pprofile - Line-granularity, thread-aware deterministic and statistic pure-python profiler
Contents
Description
This manual page documents briefly the pprofile3 command.
pprofile is a Line-granularity, thread-aware deterministic and statistic pure-python profiler
Line-granularity, thread-aware deterministic and statistic pure-python profiler. Line granularity allows
locating precisely where time is spent in code. Thread awareness automatically propagates profiling to
all threads (all threads in statistic mode, or threads spawned by profiled code in deterministic mode).
Deterministic profiling gives precise measures, but at a large speed cost (best used on minimal test
scenario). Statistic profiling gives rough measure, but has an extremely low overhead (suitable for live
code profiling). Does not require marking methods to profile, allowing non-method profiling (module
imports, class & function declarations and other module-level code). Can produce callgrind output.
Name
pprofile - Line-granularity, thread-aware deterministic and statistic pure-python profiler
Options
-h,--help
Show summary of options.
-mMODULE
Searches sys.path for the named module and runs the corresponding .py file as a
script. When given, positional argumentsbecome sys.argv[1:].
-o,--outOUT
Write annotated sources to this file. Defaults to stdout.
-z,--zipfileZIPFILE
Name of a zip file to generate from all involved source files. Useful with
callgrind output.
-t,--threadsTHREADS
If non-zero, trace threads spawned by program. Default: 1.
-f,--formatcallgrind,text
Format in which output is generated. If not set, auto-detected from filename if provided, falling
back to "text".
-v,--verbose
Enable profiler internal tracing output. Cryptic and verbose.
-s,--statisticSTATISTIC
Use this period for statistic profiling, or use deterministic profiling when 0.
Synopsis
pprofile3 [options] some-python-module [arguments]
