3, 0xF0); \
+b1 = _mm_unpackhi_epi64(m7, m0); \
+} while(0)
+
+
+#define LOAD_MSG_5_4(b0, b1) \
+do \
+{ \
+b0 = _mm_unpackhi_epi64(m6, m2); \
+b1 = _mm_blend_epi16(m7, m4, 0xF0); \
+} while(0)
+
+
+#define LOAD_MSG_6_1(b0, b1) \
+do \
+{ \
+b0 = _mm_blend_epi16(m6, m0, 0xF0); \
+b1 = _mm_unpacklo_epi64(m7, m2); \
+} while(0)
+
+
+#define LOAD_MSG_6_2(b0, b1) \
+do \
+{ \
+b0 = _mm_unpackhi_epi64(m2, m7); \
+b1 = _mm_alignr_epi8(m5, m6, 8); \
+} while(0)
+
+
+#define LOAD_MSG_6_3(b0, b1) \
+do \
+{ \
+b0 = _mm_unpacklo_epi64(m0, m3); \
+b1 = _mm_shuffle_epi32(m4, _MM_SHUFFLE(1,0,3,2)); \
+} while(0)
+
+
+#define LOAD_MSG_6_4(b0, b1) \
+do \
+{ \
+b0 = _mm_unpackhi_epi64(m3, m1); \
+b1 = _mm_blend_epi16(m1, m5, 0xF0); \
+} while(0)
+
+
+#define LOAD_MSG_7_1(b0, b1) \
+do \
+{ \
+b0 = _mm_unpackhi_epi64(m6, m3); \
+b1 = _mm_blend_epi16(m6, m1, 0xF0); \
+} while(0)
+
+
+#define LOAD_MSG_7_2(b0, b1) \
+do \
+{ \
+b0 = _mm_alignr_epi8(m7, m5, 8); \
+b1 = _mm_unpackhi_epi64(m0, m4); \
+} while(0)
+
+
+#define LOAD_MSG_7_3(b0, b1) \
+do \
+{ \
+b0 = _mm_unpackhi_epi64(m2, m7); \
+b1 = _mm_unpacklo_epi64(m4, m1); \
+} while(0)
+
+
+#define LOAD_MSG_7_4(b0, b1) \
+do \
+{ \
+b0 = _mm_unpacklo_epi64(m0, m2); \
+b1 = _mm_unpacklo_epi64(m3, m5); \
+} while(0)
+
+
+#define LOAD_MSG_8_1(b0, b1) \
+do \
+{ \
+b0 = _mm_unpacklo_epi64(m3, m7); \
+b1 = _mm_alignr_epi8(m0, m5, 8); \
+} while(0)
+
+
+#define LOAD_MSG_8_2(b0, b1) \
+do \
+{ \
+b0 = _mm_unpackhi_epi64(m7, m4); \
+b1 = _mm_alignr_epi8(m4, m1, 8); \
+} while(0)
+
+
+#define LOAD_MSG_8_3(b0, b1) \
+do \
+{ \
+b0 = m6; \
+b1 = _mm_alignr_epi8(m5, m0, 8); \
+} while(0)
+
+
+#define LOAD_MSG_8_4(b0, b1) \
+do \
+{ \
+b0 = _mm_blend_epi16(m1, m3, 0xF0); \
+b1 = m2; \
+} while(0)
+
+
+#define LOAD_MSG_9_1(b0, b1) \
+do \
+{ \
+b0 = _mm_unpacklo_epi64(m5, m4); \
+b1 = _mm_unpackhi_epi64(m3, m0); \
+} while(0)
+
+
+#define LOAD_MSG_9_2(b0, b1) \
+do \
+{ \
+b0 = _mm_unpacklo_epi64(m1, m2); \
+b1 = _mm_blend_epi16(m3, m2, 0xF0); \
+} while(0)
+
+
+#define LOAD_MSG_9_3(b0, b1) \
+do \
+{ \
+b0 = _mm_unpackhi_epi64(m7, m4); \
