qht-bench: add -p flag to precompute hash values

Precomputing the hash values allows us to perform more frequent
accesses to the hash table, thereby reaching higher throughputs.

We keep the old behaviour by default, since (1) we might confuse
users if they measured a speedup without changing anything in
the QHT implementation, and (2) benchmarking the hash function
"on line" is also valuable.

Before:
$ taskset -c 0 tests/qht-bench -n 1
 Throughput:        38.18 MT/s

After:
$ taskset -c 0 tests/qht-bench -n 1
 Throughput:        38.16 MT/s

After (with precomputing):
$ taskset -c 0 tests/qht-bench -n 1 -p
 Throughput:        50.87 MT/s
Signed-off-by: NEmilio G. Cota <cota@braap.org>
Signed-off-by: NRichard Henderson <richard.henderson@linaro.org>

tests/qht-bench.c

@@ -53,6 +53,7 @@ static unsigned long resize_delay = 1000;
 static double resize_rate; /* 0.0 to 1.0 */
 static unsigned int n_rz_threads = 1;
 static QemuThread *rz_threads;
+static bool precompute_hash;
 
 static double update_rate; /* 0.0 to 1.0 */
 static uint64_t update_threshold;
@@ -101,11 +102,18 @@ static bool is_equal(const void *ap, const void *bp)
     return *a == *b;
 }
 
-static inline uint32_t h(unsigned long v)
+static uint32_t h(unsigned long v)
 {
     return tb_hash_func7(v, 0, 0, 0, 0);
 }
 
+static uint32_t hval(unsigned long v)
+{
+    return v;
+}
+
+static uint32_t (*hfunc)(unsigned long v) = h;
+
 /*
  * From: https://en.wikipedia.org/wiki/Xorshift
  * This is faster than rand_r(), and gives us a wider range (RAND_MAX is only
@@ -149,7 +157,7 @@ static void do_rw(struct thread_info *info)
         bool read;
 
         p = &keys[info->r & (lookup_range - 1)];
-        hash = h(*p);
+        hash = hfunc(*p);
         read = qht_lookup(&ht, p, hash);
         if (read) {
             stats->rd++;
@@ -158,7 +166,7 @@ static void do_rw(struct thread_info *info)
         }
     } else {
         p = &keys[info->r & (update_range - 1)];
-        hash = h(*p);
+        hash = hfunc(*p);
         if (info->write_op) {
             bool written = false;
 
@@ -289,7 +297,9 @@ static void htable_init(void)
     /* avoid allocating memory later by allocating all the keys now */
     keys = g_malloc(sizeof(*keys) * n);
     for (i = 0; i < n; i++) {
-        keys[i] = populate_offset + i;
+        long val = populate_offset + i;
+
+        keys[i] = precompute_hash ? h(val) : hval(val);
     }
 
     /* some sanity checks */
@@ -321,7 +331,7 @@ static void htable_init(void)
 
             r = xorshift64star(r);
             p = &keys[r & (init_range - 1)];
-            hash = h(*p);
+            hash = hfunc(*p);
             if (qht_insert(&ht, p, hash, NULL)) {
                 break;
             }
@@ -412,7 +422,7 @@ static void parse_args(int argc, char *argv[])
     int c;
 
     for (;;) {
-        c = getopt(argc, argv, "d:D:g:k:K:l:hn:N:o:r:Rs:S:u:");
+        c = getopt(argc, argv, "d:D:g:k:K:l:hn:N:o:pr:Rs:S:u:");
         if (c < 0) {
             break;
         }
@@ -451,6 +461,10 @@ static void parse_args(int argc, char *argv[])
         case 'o':
             populate_offset = atol(optarg);
             break;
+        case 'p':
+            precompute_hash = true;
+            hfunc = hval;
+            break;
         case 'r':
             update_range = pow2ceil(atol(optarg));
             break;