Merge branches/quickjs to trunk. This is the way.

git-svn-id: https://www.unprompted.com/svn/projects/tildefriends/trunk@3621 ed5197a5-7fde-0310-b194-c3ffbd925b24

deps/libuv/test/benchmark-fs-stat.c

@@ -0,0 +1,136 @@
+/* Copyright Joyent, Inc. and other Node contributors. All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+ * deal in the Software without restriction, including without limitation the
+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+ * sell copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include "task.h"
+#include "uv.h"
+
+#include <stdio.h>
+#include <stdlib.h>
+
+#define NUM_SYNC_REQS         (10 * 1e5)
+#define NUM_ASYNC_REQS        (1 * (int) 1e5)
+#define MAX_CONCURRENT_REQS   32
+
+#define sync_stat(req, path)                                                  \
+  do {                                                                        \
+    uv_fs_stat(NULL, (req), (path), NULL);                                    \
+    uv_fs_req_cleanup((req));                                                 \
+  }                                                                           \
+  while (0)
+
+struct async_req {
+  const char* path;
+  uv_fs_t fs_req;
+  int* count;
+};
+
+
+static void warmup(const char* path) {
+  uv_fs_t reqs[MAX_CONCURRENT_REQS];
+  unsigned int i;
+
+  /* warm up the thread pool */
+  for (i = 0; i < ARRAY_SIZE(reqs); i++)
+    uv_fs_stat(uv_default_loop(), reqs + i, path, uv_fs_req_cleanup);
+
+  uv_run(uv_default_loop(), UV_RUN_DEFAULT);
+
+  /* warm up the OS dirent cache */
+  for (i = 0; i < 16; i++)
+    sync_stat(reqs + 0, path);
+}
+
+
+static void sync_bench(const char* path) {
+  uint64_t before;
+  uint64_t after;
+  uv_fs_t req;
+  int i;
+
+  /* do the sync benchmark */
+  before = uv_hrtime();
+
+  for (i = 0; i < NUM_SYNC_REQS; i++)
+    sync_stat(&req, path);
+
+  after = uv_hrtime();
+
+  printf("%s stats (sync): %.2fs (%s/s)\n",
+         fmt(1.0 * NUM_SYNC_REQS),
+         (after - before) / 1e9,
+         fmt((1.0 * NUM_SYNC_REQS) / ((after - before) / 1e9)));
+  fflush(stdout);
+}
+
+
+static void stat_cb(uv_fs_t* fs_req) {
+  struct async_req* req = container_of(fs_req, struct async_req, fs_req);
+  uv_fs_req_cleanup(&req->fs_req);
+  if (*req->count == 0) return;
+  uv_fs_stat(uv_default_loop(), &req->fs_req, req->path, stat_cb);
+  (*req->count)--;
+}
+
+
+static void async_bench(const char* path) {
+  struct async_req reqs[MAX_CONCURRENT_REQS];
+  struct async_req* req;
+  uint64_t before;
+  uint64_t after;
+  int count;
+  int i;
+
+  for (i = 1; i <= MAX_CONCURRENT_REQS; i++) {
+    count = NUM_ASYNC_REQS;
+
+    for (req = reqs; req < reqs + i; req++) {
+      req->path = path;
+      req->count = &count;
+      uv_fs_stat(uv_default_loop(), &req->fs_req, req->path, stat_cb);
+    }
+
+    before = uv_hrtime();
+    uv_run(uv_default_loop(), UV_RUN_DEFAULT);
+    after = uv_hrtime();
+
+    printf("%s stats (%d concurrent): %.2fs (%s/s)\n",
+           fmt(1.0 * NUM_ASYNC_REQS),
+           i,
+           (after - before) / 1e9,
+           fmt((1.0 * NUM_ASYNC_REQS) / ((after - before) / 1e9)));
+    fflush(stdout);
+  }
+}
+
+
+/* This benchmark aims to measure the overhead of doing I/O syscalls from
+ * the thread pool. The stat() syscall was chosen because its results are
+ * easy for the operating system to cache, taking the actual I/O overhead
+ * out of the equation.
+ */
+BENCHMARK_IMPL(fs_stat) {
+  const char path[] = ".";
+  warmup(path);
+  sync_bench(path);
+  async_bench(path);
+  MAKE_VALGRIND_HAPPY();
+  return 0;
+}