Legal Information
Getting Help and Support
Introduction
Coding for the Intel® Processor Graphics
Platform-Level Considerations
Application-Level Optimizations
Optimizing OpenCL™ Usage with Intel® Processor Graphics
Check-list for OpenCL™ Optimizations
Performance Debugging
Using Multiple OpenCL™ Devices
Coding for the Intel® CPU OpenCL™ Device
OpenCL™ Kernel Development for Intel® CPU OpenCL™ device
Mapping Memory Objects
Using Buffers and Images Appropriately
Using Floating Point for Calculations
Using Compiler Options for Optimizations
Using Built-In Functions
Loading and Storing Data in Greatest Chunks
Applying Shared Local Memory
Using Specialization in Branching
Considering native_ and half_ Versions of Math Built-Ins
Using the Restrict Qualifier for Kernel Arguments
Avoiding Handling Edge Conditions in Kernels
Using Shared Context for Multiple OpenCL™ Devices
Sharing Resources Efficiently
Synchronization Caveats
Writing to a Shared Resource
Partitioning the Work
Keeping Kernel Sources the Same
Basic Frequency Considerations
Eliminating Device Starvation
Limitations of Shared Context with Respect to Extensions
Why Optimizing Kernel Code Is Important?
Avoid Spurious Operations in Kernel Code
Perform Initialization in a Separate Task
Use Preprocessor for Constants
Use Signed Integer Data Types
Use Row-Wise Data Accesses
Tips for Auto-Vectorization
Local Memory Usage
Avoid Extracting Vector Components
Task-Parallel Programming Model Hints
Vectorizer Knobs
Several environment variables are related to vectorizer. The first one is CL_CONFIG_USE_VECTORIZER, which can be set to False and True respectively. Notice that just like any other environment variables this one affects the behavior of the vectorizer of the entire system (or shell instances) until variable gets unset explicitly (or shell(s) terminates). This specific variable affects code generation for Intel® Xeon Phi™ coprocessor device as well.
Another variable is CL_CONFIG_CPU_VECTORIZER_MODE (that affects code generation for CPU OpenCL device only). It effectively sets the vectorization “width” (when CL_CONFIG_USE_VECTORIZER = True):
- CL_CONFIG_CPU_VECTORIZER_MODE = 0 (default). The compiler makes heuristic decisions whether to vectorize each kernel, and if so which vector width to use.
- CL_CONFIG_CPU_VECTORIZER_MODE = 1. No vectorization by compiler. Explicit vector data types in kernels are left intact. This mode is the same as CL_CONFIG_USE_VECTORIZER = False.
- CL_CONFIG_CPU_VECTORIZER_MODE = 4. Disables heuristic and vectorizes to the width of 4.
- CL_CONFIG_CPU_VECTORIZER_MODE = 8. Disables heuristic and vectorizes to the width of 8.
NOTE:
Some kernels cannot be vectorized, so the vectorizer does not handle them, regardless the mode. Also be careful with manual overriding the compiler heuristic, build process fails if the target hardware doesn't support the specific vectorization width. Inspect the compiler output in the offline compiler tool (described in the product User’s Guide) on the messages related to vectorization.