tvm::runtime::profiling::ReportNode Class Reference

Data collected from a profiling run. Includes per-call metrics and per-device metrics. More...

#include <profiling.h>

Inheritance diagram for tvm::runtime::profiling::ReportNode:

Collaboration diagram for tvm::runtime::profiling::ReportNode:

Public Member Functions
String	AsCSV () const
	Output calls in CSV format. More...
String	AsTable (bool sort=true, bool aggregate=true, bool compute_col_sums=true) const
	Create a human readable table of profiling metrics. More...
String	AsJSON () const
	Convert this report to JSON. More...
	TVM_DECLARE_FINAL_OBJECT_INFO (ReportNode, Object)

Public Attributes
Array< Map< String, ffi::Any > >	calls
	A list of function calls and the metrics recorded for that call. More...
Map< String, Map< String, ffi::Any > >	device_metrics
	Metrics collected for the entire run of the model on a per-device basis. More...
Map< String, ffi::Any >	configuration

Static Public Attributes
static constexpr const char *	_type_key = "runtime.profiling.Report"

Detailed Description

Data collected from a profiling run. Includes per-call metrics and per-device metrics.

Member Function Documentation

AsCSV()

String tvm::runtime::profiling::ReportNode::AsCSV

(

)

const

Output calls in CSV format.

Note that this does not include device_metrics, it only includes per-call metrics.

AsJSON()

String tvm::runtime::profiling::ReportNode::AsJSON

(

)

const

Convert this report to JSON.

Output JSON will be of this format:

{
  "calls": [
    {
      "Duration (us)": {
        "microseconds": 12.3
      },
      "Name": "fused_dense",
      "Count": {
        "count": 1
      },
      "Percent": {
        "percent": 10.3
      }
    }
  ],
  "device_metrics": {
    "cpu": {
      "Duration (us)": {
        "microseconds": 334.2
      },
      "Percent": {
        "percent": 100
      }
    }
  }
}

AsTable()

String tvm::runtime::profiling::ReportNode::AsTable	(	bool	sort = true,
		bool	aggregate = true,
		bool	compute_col_sums = true
	)		const

Create a human readable table of profiling metrics.

Parameters

aggregate	Whether or not to join multiple calls to the same op into a single line.
sort	Whether or not to sort call frames by descending duration. If false and if aggregate is false, frames will be sorted by order of appearance in the program. Order is undefined if sort is false and aggregate is true.
compute_col_sums	Whether or not to include sum totals for the Count, Duation, and Percent columns.

TVM_DECLARE_FINAL_OBJECT_INFO()

tvm::runtime::profiling::ReportNode::TVM_DECLARE_FINAL_OBJECT_INFO	(	ReportNode	,
		Object
	)

Member Data Documentation

_type_key

constexpr const char* tvm::runtime::profiling::ReportNode::_type_key = "runtime.profiling.Report"

staticconstexpr

calls

Array<Map<String, ffi::Any> > tvm::runtime::profiling::ReportNode::calls

A list of function calls and the metrics recorded for that call.

Each element is a mapping from metric name to value. Some metrics that appear in every call are "Name" (the function name), "Argument Shapes", and "Duration (us)". Values are one of String, PercentNode, DurationNode, or CountNode.

configuration

Map<String, ffi::Any> tvm::runtime::profiling::ReportNode::configuration

Configuration used for this profiling run. Includes number of threads, executor.

Values must be an object type that can be used with device_metrics.

device_metrics

Map<String, Map<String, ffi::Any> > tvm::runtime::profiling::ReportNode::device_metrics

Metrics collected for the entire run of the model on a per-device basis.

device_metrics is indexed by device name then metric.

These metrics may be larger than the sum of the same metric in calls because these metrics include the overhead of the executor.

---

The documentation for this class was generated from the following file:

• include/tvm/runtime/profiling.h