Merge 416c52f into 4308cf7

randomgen/src/xoroshiro128/xoroshiro128-benchmark.c

@@ -14,14 +14,16 @@
 
 #define N 1000000000
 
-int main() {
+int main()
+{
   uint64_t count = 0, sum = 0;
   uint64_t seed = 0xDEADBEAF;
   s[0] = splitmix64_next(&seed);
   s[1] = splitmix64_next(&seed);
   int i;
   clock_t begin = clock();
-  for (i = 0; i < N; i++) {
+  for (i = 0; i < N; i++)
+  {
     sum += next();
     count++;
   }

randomgen/src/xoroshiro128/xoroshiro128-test-data-gen.c

@@ -21,50 +21,61 @@
 
 #define N 1000
 
-int main() {
+int main()
+{
   uint64_t sum = 0;
   uint64_t state, seed = 0xDEADBEAF;
   state = seed;
   int i;
-  for (i = 0; i < 2; i++) {
+  for (i = 0; i < 2; i++)
+  {
     s[i] = splitmix64_next(&state);
   }
   uint64_t store[N];
-  for (i = 0; i < N; i++) {
+  for (i = 0; i < N; i++)
+  {
     store[i] = next();
   }
 
   FILE *fp;
   fp = fopen("xoroshiro128-testset-1.csv", "w");
-  if (fp == NULL) {
+  if (fp == NULL)
+  {
     printf("Couldn't open file\n");
     return -1;
   }
   fprintf(fp, "seed, 0x%" PRIx64 "\n", seed);
-  for (i = 0; i < N; i++) {
+  for (i = 0; i < N; i++)
+  {
     fprintf(fp, "%d, 0x%" PRIx64 "\n", i, store[i]);
-    if (i == 999) {
+    if (i == 999)
+    {
       printf("%d, 0x%" PRIx64 "\n", i, store[i]);
     }
   }
   fclose(fp);
 
   seed = state = 0;
-  for (i = 0; i < 2; i++) {
+  for (i = 0; i < 2; i++)
+  {
     s[i] = splitmix64_next(&state);
   }
-  for (i = 0; i < N; i++) {
+  for (i = 0; i < N; i++)
+  {
     store[i] = next();
   }
   fp = fopen("xoroshiro128-testset-2.csv", "w");
-  if (fp == NULL) {
+  if (fp == NULL)
+  {
     printf("Couldn't open file\n");
     return -1;
   }
   fprintf(fp, "seed, 0x%" PRIx64 "\n", seed);
-  for (i = 0; i < N; i++) {
+  for (i = 0; i < N; i++)
+  {
     fprintf(fp, "%d, 0x%" PRIx64 "\n", i, store[i]);
-    if (i == 999) {
+    if (i == 999)
+    {
       printf("%d, 0x%" PRIx64 "\n", i, store[i]);
     }
   }

randomgen/src/xoroshiro128/xoroshiro128.c

@@ -1,52 +1,54 @@
-/*  Written in 2016 by David Blackman and Sebastiano Vigna (vigna@acm.org)
+/*  Written in 2016-2018 by David Blackman and Sebastiano Vigna (vigna@acm.org)
 
 To the extent possible under law, the author has dedicated all copyright
 and related and neighboring rights to this software to the public domain
 worldwide. This software is distributed without any warranty.
 
 See <http://creativecommons.org/publicdomain/zero/1.0/>. */
 
-/* This is the successor to xorshift128+. It is the fastest full-period
-   generator passing BigCrush without systematic failures, but due to the
-   relatively short period it is acceptable only for applications with a
-   mild amount of parallelism; otherwise, use a xorshift1024* generator.
-
-   Beside passing BigCrush, this generator passes the PractRand test suite
-   up to (and included) 16TB, with the exception of binary rank tests, as
-   the lowest bit of this generator is an LFSR of degree 128. The next bit
-   can be described by an LFSR of degree 8256, but in the long run it will
-   fail linearity tests, too. The other bits needs a much higher degree to
-   be represented as LFSRs.
+/* This is xoroshiro128+ 1.0, our best and fastest small-state generator
+   for floating-point numbers. We suggest to use its upper bits for
+   floating-point generation, as it is slightly faster than
+   xoroshiro128**. It passes all tests we are aware of except for the four
+   lower bits, which might fail linearity tests (and just those), so if
+   low linear complexity is not considered an issue (as it is usually the
+   case) it can be used to generate 64-bit outputs, too; moreover, this
+   generator has a very mild Hamming-weight dependency making our test
+   (http://prng.di.unimi.it/hwd.php) fail after 5 TB of output; we believe
+   this slight bias cannot affect any application. If you are concerned,
+   use xoroshiro128** or xoshiro256+.
 
