DPDK-dev Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: "Morten Brørup" <mb@smartsharesystems.com>
To: dev@dpdk.org, Jerin Jacob <jerinj@marvell.com>,
	Kiran Kumar K <kirankumark@marvell.com>,
	Nithin Dabilpuram <ndabilpuram@marvell.com>,
	Zhirun Yan <yanzhirun_163@163.com>,
	Saeed Bishara <saeed.bishara.os@gmail.com>
Cc: "Morten Brørup" <mb@smartsharesystems.com>
Subject: [PATCH v7] graph: add optional profiling stats
Date: Fri,  3 Jul 2026 13:53:01 +0000	[thread overview]
Message-ID: <20260703135301.3040663-1-mb@smartsharesystems.com> (raw)
In-Reply-To: <20260619202047.2809165-1-mb@smartsharesystems.com>

graph: add optional profiling stats

Added graph node profiling stats, build time configurable by enabling
RTE_GRAPH_PROFILE in rte_config.h.

Signed-off-by: Morten Brørup <mb@smartsharesystems.com>
---
v7:
* Use RTE_DIM() in histogram for loop.
* Added static_assert for histogram index values.
* Minor details to please checkpatch.
  Although I disagree with requiring a space when indexing into
  a constant array "(const type []){values} [idx];",
  I have changed the code to comply.
v6:
* Consolidate the four histogram entries into one array. (Saeed Bishara)
* Sample at 32 objs instead of a half burst. (Saeed Bishara)
* Moved stats to different location in rte_node structure. (Jerin)
* Minor details to please checkpatch.
v5:
* Added stats for a half burst and a full burst.
v4:
* Added documentation. (AI)
* Added more comments. (AI)
* Improved dump. (AI)
* Debug shows both cycles/call and cycles/obj.
v3:
* Debug shows cycles/obj instead of cycles/call.
* Fixed missing --in-reply-to.
v2:
* Fixed indentation.
---
 config/rte_config.h                 |  1 +
 doc/guides/prog_guide/graph_lib.rst |  1 +
 lib/graph/graph_debug.c             | 48 +++++++++++++++++++++++++++++
 lib/graph/node.c                    |  2 ++
 lib/graph/rte_graph_worker_common.h | 29 +++++++++++++++--
 5 files changed, 78 insertions(+), 3 deletions(-)

diff --git a/config/rte_config.h b/config/rte_config.h
index 0447cdf2ad..1942c1b1ec 100644
--- a/config/rte_config.h
+++ b/config/rte_config.h
@@ -106,6 +106,7 @@
 /* rte_graph defines */
 #define RTE_GRAPH_BURST_SIZE 256
 #define RTE_LIBRTE_GRAPH_STATS 1
+/* RTE_GRAPH_PROFILE is not set */
 
 /****** driver defines ********/
 
diff --git a/doc/guides/prog_guide/graph_lib.rst b/doc/guides/prog_guide/graph_lib.rst
index 8dd49c19d2..bc36042296 100644
--- a/doc/guides/prog_guide/graph_lib.rst
+++ b/doc/guides/prog_guide/graph_lib.rst
@@ -49,6 +49,7 @@ Performance tuning parameters
   RTE_GRAPH_BURST_SIZE config option.
   The testing shows, on x86 and arm64 servers, The sweet spot is 256 burst
   size. While on arm64 embedded SoCs, it is either 64 or 128.
+- Enable the ``RTE_GRAPH_PROFILE`` config option for more profiling details.
 - Disable node statistics (using ``RTE_LIBRTE_GRAPH_STATS`` config option)
   if not needed.
 
