View source for Intel VTune Tutorial/Analysis Types

Intel VTune Tutorial/Analysis Types / ← Intel VTune Tutorial: Analysis Types

You do not have permission to edit this page, for the following reason:

---

You can view and copy the source of this page.

[[Category:Tutorials|Intel VTune Tutorials: Analysis Types]]<nowiki /> {{DISPLAYTITLE:Intel VTune Tutorial: Analysis Types}}<nowiki /> {{Syllabus Intel VTune Tutorial}} __TOC__ The third Intel VTune tutorial covers a couple of important analysis types and shows their results. The hotspots analysis is discussed in detail and can tell you where your application spends most of its time. You can go into more detail with the threading analysis, microarchitecture exploration, or HPC performance characterization, each focussing of a specific topic. === Video === <!--T:5--> <youtube width="600" height="340" right>ghFn5IBzjrc</youtube> ([[Media:VTune_analysis_types_compressed.pdf |Slides as pdf]]) === Quiz === <!--T:5--> {{hidden begin |title = 1. What is a "hotspot"?}} <quiz display=simple> { |type="()"} - A burnt spot on the CPU, caused by a badly fitted cooler || Not correct. Please make sure to not melt your CPU. - A code segment that is very inefficient || Not correct. A hotspot is a good candidate for optimization in programs that never were optimized, but it is not necessarily inefficient! + A code segment where the program spends most of its time || Correct. This could be a single instruction, a function, or loop where the program spends a significant amount of time. </quiz> {{hidden end}} {{hidden begin |title = 2. What does the Bottom-Up tab show?}} <quiz display=simple> { |type="()"} - Low level performance results that are close to the hardware ("bottom") || Not correct. The closest thing to this would be the "Event Counts" tab in a microarchitecture exploration analysis. - Time spent in each code section, with the quickest sections at the top || Not correct. It does show the time spent in each code section, but the sorting is not fixed and can be changed. + List of code sections (functions, loops) with their attributed measurements || Correct. This is a great place to start looking for functions and loops to consider during optimization! </quiz> {{hidden end}} {{hidden begin |title = 3. What is a Flame Graph?}} <quiz display=simple> { |type="()"} + A graphical presentation of the call stack on a timeline || Correct. This presentation can quickly show long running code sections and the vertical structure of the call stack. - A statistic of dead CPUs from ill fitted coolers || Not correct. Please make sure to not melt your CPU. - A directed graph of a hotspots call stack. Functions are nodes and weighted edges encode the execution time || Not correct. This sounds like something you would want to use [[Gprof Tutorial | gprof2dot]] for, though! </quiz> {{hidden end}} {{hidden begin |title = 4. Can VTune present source code with the performance of each line?}} <quiz display=simple> { |type="()"} + Yes, by clicking on the function name in any of the tabs || Correct. - Yes, by calling <source enclose="none">vtune /path/to/source.cxx</source> from the commandline || Not correct. - No, VTune only analyses binaries || Not correct. The association between binaries and the source code is possible and very useful. </quiz> {{hidden end}} {{hidden begin |title = 5. How do analysis types differ?}} <quiz display=simple> { |type="()"} - All analysis types are exactly the same || Of course not! - There is exactly one analysis type for each high level metric (Memory, Back-end, Front-end, etc.) || Not correct. Analysis types often share the collected metrics, but the presentation is different. + Analysis types may collect similar data, but the presentation focuses on a certain topic || Correct. Expect some overlap between the types. </quiz> {{hidden end}}

Return to Intel VTune Tutorial: Analysis Types.