The pixie utility

The pixie utility

The pixie utility can be used under Linux or UNIX to get more accurate counts on the number of times each statement in a function is executed. Where the prof utility only produces estimates based on statistical sampling of the program counter at regular intervals throughout the execution of the program, pixie measures the number of times each basic block is executed. A basic block is a sequence of code containing no branches.

The pixie utility is applied to the already generated executable file. There is no need to recompile the executable with the -p option. The command for pixie is simply:

    pixie a.out

This will generate a new executable file, “a.out.pixie”, which when executed will generate a data file called “a.out.Counts”. A data file of function addresses called “a.out.Addrs” is also generated. The next step is to run the new executable:

    a.out.pixie

After execution, the count file can be examined using either prof or pixstats. One possible command is:

    pixstats a.out

The use of the prof command with the -pixie option is:

    prof -pixie a.out

Both of these commands will generate a variety of information. prof with the “-pixie” option will generate an ordering of functions based on instruction cycle counts, another based on invocations, and a list of instruction counts for each basic block. pixstats generates a whole wealth of useful information including summaries of opcode distributions and register usage. For more information refer to the Linux manual entries for pixie, pixstats and prof.

• Next: • Up: Table of Contents • Buy: Generative AI in C++: Coding Transformers and LLMs

The new AI programming book by Aussie AI co-founders: AI coding in C++ Transformer engine speedups LLM models Phone and desktop AI Code examples Research citations Get your copy from Amazon: Generative AI in C++