+b1 = _mm_unpackhi_epi64(m1, m6); \
+} while(0)
+
+
+#define LOAD_MSG_9_4(b0, b1) \
+do \
+{ \
+b0 = _mm_alignr_epi8(m7, m5, 8); \
+b1 = _mm_unpacklo_epi64(m6, m0); \
+} while(0)
+
+
+#define LOAD_MSG_10_1(b0, b1) \
+do \
+{ \
+b0 = _mm_unpacklo_epi64(m0, m1); \
+b1 = _mm_unpacklo_epi64(m2, m3); \
+} while(0)
+
+
+#define LOAD_MSG_10_2(b0, b1) \
+do \
+{ \
+b0 = _mm_unpackhi_epi64(m0, m1); \
+b1 = _mm_unpackhi_epi64(m2, m3); \
+} while(0)
+
+
+#define LOAD_MSG_10_3(b0, b1) \
+do \
+{ \
+b0 = _mm_unpacklo_epi64(m4, m5); \
+b1 = _mm_unpacklo_epi64(m6, m7); \
+} while(0)
+
+
+#define LOAD_MSG_10_4(b0, b1) \
+do \
+{ \
+b0 = _mm_unpackhi_epi64(m4, m5); \
+b1 = _mm_unpackhi_epi64(m6, m7); \
+} while(0)
+
+
+#define LOAD_MSG_11_1(b0, b1) \
+do \
+{ \
+b0 = _mm_unpacklo_epi64(m7, m2); \
+b1 = _mm_unpackhi_epi64(m4, m6); \
+} while(0)
+
+
+#define LOAD_MSG_11_2(b0, b1) \
+do \
+{ \
+b0 = _mm_unpacklo_epi64(m5, m4); \
+b1 = _mm_alignr_epi8(m3, m7, 8); \
+} while(0)
+
+
+#define LOAD_MSG_11_3(b0, b1) \
+do \
+{ \
+b0 = _mm_shuffle_epi32(m0, _MM_SHUFFLE(1,0,3,2)); \
+b1 = _mm_unpackhi_epi64(m5, m2); \
+} while(0)
+
+
+#define LOAD_MSG_11_4(b0, b1) \
+do \
+{ \
+b0 = _mm_unpacklo_epi64(m6, m1); \
+b1 = _mm_unpackhi_epi64(m3, m1); \
+} while(0)
+
+
+#endif
diff --git a/src/thirdparty/blake2/blake2b-round.h b/src/thirdparty/blake2/blake2b-round.h
new file mode 100644
index 0000000..6537fff
--- /dev/null
+++ b/src/thirdparty/blake2/blake2b-round.h
@@ -0,0 +1,157 @@
+/*
+   BLAKE2 reference source code package - optimized C implementations
+
+   Copyright 2012, Samuel Neves <sneves@dei.uc.pt>.  You may use this under the
+   terms of the CC0, the OpenSSL Licence, or the Apache Public License 2.0, at
+   your option.  The terms of these licenses can be found at:
+
+   - CC0 1.0 Universal : http://creativecommons.org/publicdomain/zero/1.0
+   - OpenSSL license   : https://www.openssl.org/source/license.html
+   - Apache 2.0        : http://www.apache.org/licenses/LICENSE-2.0
+
+   More information about the BLAKE2 hash function can be found at
+   https://blake2.net.
