Cycles: Add sample-based runtime profiler that measures time spent in various parts of the CPU kernel

This commit adds a sample-based profiler that runs during CPU rendering and collects statistics on time spent in different parts of the kernel (ray intersection, shader evaluation etc.) as well as time spent per material and object.

The results are currently not exposed in the user interface or per Python yet, to see the stats on the console pass the "--cycles-print-stats" argument to Cycles (e.g. "./blender -- --cycles-print-stats").

Unfortunately, there is no clear way to extend this functionality to CUDA or OpenCL, so it is CPU-only for now.

Reviewers: brecht, sergey, swerner

Reviewed By: brecht, swerner

Differential Revision: https://developer.blender.org/D3892

1 files changed, 56 insertions, 0 deletions

diff --git a/intern/cycles/render/stats.h b/intern/cycles/render/stats.h
index 2ff0ec3e0e9..ce0529bb8bd 100644
--- a/intern/cycles/render/stats.h
+++ b/intern/cycles/render/stats.h
@@ -17,6 +17,9 @@
 #ifndef __RENDER_STATS_H__
 #define __RENDER_STATS_H__
 
+#include "render/scene.h"
+
+#include "util/util_stats.h"
 #include "util/util_string.h"
 #include "util/util_vector.h"
 
@@ -61,6 +64,51 @@ public:
 	vector<NamedSizeEntry> entries;
 };
 
+class NamedNestedSampleStats {
+public:
+	NamedNestedSampleStats();
+	NamedNestedSampleStats(const string& name, uint64_t samples);
+
+	NamedNestedSampleStats& add_entry(const string& name, uint64_t samples);
+
+	/* Updates sum_samples recursively. */
+	void update_sum();
+
+	string full_report(int indent_level = 0, uint64_t total_samples = 0);
+
+	string name;
+
+	/* self_samples contains only the samples that this specific event got,
+	 * while sum_samples also includes the samples of all sub-entries. */
+	uint64_t self_samples, sum_samples;
+
+	vector<NamedNestedSampleStats> entries;
+};
+
+/* Named entry containing both a time-sample count for objects of a type and a
+ * total count of processed items.
+ * This allows to estimate the time spent per item. */
+class NamedSampleCountPair {
+public:
+	NamedSampleCountPair(const ustring& name, uint64_t samples, uint64_t hits);
+
+	ustring name;
+	uint64_t samples;
+	uint64_t hits;
+};
+
+/* Contains statistics about pairs of samples and counts as described above. */
+class NamedSampleCountStats {
+public:
+	NamedSampleCountStats();
+
+	string full_report(int indent_level = 0);
+	void add(const ustring& name, uint64_t samples, uint64_t hits);
+
+	typedef unordered_map<ustring, NamedSampleCountPair, ustringHash> entry_map;
+	entry_map entries;
+};
+
 /* Statistics about mesh in the render database. */
 class MeshStats {
 public:
@@ -95,8 +143,16 @@ public:
 	/* Return full report as string. */
 	string full_report();
 
+	/* Collect kernel sampling information from Stats. */
+	void collect_profiling(Scene *scene, Stats *stats);
+
+	bool has_profiling;
+
 	MeshStats mesh;
 	ImageStats image;
+	NamedNestedSampleStats kernel;
+	NamedSampleCountStats shaders;
+	NamedSampleCountStats objects;
 };
 
 CCL_NAMESPACE_END