AMD APP KernelAnalyzer has been succeeded by AMD’s CodeXL. With the exception of critical bug fixes, there will be no new versions of APP KernelAnalyzer. For new features, continuous support and a unified developer tools suite, please visit AMD’s CodeXL web page and check out the CodeXL Forum.
AMD APP KernelAnalyzer is a static analysis tool to compile, analyze and disassemble an OpenCL kernel for AMD GPUs. It gives you accurate performance estimates for your kernels and even allows you to view disassembly of the generated hardware kernel, all without having to run the application on actual hardware.
It can be used as a GUI tool for interactive tuning of an OpenCL kernel or in command line mode to generate detailed reports.
What’s New in Version 1.12
- Support for Catalyst™ driver 12.1-12.4
- Support for AMD APP SDK 2.7.
- Compile, analyze and disassemble the OpenCL kernel for multiple Catalyst driver versions and GPU device targets.
- View any kernel compilation errors and warnings generated by the OpenCL runtime.
- View the ISA code generated by the AMD Shader Compiler.
- View various statistics generated by analyzing the ISA code.
- View General Purpose Registers and spill registers allocated for the kernel.
Windows® Vista or 7. AMD APP SDK v2.5 or greater is required for OpenCL™ support.
Enter your kernel into the Source Code window. You can load it via the file menu, drag and drop it onto the window or type it in from scratch if you prefer. Hit Compile to generate statistics and view object code for the shader. If your input is an OpenCL, then you may need to select which kernel to work with. If your kernel is compiled successfully, you should see the disassembly of it in the Object Code window. The Compiler Statistics window will display statistics about your kernel. If you wish to see disassembly of the hardware shader as generated by the Shader Compiler you may select the AMD graphics card you’re interested in from the Format list.