    We suggest to use a sign test to extract a random Boolean value, and
    right shifts to extract subsets of bits.
 
-   Note that the generator uses a simulated rotate operation, which most C
-   compilers will turn into a single instruction. In Java, you can use
-   Long.rotateLeft(). In languages that do not make low-level rotation
-   instructions accessible xorshift128+ could be faster.
-
    The state must be seeded so that it is not everywhere zero. If you have
    a 64-bit seed, we suggest to seed a splitmix64 generator and use its
-   output to fill s. */
+   output to fill s.
+
+   NOTE: the parameters (a=24, b=16, b=37) of this version give slightly
+   better results in our test than the 2016 version (a=55, b=14, c=36).
+*/
 
 #include "xoroshiro128.h"
 
 extern INLINE uint64_t xoroshiro128_next64(xoroshiro128_state *state);
 
 extern INLINE uint32_t xoroshiro128_next32(xoroshiro128_state *state);
 
-void xoroshiro128_jump(xoroshiro128_state *state) {
+void xoroshiro128_jump(xoroshiro128_state *state)
+{
   int i, b;
   uint64_t s0;
   uint64_t s1;
-  static const uint64_t JUMP[] = {0xbeac0467eba5facb, 0xd86b048b86aa9922};
+  static const uint64_t JUMP[] = {0xdf900294d8f554a5, 0x170865df4b3201fc};
 
   s0 = 0;
   s1 = 0;
   for (i = 0; i < sizeof JUMP / sizeof *JUMP; i++)
-    for (b = 0; b < 64; b++) {
-      if (JUMP[i] & UINT64_C(1) << b) {
+    for (b = 0; b < 64; b++)
+    {
+      if (JUMP[i] & UINT64_C(1) << b)
+      {
         s0 ^= state->s[0];
         s1 ^= state->s[1];
       }

randomgen/src/xoroshiro128/xoroshiro128.h

@@ -14,35 +14,41 @@
 #define INLINE inline
 #endif
 
-typedef struct s_xoroshiro128_state {
+typedef struct s_xoroshiro128_state
+{
   uint64_t s[2];
   int has_uint32;
   uint32_t uinteger;
 } xoroshiro128_state;
 
-static INLINE uint64_t rotl(const uint64_t x, int k) {
+static INLINE uint64_t rotl(const uint64_t x, int k)
+{
   return (x << k) | (x >> (64 - k));
 }
 
-static INLINE uint64_t xoroshiro128_next(uint64_t *s) {
+static INLINE uint64_t xoroshiro128_next(uint64_t *s)
+{
   const uint64_t s0 = s[0];
   uint64_t s1 = s[1];
   const uint64_t result = s0 + s1;
 
   s1 ^= s0;
-  s[0] = rotl(s0, 55) ^ s1 ^ (s1 << 14); // a, b
-  s[1] = rotl(s1, 36);                   // c
+  s[0] = rotl(s0, 24) ^ s1 ^ (s1 << 16); // a, b
+  s[1] = rotl(s1, 37);                   // c
 
   return result;
 }
 
-static INLINE uint64_t xoroshiro128_next64(xoroshiro128_state *state) {
+static INLINE uint64_t xoroshiro128_next64(xoroshiro128_state *state)
+{
   return xoroshiro128_next(&state->s[0]);
 }
 
-static INLINE uint32_t xoroshiro128_next32(xoroshiro128_state *state) {
+static INLINE uint32_t xoroshiro128_next32(xoroshiro128_state *state)
+{
   uint64_t next;
-  if (state->has_uint32) {
+  if (state->has_uint32)
+  {
     state->has_uint32 = 0;
     return state->uinteger;
   }

randomgen/src/xoroshiro128/xoroshiro128plus.orig.c

@@ -1,4 +1,4 @@
-/*  Written in 2016 by David Blackman and Sebastiano Vigna (vigna@acm.org)
+/*  Written in 2016-2018 by David Blackman and Sebastiano Vigna (vigna@acm.org)
 
 To the extent possible under law, the author has dedicated all copyright
 and related and neighboring rights to this software to the public domain
@@ -8,44 +8,45 @@ See <http://creativecommons.org/publicdomain/zero/1.0/>. */
 
