zipf/pareto/gauss: hash cleanup

[fio.git] / lib / zipf.c

diff --git a/lib/zipf.c b/lib/zipf.c
index 527ae294cf9afd6212151e004e2220163e25f888..681df70069ac27419b4545dcbac0ee2de81fdbee 100644 (file)
--- a/lib/zipf.c
+++ b/lib/zipf.c
@@ -10,104 +10,48 @@
 #include "zipf.h"
 #include "../minmax.h"
 #include "../hash.h"
-#include "../os/os.h"
 
-struct fio_zipf_disk {
-       uint64_t ver_magic;
-       uint64_t nranges;
-       uint64_t zetan;
-};
-
-#define FIO_ZIPF_DISK_MAGIC    0x7a697066
-#define FIO_ZIPF_DISK_VER      1
-#define FIO_ZIPF_MAGIC         ((FIO_ZIPF_DISK_MAGIC << 16) | FIO_ZIPF_DISK_VER)
-
-static void write_zipf(struct zipf_state *zs)
-{
-       struct fio_zipf_disk f;
-       char tmp[80];
-       int fd;
-
-       sprintf(tmp, "fio.zipf.%f.%llu", zs->theta, (unsigned long long) zs->nranges);
-       fd = open(tmp, O_CREAT | O_WRONLY, 0644);
-       if (fd == -1)
-               return;
-
-       f.ver_magic = __cpu_to_le64(FIO_ZIPF_MAGIC);
-       f.nranges = __cpu_to_le64(zs->nranges);
-       f.zetan = __cpu_to_le64(fio_double_to_uint64(zs->zetan));
-       if (write(fd, &f, sizeof(f)) != sizeof(f))
-               unlink(tmp);
-
-       close(fd);
-}
+#define ZIPF_MAX_GEN   10000000UL
 
 static void zipf_update(struct zipf_state *zs)
 {
+       unsigned long to_gen;
        unsigned int i;
 
-       log_info("fio: generating zetan for theta=%f, ranges=%lu\n", zs->theta, zs->nranges);
+       /*
+        * It can become very costly to generate long sequences. Just cap it at
+        * 10M max, that should be doable in 1-2s on even slow machines.
+        * Precision will take a slight hit, but nothing major.
+        */
+       to_gen = min(zs->nranges, (uint64_t) ZIPF_MAX_GEN);
 
-       for (i = 0; i < zs->nranges; i++)
+       for (i = 0; i < to_gen; i++)
                zs->zetan += pow(1.0 / (double) (i + 1), zs->theta);
-
-       write_zipf(zs);
 }
 
-static void zipf_load_gen_zeta(struct zipf_state *zs)
+static void shared_rand_init(struct zipf_state *zs, unsigned long nranges,
+                            unsigned int seed)
 {
-       struct fio_zipf_disk f;
-       char tmp[80];
-       int fd;
-
-       sprintf(tmp, "fio.zipf.%f.%llu", zs->theta, (unsigned long long) zs->nranges);
-       fd = open(tmp, O_RDONLY);
-       if (fd == -1) {
-punt:
-               zipf_update(zs);
-               return;
-       }
-
-       if (read(fd, &f, sizeof(f)) != sizeof(f)) {
-               close(fd);
-               goto punt;
-       }
-
-       close(fd);
-
-       f.ver_magic = le64_to_cpu(f.ver_magic);
-       f.nranges = le64_to_cpu(f.nranges);
-       f.zetan = le64_to_cpu(f.zetan);
-
-       if (f.ver_magic != FIO_ZIPF_MAGIC) {
-               unlink(tmp);
-               goto punt;
-       }
-
-       zs->zetan = fio_uint64_to_double(f.zetan);
-       zs->nranges = f.nranges;
+       memset(zs, 0, sizeof(*zs));
+       zs->nranges = nranges;
+
+       init_rand_seed(&zs->rand, seed, 0);
+       zs->rand_off = __rand(&zs->rand);
 }
 
-void zipf_init(struct zipf_state *zs, unsigned long nranges, double theta)
+void zipf_init(struct zipf_state *zs, unsigned long nranges, double theta,
+              unsigned int seed)
 {
-       unsigned int i;
-
-       memset(zs, 0, sizeof(*zs));
+       shared_rand_init(zs, nranges, seed);
 
-       zs->nranges = nranges;
        zs->theta = theta;
+       zs->zeta2 = pow(1.0, zs->theta) + pow(0.5, zs->theta);
 
-       for (i = 1; i <= 2; i++)
-               zs->zeta2 += pow(1.0 / (double) i, zs->theta);
-
-       init_rand(&zs->rand);
-
-       zipf_load_gen_zeta(zs);
+       zipf_update(zs);
 }
 
 unsigned long long zipf_next(struct zipf_state *zs)
 {
-
        double alpha, eta, rand_uni, rand_z;
        unsigned long long n = zs->nranges;
        unsigned long long val;
@@ -115,7 +59,7 @@ unsigned long long zipf_next(struct zipf_state *zs)
        alpha = 1.0 / (1.0 - zs->theta);
        eta = (1.0 - pow(2.0 / n, 1.0 - zs->theta)) / (1.0 - zs->zeta2 / zs->zetan);
 
-       rand_uni = (double) __rand(&zs->rand) / (double) FRAND_MAX;
+       rand_uni = (double) __rand(&zs->rand) / (double) FRAND32_MAX;
        rand_z = rand_uni * zs->zetan;
 
        if (rand_z < 1.0)
@@ -125,23 +69,35 @@ unsigned long long zipf_next(struct zipf_state *zs)
        else
                val = 1 + (unsigned long long)(n * pow(eta*rand_uni - eta + 1.0, alpha));
 
-       return __hash_long(val - 1) % zs->nranges;
+       val--;
+
+       if (!zs->disable_hash)
+               val = __hash_u64(val);
+
+       return (val + zs->rand_off) % zs->nranges;
 }
 
-void pareto_init(struct zipf_state *zs, unsigned long nranges, double h)
+void pareto_init(struct zipf_state *zs, unsigned long nranges, double h,
+                unsigned int seed)
 {
-       memset(zs, 0, sizeof(*zs));
-
-       zs->nranges = nranges;
+       shared_rand_init(zs, nranges, seed);
        zs->pareto_pow = log(h) / log(1.0 - h);
-
-       init_rand(&zs->rand);
 }
 
 unsigned long long pareto_next(struct zipf_state *zs)
 {
-       double rand = (double) __rand(&zs->rand) / (double) FRAND_MAX;
-       unsigned long long n = zs->nranges - 1;
+       double rand = (double) __rand(&zs->rand) / (double) FRAND32_MAX;
+       unsigned long long n;
+
+       n = (zs->nranges - 1) * pow(rand, zs->pareto_pow);
+
+       if (!zs->disable_hash)
+               n = __hash_u64(n);
 
-       return __hash_long(n * pow(rand, zs->pareto_pow)) % zs->nranges;
+       return (n + zs->rand_off)  % zs->nranges;
+}
+
+void zipf_disable_hash(struct zipf_state *zs)
+{
+       zs->disable_hash = true;
 }