Quantcast
Viewing all articles
Browse latest Browse all 1853

Pane: Analysis Type - CPU/GPU Concurrency

To access this pane:

  1. Click the Image may be NSFW.
    Clik here to view.
    (standalone GUI)/Image may be NSFW.
    Clik here to view.
    (Visual Studio IDE) New Analysis button on the Intel® VTune™ Amplifier toolbar.

    The New Amplifier Result tab opens with the Analysis Type window active.

  2. From the analysis tree on the left pane, select Platform Analysis > CPU/GPU Concurrency analysis type.

    The analysis configuration pane opens on the right.

The CPU/GPU Concurrency pane provides a default configuration that can be used as is. If you need to modify any of the supported configuration options, click the Copy button in the upper right corner and create your own custom analysis type based on the existing configuration.

The Details section of the CPU/GPU Concurrency pane displays the settings used for this predefined configuration:

Events Table

For the CPU/GPU Concurrency analysis, the VTune Amplifier uses the Clockticks and Instructions Retired hardware events to sample your system and analyze CPU usage. The events table provides details on the event configuration used for the analysis:

Event Data

Description

Event Name

Name of the event to monitor.

Sample After

Number of events after which the VTune Amplifier interrupts the event data collection. The Sample After value depends on the target duration. Based on the duration value, the VTune Amplifier adjusts the Sample After value with a multiplier.

LBR Filter

Last branch records (LBRs) filter applied to the event collection.

Event Description

Brief information on the event. For more details, see Help > Intel Processor Event Reference.

Collection Options

VTune Amplifier uses the following collection options for the CPU/GPU Concurrency analysis:

Use This Option

To Do This

Default Value

Analyze system-wide context switches

Analyze detailed scheduling layout for all threads on the system and identify the nature of context switches for a thread (preemption or synchronization).

Yes

Analyze GPU usage

Analyze GPU usage and frame rate and identify whether your application is GPU or CPU bound.

Yes

Collect stacks

Analyze execution paths based on performance, parallelism, and power consumption metrics.

No

Stack size, in bytes

Specify the size of a raw stack (in bytes) to process. Zero value means unlimited size. Possible values are numbers between 0 and 2147483647.

Note

For driverless collection, the default value is 1024 bytes.

0

Estimate call counts

Obtain statistical estimation of call counts based on the hardware events.

No

Chipset events

Specify a comma-separated list of Android chipset events (up to 5 events) to monitor with the hardware event-based sampling collector.

Note

This option is supported with the VTune Amplifier for Systems only.

No

Analyze memory bandwidth

Collect data to identify where your application is generating significant bandwidth to DRAM.

Note

This option is enabled by default for the analysis types on particular platforms only; for example, for Bandwidth analysis on the Intel Xeon Phi coprocessor. To view the collected bandwidth data, enable the Bandwidth viewpoint.

Yes

Analyze user tasks

Analyze tasks in your code specified via Task API. This option causes higher overhead and increases result size.

Note

This option is available only for analysis types targeted for the Intel Xeon Phi™ coprocessor (code name: Knights Corner) and Intel microarchitecture code name Haswell.

No

Analyze Processor Graphics events

Analyze performance data from Intel HD Graphics based on the predefined groups of GPU metrics (for Windows targets only).

No

GPU sampling interval, us

Specify an interval between GPU samples.

1000

Trace OpenCL and Intel Media SDK programs (Intel HD Graphics only)

Capture the execution time of Intel® Media SDK programs (for Linux targets only) and OpenCL™ kernels on a GPU, identify performance-critical GPU computing tasks, and analyze the performance of OpenCL kernels per GPU hardware metrics (for Windows targets only).

No

Event mode

Limit event-based sampling collection to USER (user events) or OS(system events) mode. By default, all event types are collected.

All

Analyze active power consumption

Analyze energy counters to evaluate the energy dissipated by processor cores, integrated graphics, DRAM, or the entire processor package while executing an application.

No

Analyze idle power consumption

Collect additional metrics to analyze system sleep states during the application execution and understand the power consumption for system wake-ups.

No

Note

You may copy the command line for this configuration using the Command Line... button at the bottom and run this analysis remotely.

See Also


Viewing all articles
Browse latest Browse all 1853

Trending Articles



<script src="https://jsc.adskeeper.com/r/s/rssing.com.1596347.js" async> </script>