+*/
+#ifndef BLAKE2B_ROUND_H
+#define BLAKE2B_ROUND_H
+
+#define LOADU(p)  _mm_loadu_si128( (const __m128i *)(p) )
+#define STOREU(p,r) _mm_storeu_si128((__m128i *)(p), r)
+
+#define TOF(reg) _mm_castsi128_ps((reg))
+#define TOI(reg) _mm_castps_si128((reg))
+
+#define LIKELY(x) __builtin_expect((x),1)
+
+
+/* Microarchitecture-specific macros */
+#ifndef HAVE_XOP
+#ifdef HAVE_SSSE3
+#define _mm_roti_epi64(x, c) \
+    (-(c) == 32) ? _mm_shuffle_epi32((x), _MM_SHUFFLE(2,3,0,1))  \
+    : (-(c) == 24) ? _mm_shuffle_epi8((x), r24) \
+    : (-(c) == 16) ? _mm_shuffle_epi8((x), r16) \
+    : (-(c) == 63) ? _mm_xor_si128(_mm_srli_epi64((x), -(c)), _mm_add_epi64((x), (x)))  \
+    : _mm_xor_si128(_mm_srli_epi64((x), -(c)), _mm_slli_epi64((x), 64-(-(c))))
+#else
+#define _mm_roti_epi64(r, c) _mm_xor_si128(_mm_srli_epi64( (r), -(c) ),_mm_slli_epi64( (r), 64-(-(c)) ))
+#endif
+#else
+/* ... */
+#endif
+
+
+
+#define G1(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1) \
+  row1l = _mm_add_epi64(_mm_add_epi64(row1l, b0), row2l); \
+  row1h = _mm_add_epi64(_mm_add_epi64(row1h, b1), row2h); \
+  \
+  row4l = _mm_xor_si128(row4l, row1l); \
+  row4h = _mm_xor_si128(row4h, row1h); \
+  \
+  row4l = _mm_roti_epi64(row4l, -32); \
+  row4h = _mm_roti_epi64(row4h, -32); \
+  \
+  row3l = _mm_add_epi64(row3l, row4l); \
+  row3h = _mm_add_epi64(row3h, row4h); \
+  \
+  row2l = _mm_xor_si128(row2l, row3l); \
+  row2h = _mm_xor_si128(row2h, row3h); \
+  \
+  row2l = _mm_roti_epi64(row2l, -24); \
+  row2h = _mm_roti_epi64(row2h, -24); \
+
+#define G2(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1) \
+  row1l = _mm_add_epi64(_mm_add_epi64(row1l, b0), row2l); \
+  row1h = _mm_add_epi64(_mm_add_epi64(row1h, b1), row2h); \
+  \
+  row4l = _mm_xor_si128(row4l, row1l); \
+  row4h = _mm_xor_si128(row4h, row1h); \
+  \
+  row4l = _mm_roti_epi64(row4l, -16); \
+  row4h = _mm_roti_epi64(row4h, -16); \
+  \
+  row3l = _mm_add_epi64(row3l, row4l); \
+  row3h = _mm_add_epi64(row3h, row4h); \
+  \
+  row2l = _mm_xor_si128(row2l, row3l); \
+  row2h = _mm_xor_si128(row2h, row3h); \
+  \
+  row2l = _mm_roti_epi64(row2l, -63); \
+  row2h = _mm_roti_epi64(row2h, -63); \
+
+#if defined(HAVE_SSSE3)
+#define DIAGONALIZE(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h) \
+  t0 = _mm_alignr_epi8(row2h, row2l, 8); \
+  t1 = _mm_alignr_epi8(row2l, row2h, 8); \
+  row2l = t0; \
+  row2h = t1; \
+  \
+  t0 = row3l; \
+  row3l = row3h; \
+  row3h = t0;    \
+  \
+  t0 = _mm_alignr_epi8(row4h, row4l, 8); \
+  t1 = _mm_alignr_epi8(row4l, row4h, 8); \
+  row4l = t1; \
+  row4h = t0;
+
+#define UNDIAGONALIZE(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h) \
+  t0 = _mm_alignr_epi8(row2l, row2h, 8); \
+  t1 = _mm_alignr_epi8(row2h, row2l, 8); \
+  row2l = t0; \
+  row2h = t1; \
+  \
+  t0 = row3l; \
+  row3l = row3h; \
+  row3h = t0; \
+  \
+  t0 = _mm_alignr_epi8(row4l, row4h, 8); \
+  t1 = _mm_alignr_epi8(row4h, row4l, 8); \
+  row4l = t1; \
+  row4h = t0;
+#else
+
+#define DIAGONALIZE(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h) \
+  t0 = row4l;\
+  t1 = row2l;\
