Intel® Advisor Performance Optimization Cookbook
The Intel® Advisor is a tool to help design and optimize high-performing code for modern computer architectures. Each chapter in the Intel® Advisor Cookbook contains step-by-step instructions to help effectively use more cores, vectorization, or heterogeneous processing using Intel Advisor.
What You Will Learn
Use GPU Roofline to Identify Optimization Opportunities
Learn how to analyze performance of a SYCL application running on the Intel® GPU platform, using the Intel® Advisor GPU Roofline perspective.
Analyze a SYCL Application with GPU Roofline
Learn how to evaluate performance of a SYCL application running on graphics processing unit (GPU), pinpoint and prioritize bottlenecks, and define room for optimization using GPU Roofline Insights perspective from command line interface.
This recipe now has a link to an interactive HTML report that you can download and use to follow the workflow and view results.
Model Performance of a C++ App Ported to a GPU.
Learn how to check profitability of porting a native C++ application to a target GPU from command line using a single command to run the Offload Modeling perspective.
This recipe now has a link to an interactive HTML report that you can download and use to follow the workflow and view results.
Model GPU Application Performance for a Different GPU Device.
Learn how to check profitability of porting a SYCL application running on a baseline GPU to a different GPU from command line with the Offload Modeling perspective.
This recipe now has a link to an interactive HTML report that you can download and use to follow the workflow and view results.
Estimate the C++ Application Speedup on a Target GPU.
Learn how to analyze a native C++ application and estimate the profitability of porting the code to SYCL with the Offload Modeling feature of the Intel Advisor.
Identify Code Regions to Offload to GPU and Visualize GPU Usage.
Learn how to identify loops and functions to offload to a GPU and find bottlenecks after offloading to the GPU using Offload Modeling and GPU Roofline Insights features.
Identify Bottlenecks Iteratively: Cache-Aware Roofline.
Learn how to take advantage of Roofline Analysis to identify and address performance bottlenecks.
Optimize Vectorization Aspects of a Real-Time 3D Cardiac Electrophysiology Simulation.
Learn how to improve performance of your application by optimizing its vectorization aspects and efficiency following step-by-step recommendations suggested by Intel Advisor.
Optimize Memory Access Patterns using Loop Interchange and Cache Blocking Techniques.
Learn how to identify memory bottlenecks and improve performance by optimizing memory access patterns.
Visualize Performance Improvements with Roofline Compare.
Learn how to compare optimization strategies by visualizing multiple analysis results on the same chart.
Analyze Vectorization and Memory Aspects of an MPI Application.
Learn how to identify vectorization issues and memory bottlenecks of an MPI application.
Analyze Performance Remotely and Visualize Results on a Local macOS* System.
Learn how to analyze your application performance on a remote system using the Intel Advisor command line interface (CLI) and visualize the results on a local system like macOS using the Intel Advisor GUI.
Analyze Performance in Amazon Web Services* (AWS*) EC2* Instances.
Learn how to set up Intel Advisor to analyze performance data from applications in AWS EC2 instances.
Analyze Performance on Cray* Systems.
Learn how to set up and use Intel Advisor to interpret performance data for applications running on Cray systems.
Intel® Advisor is available as a standalone product and as part of the Intel® oneAPI Base Toolkit.
Documentation for older versions of Intel® Advisor is available for download only. For a list of available documentation downloads by product version, see these pages:
To download the Intel Advisor and try one of the recipes, visit Intel Developer Zone.
The intended audience for this cookbook is intermediate to expert Intel Advisor users; however, novice users may also benefit with prior assistance from the Get Started with Intel Advisor and follow-up assistance from the Intel Advisor User Guide.
TIP:
This cookbook is dynamic; recipes will be added on a regular basis. Please use the Analyzers Forum to communicate suggestions for new recipes.