Chapter 12. OpenCL Profiling and Debugging
In this chapter, we examine aspects of profiling and debugging OpenCL programs. OpenCL tools such as the Accelerated Parallel Processing Profiler and the KernelAnalyzer help us to understand the location of performance bottlenecks in our code. gDEBugger gives us a high-level debugging interface above kernel-level options such as inserting calls to the printf function.
Keywords: debug, profile, performance analysis, gDEBugger, KernelAnalyzer, performance counter
Introduction
Our motivation for writing programs in OpenCL is not limited to writing isolated high-performance kernels but to speed up parallel applications. Previous chapters discussed how we can optimize kernels running on OpenCL devices by targeting ...
Get Heterogeneous Computing with OpenCL now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