diff --git a/lib/graph/graph_debug.c b/lib/graph/graph_debug.c
index e3b8cccdc1..427c6a6e77 100644
--- a/lib/graph/graph_debug.c
+++ b/lib/graph/graph_debug.c
@@ -93,6 +93,54 @@ rte_graph_obj_dump(FILE *f, struct rte_graph *g, bool all)
 				n->dispatch.total_sched_fail);
 		}
 		fprintf(f, "       total_calls=%" PRId64 "\n", n->total_calls);
+		if (rte_graph_has_stats_feature())
+			fprintf(f, "       total_cycles=%" PRIu64 ", avg cycles/call=%.1f\n",
+					n->total_cycles,
+					n->total_calls == 0 ? 0.0 :
+					(double)n->total_cycles / (double)n->total_calls);
+#ifdef RTE_GRAPH_PROFILE
+		int64_t calls_other = n->total_calls;
+		int64_t cycles_other = n->total_cycles;
+		int64_t objs_other = n->total_objs;
+		for (int idx = 0; idx < RTE_DIM(n->usage_stats) + 1; idx++) {
+			uint64_t calls;
+			uint64_t cycles;
+			double objs_per_call;
+			if (idx < RTE_DIM(n->usage_stats)) {
+				static_assert(RTE_DIM(n->usage_stats) == 4);
+				uint16_t idx_objs = (const uint16_t []){0, 1, 32,
+						RTE_GRAPH_BURST_SIZE} [idx];
+				fprintf(f, "       objs[%u]\n", idx_objs);
+				calls = n->usage_stats[idx].calls;
+				cycles = n->usage_stats[idx].cycles;
+				objs_per_call = (double)idx_objs;
+				calls_other -= calls;
+				cycles_other -= cycles;
+				objs_other -= idx_objs * calls;
+			} else {
+				fprintf(f, "       objs[other]\n");
+				if (calls_other > 0 && cycles_other > 0 && objs_other > 0) {
+					calls = calls_other;
+					cycles = cycles_other;
+					objs_per_call = (double)objs_other / (double)calls_other;
+					fprintf(f, "         avg objs/call=%.1f\n", objs_per_call);
+				} else {
+					calls = 0;
+					cycles = 0;
+					objs_per_call = 0.0;
+				}
+			}
+			fprintf(f, "         calls=%" PRIu64, calls);
+			if (calls != 0)
+				fprintf(f, ", cycles=%" PRIu64 ", avg cycles/call=%.1f",
+						cycles,
+						(double)cycles / (double)calls);
+			if (calls != 0 && objs_per_call != 0.0)
+				fprintf(f, ", avg cycles/obj=%.1f",
+						(double)cycles / (double)calls / objs_per_call);
+			fprintf(f, "\n");
+		}
+#endif
 		for (i = 0; i < n->nb_edges; i++)
 			fprintf(f, "          edge[%d] <%s>\n", i,
 				n->nodes[i]->name);
diff --git a/lib/graph/node.c b/lib/graph/node.c
index 1fce3e6632..19b38881ae 100644
--- a/lib/graph/node.c
+++ b/lib/graph/node.c
@@ -110,10 +110,12 @@ __rte_node_register(const struct rte_node_register *reg)
 	rte_edge_t i;
 	size_t sz;
 
+#ifndef RTE_GRAPH_PROFILE
 	/* Limit Node specific metadata to one cacheline on 64B CL machine */
 	RTE_BUILD_BUG_ON((offsetof(struct rte_node, nodes) -
 			  offsetof(struct rte_node, ctx)) !=
 			 RTE_CACHE_LINE_MIN_SIZE);
+#endif
 
 	graph_spinlock_lock();
 
diff --git a/lib/graph/rte_graph_worker_common.h b/lib/graph/rte_graph_worker_common.h
index 4ab53a533e..bac0fc534e 100644
--- a/lib/graph/rte_graph_worker_common.h
+++ b/lib/graph/rte_graph_worker_common.h
@@ -121,6 +121,14 @@ struct __rte_cache_aligned rte_node {
 	rte_graph_off_t xstat_off; /**< Offset to xstat counters. */
 