+  row4l = row3l;\
+  row3l = row3h;\
+  row3h = row4l;\
+  row4l = _mm_unpackhi_epi64(row4h, _mm_unpacklo_epi64(t0, t0)); \
+  row4h = _mm_unpackhi_epi64(t0, _mm_unpacklo_epi64(row4h, row4h)); \
+  row2l = _mm_unpackhi_epi64(row2l, _mm_unpacklo_epi64(row2h, row2h)); \
+  row2h = _mm_unpackhi_epi64(row2h, _mm_unpacklo_epi64(t1, t1))
+
+#define UNDIAGONALIZE(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h) \
+  t0 = row3l;\
+  row3l = row3h;\
+  row3h = t0;\
+  t0 = row2l;\
+  t1 = row4l;\
+  row2l = _mm_unpackhi_epi64(row2h, _mm_unpacklo_epi64(row2l, row2l)); \
+  row2h = _mm_unpackhi_epi64(t0, _mm_unpacklo_epi64(row2h, row2h)); \
+  row4l = _mm_unpackhi_epi64(row4l, _mm_unpacklo_epi64(row4h, row4h)); \
+  row4h = _mm_unpackhi_epi64(row4h, _mm_unpacklo_epi64(t1, t1))
+
+#endif
+
+#if defined(HAVE_SSE41)
+#include "blake2b-load-sse41.h"
+#else
+#include "blake2b-load-sse2.h"
+#endif
+
+#define ROUND(r) \
+  LOAD_MSG_ ##r ##_1(b0, b1); \
+  G1(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1); \
+  LOAD_MSG_ ##r ##_2(b0, b1); \
+  G2(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1); \
+  DIAGONALIZE(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h); \
+  LOAD_MSG_ ##r ##_3(b0, b1); \
+  G1(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1); \
+  LOAD_MSG_ ##r ##_4(b0, b1); \
+  G2(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h,b0,b1); \
+  UNDIAGONALIZE(row1l,row2l,row3l,row4l,row1h,row2h,row3h,row4h);
+
+#endif
diff --git a/src/thirdparty/blake2/blake2b.c b/src/thirdparty/blake2/blake2b.c
new file mode 100644
index 0000000..c8c1c5f
--- /dev/null
+++ b/src/thirdparty/blake2/blake2b.c
@@ -0,0 +1,373 @@
+/*
+   BLAKE2 reference source code package - optimized C implementations
+
+   Copyright 2012, Samuel Neves <sneves@dei.uc.pt>.  You may use this under the
+   terms of the CC0, the OpenSSL Licence, or the Apache Public License 2.0, at
+   your option.  The terms of these licenses can be found at:
+
+   - CC0 1.0 Universal : http://creativecommons.org/publicdomain/zero/1.0
+   - OpenSSL license   : https://www.openssl.org/source/license.html
+   - Apache 2.0        : http://www.apache.org/licenses/LICENSE-2.0
+
+   More information about the BLAKE2 hash function can be found at
+   https://blake2.net.
+*/
+
+#include <stdint.h>
+#include <string.h>
+#include <stdio.h>
+
+#include "blake2.h"
+#include "blake2-impl.h"
+
+#include "blake2-config.h"
+
+#ifdef _MSC_VER
+#include <intrin.h> /* for _mm_set_epi64x */
+#endif
+#include <emmintrin.h>
+#if defined(HAVE_SSSE3)
+#include <tmmintrin.h>
+#endif
+#if defined(HAVE_SSE41)
+#include <smmintrin.h>
+#endif
+#if defined(HAVE_AVX)
+#include <immintrin.h>
+#endif
+#if defined(HAVE_XOP)
+#include <x86intrin.h>
+#endif
+
+#include "blake2b-round.h"
+
+static const uint64_t blake2b_IV[8] =
+{
+  0x6a09e667f3bcc908ULL, 0xbb67ae8584caa73bULL,
+  0x3c6ef372fe94f82bULL, 0xa54ff53a5f1d36f1ULL,
+  0x510e527fade682d1ULL, 0x9b05688c2b3e6c1fULL,
+  0x1f83d9abfb41bd6bULL, 0x5be0cd19137e2179ULL
+};
+
+/* Some helper functions */
+static void blake2b_set_lastnode( blake2b_state *S )
+{
+  S->f[1] = (uint64_t)-1;
+}
+
+static int blake2b_is_lastblock( const blake2b_state *S )
+{
+  return S->f[0] != 0;
+}
+
+static void blake2b_set_lastblock( blake2b_state *S )
+{
+  if( S->last_node ) blake2b_set_lastnode( S );
+
+  S->f[0] = (uint64_t)-1;
+}
+
+static void blake2b_increment_counter( blake2b_state *S, const uint64_t inc )
+{
+  S->t[0] += inc;
+  S->t[1] += ( S->t[0] < inc );
+}
+
+/* init xors IV with input parameter block */
+int blake2b_init_param( blake2b_state *S, const blake2b_param *P )
+{
+  size_t i;
+  /*blake2b_init0( S ); */
+  const unsigned char * v = ( const unsigned char * )( blake2b_IV );
+  const unsigned char * p = ( const unsigned char * )( P );
+  unsigned char * h = ( unsigned char * )( S->h );
+  /* IV XOR ParamBlock */
+  memset( S, 0, sizeof( blake2b_state ) );
+
+  for( i = 0; i < BLAKE2B_OUTBYTES; ++i ) h[i] = v[i] ^ p[i];
+
+  S->outlen = P->digest_length;
+  return 0;
+}
+
+
+/* Some sort of default parameter block initialization, for sequential blake2b */
+int blake2b_init( blake2b_state *S, size_t outlen )
+{
+  blake2b_param P[1];
+
+  if ( ( !outlen ) || ( outlen > BLAKE2B_OUTBYTES ) ) return -1;
+
+  P->digest_length = (uint8_t)outlen;
+  P->key_length    = 0;
+  P->fanout        = 1;
+  P->depth         = 1;
+  store32( &P->leaf_length, 0 );
+  store32( &P->node_offset, 0 );
+  store32( &P->xof_length, 0 );
+  P->node_depth    = 0;
+  P->inner_length  = 0;
+  memset( P->reserved, 0, sizeof( P->reserved ) );
+  memset( P->salt,     0, sizeof( P->salt ) );
+  memset( P->personal, 0, sizeof( P->personal ) );
+
+  return blake2b_init_param( S, P );
+}
+
+int blake2b_init_key( blake2b_state *S, size_t outlen, const void *key, size_t keylen )
+{
+  blake2b_param P[1];
+
+  if ( ( !outlen ) || ( outlen > BLAKE2B_OUTBYTES ) ) return -1;
+
+  if ( ( !keylen ) || keylen > BLAKE2B_KEYBYTES ) return -1;
+
+  P->digest_length = (uint8_t)outlen;
+  P->key_length    = (uint8_t)keylen;
+  P->fanout        = 1;
+  P->depth         = 1;
+  store32( &P->leaf_length, 0 );
+  store32( &P->node_offset, 0 );
+  store32( &P->xof_length, 0 );
+  P->node_depth    = 0;
+  P->inner_length  = 0;
+  memset( P->reserved, 0, sizeof( P->reserved ) );
+  memset( P->salt,     0, sizeof( P->salt ) );
+  memset( P->personal, 0, sizeof( P->personal ) );
+
+  if( blake2b_init_param( S, P ) < 0 )
+    return 0;
+
+  {
+    uint8_t block[BLAKE2B_BLOCKBYTES];
+    memset( block, 0, BLAKE2B_BLOCKBYTES );
+    memcpy( block, key, keylen );
+    blake2b_update( S, block, BLAKE2B_BLOCKBYTES );
+    secure_zero_memory( block, BLAKE2B_BLOCKBYTES ); /* Burn the key from stack */
+  }
+  return 0;
+}
+
+static void blake2b_compress( blake2b_state *S, const uint8_t block[BLAKE2B_BLOCKBYTES] )
+{
+  __m128i row1l, row1h;
+  __m128i row2l, row2h;
+  __m128i row3l, row3h;
+  __m128i row4l, row4h;
+  __m128i b0, b1;
+  __m128i t0, t1;
+#if defined(HAVE_SSSE3) && !defined(HAVE_XOP)
+  const __m128i r16 = _mm_setr_epi8( 2, 3, 4, 5, 6, 7, 0, 1, 10, 11, 12, 13, 14, 15, 8, 9 );
+  const __m128i r24 = _mm_setr_epi8( 3, 4, 5, 6, 7, 0, 1, 2, 11, 12, 13, 14, 15, 8, 9, 10 );
