• Intel® Graphics Performance Analyzers 2020 R1
  • 03/31/2020
  • Public Content
Contents

Analyzing Duration of DirectX 9/10 Frames

Graphics Frame Analyzer displays the time for both individual ergs and for groups of ergs submitted to the hardware together, to better represent a pipelined workload.
To optimize frame duration, you can use the following strategy:
  • Analyze individual erg timings to isolate the most expensive ergs that might provide significant optimization opportunities.
    NOTE
    In the context of the overall frame time, optimizing a single erg may prove to be less effective than indicated because the pipeline might be hiding latency caused by this particular call.
  • Review region timings to judge how much difference a change will make in the overall frame time.
To compute the time for an individual erg, Graphics Frame Analyzer brackets each individual erg with timing markers. Time for groups (such as render target regions) is calculated as a single unit, so that hardware utilization matches the expected workload. Therefore, the sum of the individual erg timings might not match the total region time.
The following table summarizes the available options for metrics data representation:
 
The metric values shown…
Graphics Frame Analyzer calculates the metric values …
Frame Overview Tab
… represent all ergs within the frame, independent of which ergs are selected.
… by gathering the metrics from the hardware as a single call to the hardware for the entire frame.
Details Tab
… represent only the selected ergs.
… by gathering the metrics from the hardware as a single call to the hardware for all selected ergs. If you select several non-overlapping regions, their values are summed up.
Scene Overview Pane
… represent individual ergs (for individual ergs), or the total for that group of ergs (for all groupings of ergs).
… as a single call to the hardware for both individual ergs and regions.
The following examples demonstrate how the Graphics Frame Analyzer calculates and displays erg metric values.

Analyzing all Ergs in the Frame

Select all ergs in the frame (checkbox labeled Entire Frame). Values for the GPU Duration metric in the
Scene Overview
pane (4,527.5) match those in the
Frame Overview
tab (4,527.5), since both were calculated as a single call to the hardware.

Analyzing a Render Target and all Ergs within It

Now select a single render target and all ergs within that render target. In the
Scene Overview
pane, the values for the entire render target is not the sum of the individual ergs (41.5 versus 56), since the value for the render target was calculated by making a single call for the entire group rather than summing the values for the individual ergs. Similarly, the value shown in the
Details
tab is also calculated for the entire selected region:

Product and Performance Information

1

Intel's compilers may or may not optimize to the same degree for non-Intel microprocessors for optimizations that are not unique to Intel microprocessors. These optimizations include SSE2, SSE3, and SSSE3 instruction sets and other optimizations. Intel does not guarantee the availability, functionality, or effectiveness of any optimization on microprocessors not manufactured by Intel. Microprocessor-dependent optimizations in this product are intended for use with Intel microprocessors. Certain optimizations not specific to Intel microarchitecture are reserved for Intel microprocessors. Please refer to the applicable product User and Reference Guides for more information regarding the specific instruction sets covered by this notice.

Notice revision #20110804