Visible to Intel only — GUID: GUID-D4A42041-A52F-417E-973D-22BE83B8E17D
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
Visible to Intel only — GUID: GUID-D4A42041-A52F-417E-973D-22BE83B8E17D
Use Row-Wise Data Accesses
OpenCL™ enables you to submit kernels on one-, two- or three-dimensional index space. Consider using one-dimensional ranges for cache locality and to save index computations.
If a two- or three-dimensional range naturally fits your data dimensions, try to keep work-items scanning along rows, not columns. For example:
__kernel void smooth(const __global float* input,
uint image_width, uint image_height,
__global float* output)
{
int myX = get_global_id(
0);
int myY = get_global_id(
1);
int myPixel = myY * image_width + myX;
float data = input[myPixel];
…
}
In the example above, the first dimension is the image width and the second is the image height. The following code is less effective:
__kernel void smooth(const __global float* input,
uint image_width, uint image_height,
__global float* output)
{
int myY = get_global_id(
0);
int myX = get_global_id(
1);
int myPixel = myY * image_width + myX;
float data = input[myPixel];
…
}
In the second code example, the image height is the first dimension and the image width is the second dimension. The resulting column-wise data access is inefficient, since CPU OpenCL™ framework initially iterates over the first dimension.
The same rule applies if each work-item calculates several elements. To optimize performance, make sure work-items read from consecutive memory addresses.