+#endif
+#if defined(HAVE_SSE41)
+  const __m128i m0 = LOADU( block + 00 );
+  const __m128i m1 = LOADU( block + 16 );
+  const __m128i m2 = LOADU( block + 32 );
+  const __m128i m3 = LOADU( block + 48 );
+  const __m128i m4 = LOADU( block + 64 );
+  const __m128i m5 = LOADU( block + 80 );
+  const __m128i m6 = LOADU( block + 96 );
+  const __m128i m7 = LOADU( block + 112 );
+#else
+  const uint64_t  m0 = load64(block +  0 * sizeof(uint64_t));
+  const uint64_t  m1 = load64(block +  1 * sizeof(uint64_t));
+  const uint64_t  m2 = load64(block +  2 * sizeof(uint64_t));
+  const uint64_t  m3 = load64(block +  3 * sizeof(uint64_t));
+  const uint64_t  m4 = load64(block +  4 * sizeof(uint64_t));
+  const uint64_t  m5 = load64(block +  5 * sizeof(uint64_t));
+  const uint64_t  m6 = load64(block +  6 * sizeof(uint64_t));
+  const uint64_t  m7 = load64(block +  7 * sizeof(uint64_t));
+  const uint64_t  m8 = load64(block +  8 * sizeof(uint64_t));
+  const uint64_t  m9 = load64(block +  9 * sizeof(uint64_t));
+  const uint64_t m10 = load64(block + 10 * sizeof(uint64_t));
+  const uint64_t m11 = load64(block + 11 * sizeof(uint64_t));
+  const uint64_t m12 = load64(block + 12 * sizeof(uint64_t));
+  const uint64_t m13 = load64(block + 13 * sizeof(uint64_t));
+  const uint64_t m14 = load64(block + 14 * sizeof(uint64_t));
+  const uint64_t m15 = load64(block + 15 * sizeof(uint64_t));
+#endif
+  row1l = LOADU( &S->h[0] );
+  row1h = LOADU( &S->h[2] );
+  row2l = LOADU( &S->h[4] );
+  row2h = LOADU( &S->h[6] );
+  row3l = LOADU( &blake2b_IV[0] );
+  row3h = LOADU( &blake2b_IV[2] );
+  row4l = _mm_xor_si128( LOADU( &blake2b_IV[4] ), LOADU( &S->t[0] ) );
+  row4h = _mm_xor_si128( LOADU( &blake2b_IV[6] ), LOADU( &S->f[0] ) );
+  ROUND( 0 );
+  ROUND( 1 );
+  ROUND( 2 );
+  ROUND( 3 );
+  ROUND( 4 );
+  ROUND( 5 );
+  ROUND( 6 );
+  ROUND( 7 );
+  ROUND( 8 );
+  ROUND( 9 );
+  ROUND( 10 );
+  ROUND( 11 );
+  row1l = _mm_xor_si128( row3l, row1l );
+  row1h = _mm_xor_si128( row3h, row1h );
+  STOREU( &S->h[0], _mm_xor_si128( LOADU( &S->h[0] ), row1l ) );
+  STOREU( &S->h[2], _mm_xor_si128( LOADU( &S->h[2] ), row1h ) );
+  row2l = _mm_xor_si128( row4l, row2l );
+  row2h = _mm_xor_si128( row4h, row2h );
+  STOREU( &S->h[4], _mm_xor_si128( LOADU( &S->h[4] ), row2l ) );
+  STOREU( &S->h[6], _mm_xor_si128( LOADU( &S->h[6] ), row2h ) );
+}
+
+
+int blake2b_update( blake2b_state *S, const void *pin, size_t inlen )
+{
+  const unsigned char * in = (const unsigned char *)pin;
+  if( inlen > 0 )
+  {
+    size_t left = S->buflen;
+    size_t fill = BLAKE2B_BLOCKBYTES - left;
+    if( inlen > fill )
+    {
+      S->buflen = 0;
+      memcpy( S->buf + left, in, fill ); /* Fill buffer */
+      blake2b_increment_counter( S, BLAKE2B_BLOCKBYTES );
+      blake2b_compress( S, S->buf ); /* Compress */
+      in += fill; inlen -= fill;
+      while(inlen > BLAKE2B_BLOCKBYTES) {
+        blake2b_increment_counter(S, BLAKE2B_BLOCKBYTES);
+        blake2b_compress( S, in );
+        in += BLAKE2B_BLOCKBYTES;
+        inlen -= BLAKE2B_BLOCKBYTES;
+      }
+    }