 #include <stdint.h>
 
-/* This is the successor to xorshift128+. It is the fastest full-period
-   generator passing BigCrush without systematic failures, but due to the
-   relatively short period it is acceptable only for applications with a
-   mild amount of parallelism; otherwise, use a xorshift1024* generator.
-
-   Beside passing BigCrush, this generator passes the PractRand test suite
-   up to (and included) 16TB, with the exception of binary rank tests, as
-   the lowest bit of this generator is an LFSR of degree 128. The next bit
-   can be described by an LFSR of degree 8256, but in the long run it will
-   fail linearity tests, too. The other bits needs a much higher degree to
-   be represented as LFSRs.
+/* This is xoroshiro128+ 1.0, our best and fastest small-state generator
+   for floating-point numbers. We suggest to use its upper bits for
+   floating-point generation, as it is slightly faster than
+   xoroshiro128**. It passes all tests we are aware of except for the four
+   lower bits, which might fail linearity tests (and just those), so if
+   low linear complexity is not considered an issue (as it is usually the
+   case) it can be used to generate 64-bit outputs, too; moreover, this
+   generator has a very mild Hamming-weight dependency making our test
+   (http://prng.di.unimi.it/hwd.php) fail after 5 TB of output; we believe
+   this slight bias cannot affect any application. If you are concerned,
+   use xoroshiro128** or xoshiro256+.
 
    We suggest to use a sign test to extract a random Boolean value, and
    right shifts to extract subsets of bits.
 
-   Note that the generator uses a simulated rotate operation, which most C
-   compilers will turn into a single instruction. In Java, you can use
-   Long.rotateLeft(). In languages that do not make low-level rotation
-   instructions accessible xorshift128+ could be faster.
-
    The state must be seeded so that it is not everywhere zero. If you have
    a 64-bit seed, we suggest to seed a splitmix64 generator and use its
-   output to fill s. */
+   output to fill s.
+
+   NOTE: the parameters (a=24, b=16, b=37) of this version give slightly
+   better results in our test than the 2016 version (a=55, b=14, c=36).
+*/
 
 uint64_t s[2];
 
-static inline uint64_t rotl(const uint64_t x, int k) {
+static inline uint64_t rotl(const uint64_t x, int k)
+{
   return (x << k) | (x >> (64 - k));
 }
 
-uint64_t next(void) {
+uint64_t next(void)
+{
   const uint64_t s0 = s[0];
   uint64_t s1 = s[1];
   const uint64_t result = s0 + s1;
 
   s1 ^= s0;
-  s[0] = rotl(s0, 55) ^ s1 ^ (s1 << 14); // a, b
-  s[1] = rotl(s1, 36);                   // c
+  s[0] = rotl(s0, 24) ^ s1 ^ (s1 << 16); // a, b
+  s[1] = rotl(s1, 37);                   // c
 
   return result;
 }
@@ -54,14 +55,42 @@ uint64_t next(void) {
    to 2^64 calls to next(); it can be used to generate 2^64
    non-overlapping subsequences for parallel computations. */
 
-void jump(void) {
-  static const uint64_t JUMP[] = {0xbeac0467eba5facb, 0xd86b048b86aa9922};
+void jump(void)
+{
+  static const uint64_t JUMP[] = {0xdf900294d8f554a5, 0x170865df4b3201fc};
 
   uint64_t s0 = 0;
   uint64_t s1 = 0;
   for (int i = 0; i < sizeof JUMP / sizeof *JUMP; i++)
-    for (int b = 0; b < 64; b++) {
-      if (JUMP[i] & UINT64_C(1) << b) {
+    for (int b = 0; b < 64; b++)
+    {
+      if (JUMP[i] & UINT64_C(1) << b)
+      {
+        s0 ^= s[0];
+        s1 ^= s[1];
+      }
+      next();
+    }
+  s[0] = s0;
+  s[1] = s1;
+}
+
+/* This is the long-jump function for the generator. It is equivalent to
+   2^96 calls to next(); it can be used to generate 2^32 starting points,
+   from each of which jump() will generate 2^32 non-overlapping
+   subsequences for parallel distributed computations. */
+
+void long_jump(void)
+{
+  static const uint64_t LONG_JUMP[] = {0xd2a98b26625eee7b, 0xdddf9b1090aa7ac1};
+
+  uint64_t s0 = 0;
+  uint64_t s1 = 0;
+  for (int i = 0; i < sizeof LONG_JUMP / sizeof *LONG_JUMP; i++)
+    for (int b = 0; b < 64; b++)
+    {
+      if (LONG_JUMP[i] & UINT64_C(1) << b)
+      {
         s0 ^= s[0];
         s1 ^= s[1];
       }