vlib: calculate per-worker loops/second metric
Use exponential smoothing. Each sample has a half-life of 1
second. reported_rate(t) = reported_rate(t-1) * K + rate(t)*(1-K)
Sample every 20ms, i.e. 50 samples per second
K = exp (-1.0/20.0);
K = 0.95;
Type: feature
Signed-off-by: Dave Barach <dave@barachs.net>
Change-Id: I9aea5dd5fecfaefffb78245316adb4bf62eb2bd4
diff --git a/src/vlib/main.c b/src/vlib/main.c
index 94218d3..f723d10 100644
--- a/src/vlib/main.c
+++ b/src/vlib/main.c
@@ -1683,6 +1683,7 @@
vlib_thread_main_t *tm = vlib_get_thread_main ();
uword i;
u64 cpu_time_now;
+ f64 now;
vlib_frame_queue_main_t *fqm;
u32 *last_node_runtime_indices = 0;
u32 frame_queue_check_counter = 0;
@@ -1926,6 +1927,33 @@
/* Record time stamp in case there are no enabled nodes and above
calls do not update time stamp. */
cpu_time_now = clib_cpu_time_now ();
+ vm->loops_this_reporting_interval++;
+ now = clib_time_now_internal (&vm->clib_time, cpu_time_now);
+ /* Time to update loops_per_second? */
+ if (PREDICT_FALSE (now >= vm->loop_interval_end))
+ {
+ /* Next sample ends in 20ms */
+ if (vm->loop_interval_start)
+ {
+ f64 this_loops_per_second;
+
+ this_loops_per_second =
+ ((f64) vm->loops_this_reporting_interval) / (now -
+ vm->loop_interval_start);
+
+ vm->loops_per_second =
+ vm->loops_per_second * vm->damping_constant +
+ (1.0 - vm->damping_constant) * this_loops_per_second;
+ if (vm->loops_per_second != 0.0)
+ vm->seconds_per_loop = 1.0 / vm->loops_per_second;
+ else
+ vm->seconds_per_loop = 0.0;
+ }
+ /* New interval starts now, and ends in 20ms */
+ vm->loop_interval_start = now;
+ vm->loop_interval_end = now + 2e-4;
+ vm->loops_this_reporting_interval = 0;
+ }
}
}
@@ -2122,6 +2150,16 @@
if ((error = vlib_call_all_config_functions (vm, input, 0 /* is_early */ )))
goto done;
+ /*
+ * Use exponential smoothing, with a half-life of 1 second
+ * reported_rate(t) = reported_rate(t-1) * K + rate(t)*(1-K)
+ *
+ * Sample every 20ms, aka 50 samples per second
+ * K = exp (-1.0/20.0);
+ * K = 0.95
+ */
+ vm->damping_constant = exp (-1.0 / 20.0);
+
/* Sort per-thread init functions before we start threads */
vlib_sort_init_exit_functions (&vm->worker_init_function_registrations);
diff --git a/src/vlib/main.h b/src/vlib/main.h
index af6539e..0e65c81 100644
--- a/src/vlib/main.h
+++ b/src/vlib/main.h
@@ -251,6 +251,14 @@
*/
int need_vlib_worker_thread_node_runtime_update;
+ /* Dispatch loop time accounting */
+ u64 loops_this_reporting_interval;
+ f64 loop_interval_end;
+ f64 loop_interval_start;
+ f64 loops_per_second;
+ f64 seconds_per_loop;
+ f64 damping_constant;
+
/*
* Barrier epoch - Set to current time, each time barrier_sync or
* barrier_release is called with zero recursion.
diff --git a/src/vlib/node_cli.c b/src/vlib/node_cli.c
index bf18702..1ba0575 100644
--- a/src/vlib/node_cli.c
+++ b/src/vlib/node_cli.c
@@ -442,11 +442,12 @@
dt = time_now - nm->time_last_runtime_stats_clear;
vlib_cli_output
(vm,
- "Time %.1f, %f sec internal node vector rate %.2f \n"
+ "Time %.1f, %f sec internal node vector rate %.2f loops/sec %.2f\n"
" vector rates in %.4e, out %.4e, drop %.4e, punt %.4e",
dt,
vlib_get_stat_segment_update_rate (),
internal_node_vector_rates[j],
+ stat_vm->loops_per_second,
(f64) n_input / dt,
(f64) n_output / dt, (f64) n_drop / dt, (f64) n_punt / dt);