+    memcpy( S->buf + S->buflen, in, inlen );
+    S->buflen += inlen;
+  }
+  return 0;
+}
+
+
+int blake2b_final( blake2b_state *S, void *out, size_t outlen )
+{
+  if( out == NULL || outlen < S->outlen )
+    return -1;
+
+  if( blake2b_is_lastblock( S ) )
+    return -1;
+
+  blake2b_increment_counter( S, S->buflen );
+  blake2b_set_lastblock( S );
+  memset( S->buf + S->buflen, 0, BLAKE2B_BLOCKBYTES - S->buflen ); /* Padding */
+  blake2b_compress( S, S->buf );
+
+  memcpy( out, &S->h[0], S->outlen );
+  return 0;
+}
+
+
+int blake2b( void *out, size_t outlen, const void *in, size_t inlen, const void *key, size_t keylen )
+{
+  blake2b_state S[1];
+
+  /* Verify parameters */
+  if ( NULL == in && inlen > 0 ) return -1;
+
+  if ( NULL == out ) return -1;
+
+  if( NULL == key && keylen > 0 ) return -1;
+
+  if( !outlen || outlen > BLAKE2B_OUTBYTES ) return -1;
+
+  if( keylen > BLAKE2B_KEYBYTES ) return -1;
+
+  if( keylen )
+  {
+    if( blake2b_init_key( S, outlen, key, keylen ) < 0 ) return -1;
+  }
+  else
+  {
+    if( blake2b_init( S, outlen ) < 0 ) return -1;
+  }
+
+  blake2b_update( S, ( const uint8_t * )in, inlen );
+  blake2b_final( S, out, outlen );
+  return 0;
+}
+
+int blake2( void *out, size_t outlen, const void *in, size_t inlen, const void *key, size_t keylen ) {
+  return blake2b(out, outlen, in, inlen, key, keylen);
+}
+
+#if defined(SUPERCOP)
+int crypto_hash( unsigned char *out, unsigned char *in, unsigned long long inlen )
+{
+  return blake2b( out, BLAKE2B_OUTBYTES, in, inlen, NULL, 0 );
+}
+#endif
+
+#if defined(BLAKE2B_SELFTEST)
+#include <string.h>
+#include "blake2-kat.h"
+int main( void )
+{
+  uint8_t key[BLAKE2B_KEYBYTES];
+  uint8_t buf[BLAKE2_KAT_LENGTH];
+  size_t i, step;
+
+  for( i = 0; i < BLAKE2B_KEYBYTES; ++i )
+    key[i] = ( uint8_t )i;
+
+  for( i = 0; i < BLAKE2_KAT_LENGTH; ++i )
+    buf[i] = ( uint8_t )i;
+
+  /* Test simple API */
+  for( i = 0; i < BLAKE2_KAT_LENGTH; ++i )
+  {
+    uint8_t hash[BLAKE2B_OUTBYTES];
+    blake2b( hash, BLAKE2B_OUTBYTES, buf, i, key, BLAKE2B_KEYBYTES );
+
+    if( 0 != memcmp( hash, blake2b_keyed_kat[i], BLAKE2B_OUTBYTES ) )
+    {
+      goto fail;
+    }
+  }
+
+  /* Test streaming API */
+  for(step = 1; step < BLAKE2B_BLOCKBYTES; ++step) {
+    for (i = 0; i < BLAKE2_KAT_LENGTH; ++i) {
+      uint8_t hash[BLAKE2B_OUTBYTES];
+      blake2b_state S;
+      uint8_t * p = buf;
+      size_t mlen = i;
+      int err = 0;
+
+      if( (err = blake2b_init_key(&S, BLAKE2B_OUTBYTES, key, BLAKE2B_KEYBYTES)) < 0 ) {
+        goto fail;
+      }
+
+      while (mlen >= step) {
+        if ( (err = blake2b_update(&S, p, step)) < 0 ) {
+          goto fail;
+        }
+        mlen -= step;
+        p += step;
+      }
+      if ( (err = blake2b_update(&S, p, mlen)) < 0) {
+        goto fail;
+      }
+      if ( (err = blake2b_final(&S, hash, BLAKE2B_OUTBYTES)) < 0) {
+        goto fail;
+      }
+
+      if (0 != memcmp(hash, blake2b_keyed_kat[i], BLAKE2B_OUTBYTES)) {
+        goto fail;
+      }
+    }
+  }
+
+  puts( "ok" );
+  return 0;
+fail:
+  puts("error");
+  return -1;
+}
+#endif