Skip to content

Latest commit

 

History

History
78 lines (57 loc) · 2.76 KB

README.md

File metadata and controls

78 lines (57 loc) · 2.76 KB

PProf.jl

CI

Sometimes I need a hammer, sometimes I need a drill, this is a hammer-drill

using Profile
using PProf

# Collect a profile
Profile.clear()
@profile peakflops()

# Export pprof profile and open interactive profiling web interface.
pprof()

This prints a link to a local webserver where you can inspect the profile you've collected. It produces a file called profile.pb.gz in the pprof format, and then opens the pprof tool in interactive "web" mode.

To profile allocations instead of CPU time, simply use the equivalent functions from the Allocs submodule instead:

# Collect an allocation profile
Profile.Allocs.clear()
Profile.Allocs.@profile peakflops()

# Export pprof allocation profile and open interactive profiling web interface.
PProf.Allocs.pprof()

For more usage examples see the pprof docs: https://github.com/google/pprof/blob/master/doc/README.md

Dependencies

  • Graphviz
    • In order to use pprof's web graph view (which is one of the best parts of pprof), you need to have graphviz installed. Click the above link to install.

Usage

help?> pprof
  pprof([data, [lidict]];
          web = true, webhost = "localhost", webport = 57599,
          out = "profile.pb.gz", from_c = true, full_signatures = true, drop_frames = "",
          keep_frames = "", ui_relative_percentages = true, sampling_delay = nothing,
        )
  pprof(FlameGraphs.flamegraph(); kwargs...)

  Fetches the collected `Profile` data, exports to the `pprof` format, and (optionally) opens
  a `pprof` web-server for interactively viewing the results.
help?> @pprof
  @pprof ex

  Profiles the expression using @profile and starts or restarts the pprof() web UI with default arguments.

Calling pprof() or @pprof exports the profiling results to disk, and launches the web server in the background pointed at the results. Calling it again overwrites the results and refreshes the server, so you can reload the web-page to see your updated results. The link to the webserver is printed to your terminal, which you can then open in a browser.

If you manually modify the output profile file, you can refresh the web server without overwriting the out file by calling PProf.refresh().

Example

julia> using PProf

julia> @pprof peakflops()
"profile.pb.gz"

Main binary filename not available.
Serving web UI on http://localhost:57599

Google PProf Web View

graph

flamegraph