 	/** Fast path area cache line 2. */
+#ifdef RTE_GRAPH_PROFILE
+	/** Usage when this node processed 0, 1, 32 or a full burst of objects. */
+	struct __rte_cache_aligned {
+		uint64_t calls;     /**< Calls done. */
+		uint64_t cycles;    /**< Cycles spent. */
+	} usage_stats[4];
+	/** Fast path area cache line 3. */
+#endif
 	__extension__ struct __rte_cache_aligned {
 #define RTE_NODE_CTX_SZ 16
 		union {
@@ -148,8 +156,10 @@ struct __rte_cache_aligned rte_node {
 	};
 };
 
+#ifndef RTE_GRAPH_PROFILE
 static_assert(offsetof(struct rte_node, nodes) - offsetof(struct rte_node, ctx)
 	== RTE_CACHE_LINE_MIN_SIZE, "rte_node fast path area must fit in 64 bytes");
+#endif
 
 /**
  * @internal
@@ -197,7 +207,7 @@ void __rte_node_stream_alloc_size(struct rte_graph *graph,
 static __rte_always_inline void
 __rte_node_process(struct rte_graph *graph, struct rte_node *node)
 {
-	uint64_t start;
+	uint64_t cycles;
 	uint16_t rc;
 	void **objs;
 
@@ -206,11 +216,24 @@ __rte_node_process(struct rte_graph *graph, struct rte_node *node)
 	rte_prefetch0(objs);
 
 	if (rte_graph_has_stats_feature()) {
-		start = rte_rdtsc();
+		cycles = -rte_rdtsc();
 		rc = node->process(graph, node, objs, node->idx);
-		node->total_cycles += rte_rdtsc() - start;
+		cycles += rte_rdtsc();
+		node->total_cycles += cycles;
 		node->total_calls++;
 		node->total_objs += rc;
+#ifdef RTE_GRAPH_PROFILE
+		if (rc <= 1) {
+			node->usage_stats[rc].calls++;
+			node->usage_stats[rc].cycles += cycles;
+		} else if (rc == 32) {
+			node->usage_stats[2].calls++;
+			node->usage_stats[2].cycles += cycles;
+		} else if (rc == RTE_GRAPH_BURST_SIZE) {
+			node->usage_stats[3].calls++;
+			node->usage_stats[3].cycles += cycles;
+		}
+#endif
 	} else {
 		node->process(graph, node, objs, node->idx);
 	}
-- 
2.43.0


  parent reply	other threads:[~2026-07-03 13:53 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-06-19 20:20 [PATCH] graph: add optional profiling stats Morten Brørup
2026-06-19 20:56 ` [PATCH v3] " Morten Brørup
2026-06-21 17:55 ` [PATCH v4] " Morten Brørup
2026-06-21 18:41 ` [PATCH v5] " Morten Brørup
2026-06-23  5:13   ` Jerin Jacob
2026-06-23  6:45     ` Morten Brørup
2026-06-23  6:56       ` Jerin Jacob
2026-06-23  7:10         ` Morten Brørup
2026-06-23  9:08           ` Jerin Jacob
2026-06-23  8:33         ` saeed bishara
2026-06-23 12:04           ` Morten Brørup
2026-06-23 14:10             ` saeed bishara
2026-06-24  7:59               ` Morten Brørup
2026-06-24 13:09                 ` saeed bishara
2026-07-03 13:18 ` [PATCH v6] " Morten Brørup
2026-07-03 13:53 ` Morten Brørup [this message]
2026-07-03 14:22 ` [PATCH v8] " Morten Brørup
2026-07-03 15:43 ` [PATCH v9] " Morten Brørup

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20260703135301.3040663-1-mb@smartsharesystems.com \
    --to=mb@smartsharesystems.com \
    --cc=dev@dpdk.org \
    --cc=jerinj@marvell.com \
    --cc=kirankumark@marvell.com \
    --cc=ndabilpuram@marvell.com \
    --cc=saeed.bishara.os@gmail.com \
    --cc=yanzhirun_163@163.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox