NAME

coz - profiler running experiements on multithreaded code

SYNOPSIS

coz run [profiling options] --- <command> [args]

coz plot [-h]

DESCRIPTION

Coz is a new kind of profiler that unlocks optimization opportunities missed by traditional profilers. Coz employs a novel technique we call causal profiling that measures optimization potential. This measurement matches developers' assumptions about profilers: that optimizing highly-ranked code will have the greatest impact on performance. Causal profiling measures optimization potential for serial, parallel, and asynchronous programs without instrumentation of special handling for library calls and concurrency primitives. Instead, a causal profiler uses performance experiments to predict the effect of optimizations. This allows the profiler to establish causality: "optimizing function X will have effect Y," exactly the measurement developers had assumed they were getting all along.

OPTIONS

-h, --help: show this help message and exit
--binary-scope <filepattern>, -b <filepattern>: Profile matching executables. Use '%' as a wildcard, or 'MAIN' to include the main executable (default=MAIN)
--source-scope <filepattern>, -s <filepattern>: Profile matching source files. Use '%' as a wildcard. (default=%)

--progress <source file>:<line number>, -p <source file>:<line number>: [NOT SUPPORTED] Add a sampling-based progress point

--output <profileoutput>, -o <profileoutput>: Profiler output (default=`profile.coz`)
--end-to-end: Run a single performance experiment per-execution

--fixed-line <source file>:<line number>: Evaluate optimizations of a specific source line
--fixed-speedup <speedup> (0-100): Evaluate optimizations of a specific amount

AUTHOR

Petter Reinholdtsen - pere@hungry.com

COPYRIGHT