Vectorization Accuracy Presets
For each perspective,
Intel® Advisor
has several levels of collection accuracy. Each accuracy level is a set of analyses and properties that control what data is collected and the level of collection details. The higher accuracy value you choose, the higher runtime overhead is added. The following accuracy levels are available:
Comparison / Accuracy Level | Low | Medium | High |
---|---|---|---|
Overhead | 1.1x
| 5 - 8x
| 10 - 40x
|
Goal | Get basic insights about how well your application is vectorized and how you can improve vectorization efficiency
| Get more insights about how well your application is vectorized and the number of iterations in loops/functions
| Get detailed insights about your application performance, including performance issues and detailed optimization recommendations
|
Analyses | Survey
| Survey + Characterization (Trip Counts)
| Survey + Characterization (Trip Counts, FLOP, Call Stacks) + Memory Access Patterns
|
Result | Basic Survey report
| Survey report extended with trip count data
| Extended Survey report with trip counts and floating-point and integer operations (FLOP and INTOP)
Memory Access Patters with memory traffic data and memory usage issues
|
You can choose custom accuracy and set a custom perspective flow for your application. For more information, see
Customize
Vectorization and Code Insights
Perspective.
There is a variety of techniques available to minimize data collection, result size, and execution overhead. Check
Minimize Analysis Overhead.