Commit de166958 authored by Changpeng Liu's avatar Changpeng Liu Committed by Daniel Verkamp
Browse files

spdk/perf: add average latency statistics to perf tool 



Change-Id: Ic8042e0c7b1e727292af211a3857fda987dfb2b0
Signed-off-by: default avatarChangpeng Liu <changpeng.liu@intel.com>
parent aaa533ad

examples/nvme/perf/perf.c

+22 −6
Original line number Diff line number Diff line
@@ -90,6 +90,7 @@ struct ns_entry { 
struct ns_worker_ctx {

	struct ns_entry		*entry;

	uint64_t		io_completed;

	uint64_t		total_tsc;

	uint64_t		current_queue_depth;

	uint64_t		offset_in_ios;

	bool			is_draining;
@@ -113,6 +114,7 @@ struct ns_worker_ctx { 
struct perf_task {

	struct ns_worker_ctx	*ns_ctx;

	void			*buf;

	uint64_t		submit_tsc;

#if HAVE_LIBAIO

	struct iocb		iocb;

#endif
@@ -406,6 +408,8 @@ submit_single_io(struct ns_worker_ctx *ns_ctx) 
		}

	}



	task->submit_tsc = rte_get_timer_cycles();



	if ((g_rw_percentage == 100) ||

	    (g_rw_percentage != 0 && ((rand_r(&seed) % 100) < g_rw_percentage))) {

#if HAVE_LIBAIO
@@ -448,6 +452,7 @@ task_complete(struct perf_task *task) 
	ns_ctx = task->ns_ctx;

	ns_ctx->current_queue_depth--;

	ns_ctx->io_completed++;

	ns_ctx->total_tsc += rte_get_timer_cycles() - task->submit_tsc;



	rte_mempool_put(task_pool, task);
@@ -615,13 +620,16 @@ static void usage(char *program_name) 
static void

print_performance(void)

{

	float io_per_second, mb_per_second;

	float total_io_per_second, total_mb_per_second;

	uint64_t total_tsc, total_io_completed;

	float io_per_second, mb_per_second, average_latency;

	float total_io_per_second, total_mb_per_second, total_average_latency;

	struct worker_thread	*worker;

	struct ns_worker_ctx	*ns_ctx;



	total_io_per_second = 0;

	total_mb_per_second = 0;

	total_tsc = 0;

	total_io_completed = 0;



	worker = g_workers;

	while (worker) {
@@ -629,18 +637,26 @@ print_performance(void) 
		while (ns_ctx) {

			io_per_second = (float)ns_ctx->io_completed / g_time_in_sec;

			mb_per_second = io_per_second * g_io_size_bytes / (1024 * 1024);

			printf("%-43.43s from core %u: %10.2f IO/s %10.2f MB/s\n",

			average_latency = (float)(ns_ctx->total_tsc / ns_ctx->io_completed) * 1000 * 1000 / g_tsc_rate;

			printf("%-43.43s from core %u: %10.2f IO/s %10.2f MB/s %10.2f us(average latency)\n",

			       ns_ctx->entry->name, worker->lcore,

			       io_per_second, mb_per_second);

			       io_per_second, mb_per_second,

			       average_latency);

			total_io_per_second += io_per_second;

			total_mb_per_second += mb_per_second;

			total_tsc += ns_ctx->total_tsc;

			total_io_completed += ns_ctx->io_completed;

			ns_ctx = ns_ctx->next;

		}

		worker = worker->next;

	}



	assert(total_io_completed != 0);

	total_average_latency = (float)(total_tsc / total_io_completed) * 1000 * 1000 / g_tsc_rate;;



	printf("========================================================\n");

	printf("%-55s: %10.2f IO/s %10.2f MB/s\n",

	       "Total", total_io_per_second, total_mb_per_second);

	printf("%-55s: %10.2f IO/s %10.2f MB/s %10.2f us(average latency)\n",

	       "Total", total_io_per_second, total_mb_per_second, total_average_latency);



	printf("\n");

}