Visible to Intel only — GUID: GUID-569CC758-E3D8-44A3-A67B-29BF18FA56ED
Legal Information
Getting Help and Support
Introduction
Check-list for OpenCL™ Optimizations
Tips and Tricks for Kernel Development
Application-Level Optimizations
Debugging OpenCL™ Kernels on Linux* OS
Performance Debugging with Intel® SDK for OpenCL™ Applications
Coding for the Intel® Architecture Processors
Why Optimizing Kernels Is Important?
Avoid Spurious Operations in Kernels
Avoid Handling Edge Conditions in Kernels
Use the Preprocessor for Constants
Prefer (32-bit) Signed Integer Data Types
Prefer Row-Wise Data Accesses
Use Built-In Functions
Avoid Extracting Vector Components
Task-Parallel Programming Model Hints
Common Mistakes in OpenCL™ Applications
Introduction for OpenCL™ Coding on Intel® Architecture Processors
Vectorization Basics for Intel® Architecture Processors
Vectorization: SIMD Processing Within a Work Group
Benefitting from Implicit Vectorization
Vectorizer Knobs
Targeting a Different CPU Architecture
Using Vector Data Types
Writing Kernels to Directly Target the Intel® Architecture Processors
Work-Group Size Considerations
Threading: Achieving Work-Group Level Parallelism
Efficient Data Layout
Using the Blocking Technique
Intel® Turbo Boost Technology Support
Global Memory Size
Visible to Intel only — GUID: GUID-569CC758-E3D8-44A3-A67B-29BF18FA56ED
Enabling Debugging in OpenCL™ CPU Compiler and Runtime
To enable the debugging mode in the OpenCL™ CPU Compiler and Runtime, specific options should be passed to the build options string parameter in the clBuildProgram function:
- -g flag enables source-level debug information generation.
NOTE:Passing a -g flag means that no optimizations (such as inlining, unrolling, vectorization) are performed, the same as if a -cl-opt-disable option was passed.
-s /full/path/to/OpenCL/source/file.cl option specifies the full path to the OpenCL™ source file. If the path includes spaces, the entire path should be enclosed with double or single quotes.
This option is required in cases where the source of an OpenCL™ kernel is passed to the clCreateProgramWithSource function as a string.
Only one file can be specified with the -s option. If the file contains one or more #include directives, multiple paths for files with the -s option are not needed.
Consider the following example of using the clBuildProgram function:
err = clBuildProgram(
g_program,
0,
NULL,
"-g -s \"<path_to_opencl_source_file>\"",
NULL,
NULL);
NOTE:
The OpenCL™ kernel code must exist in a text file that is separate from the host code. Debugging OpenCL™ code that appears only in a string embedded in the host application is not supported.
Instead of passing the -s option to specify a path to an OpenCL source file, you can use one of the following approaches:
Use an Intel® SDK for OpenCL™ Applications - Offline Compiler.
Specifying the -s option is not necessary, as the path to an OpenCL source file is defined by the -input option:
ioc -cmd=build -input=kernel.cl '-bo=-g'
Pass a string with one or more #include directives to clCreateProgramWithSource.
In this case, the paths to included files are detected automatically:
const char* src = "#include kernel.cl"; cl_program program = clCreateProgramWithSource( m_context, 1, &src, NULL, &error);