User Guide

Contents

Compare GPU Roofline Results

Use the
Roofline Compare
functionality to display Roofline chart data from other
Intel® Advisor
results or non-archived snapshots for comparison purposes to track optimization progress.
To compare the GPU Roofline results you need:
  • A baseline GPU Roofline
    project
    or
    snapshot
  • One or more GPU Roofline projects or snapshots of the same application with an optimization applied
You can only compare Roofline results of the same type: CPU Roofline or GPU Roofline.
For example, to compare the results using snapshots:
  1. Open a baseline GPU Roofline snapshot.
  2. From the
    Compare
    drop-down toolbar, click
    +
    to load a comparison snapshot. You can load multiple snapshots for comparison one by one.
Use the Roofline Compare functionality of Intel Advisor to compare different Roofline results before and after optimization.
When the comparison snapshot is uploaded:
  • The filenames for uploaded results/snapshots are displayed in the
    Compared Results
    region.
  • The Roofline Compare feature automatically recognizes similar loops from both snapshots. It connects related loops with a dashed arrow line and displays the performance improvement between the loops, in per cent, which is calculated as the difference in FLOPS (or INTOPS or OPS) and Total Time.
    The arrowed lines showing the relationship among loops/functions do not reappear if you upload the comparison file.
  • Loops from different snapshots are shown as different icons on the chart. For example, on the picture below, the baseline loops are shown as circles and comparison loops are triangles and diamonds.
  • To highlight all dots from a specific result, open the
    Compare
    drop-down and hover over the result name.
  • Each time you change the Roofline configuration or filter the dots on the chart, the comparison is updated automatically.
  • You can remove a selected result/snapshot from Compared Results and move it to the
    Ready for comparison
    region.
    Click a filename in the
    Ready for comparison
    region to reload the result/snapshot.
  • You can save the comparison itself to a file using the export feature.
  • To find the same loops among the results,
    Intel Advisor
    compares several loop features, such as loop type, nesting level, source code file name and line, and name of the function. When a certain threshold of similar or equal features is reached, the two loops are considered a match and connected with a dashed line.
  • However, this method still has few limitations. Sometimes there can be no match for the same loop if one is optimized or parallelized or moved in the source code to four or more lines from the original code.
  • Intel Advisor
    tries to ensure some balance between matching source code changes and false positives.

Product and Performance Information

1

Performance varies by use, configuration and other factors. Learn more at www.Intel.com/PerformanceIndex.