https://github.com/plasma-umass/scalene Skip to content Toggle navigation Sign up * Product + Actions Automate any workflow + Packages Host and manage packages + Security Find and fix vulnerabilities + Codespaces Instant dev environments + Copilot Write better code with AI + Code review Manage code changes + Issues Plan and track work + Discussions Collaborate outside of code Explore + All features + Documentation + GitHub Skills + Blog * Solutions For + Enterprise + Teams + Startups + Education By Solution + CI/CD & Automation + DevOps + DevSecOps Case Studies + Customer Stories + Resources * Open Source + GitHub Sponsors Fund open source developers + The ReadME Project GitHub community articles Repositories + Topics + Trending + Collections * Pricing [ ] * # In this repository All GitHub | Jump to | * No suggested jump to results * # In this repository All GitHub | Jump to | * # In this organization All GitHub | Jump to | * # In this repository All GitHub | Jump to | Sign in Sign up You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session. You switched accounts on another tab or window. Reload to refresh your session. {{ message }} plasma-umass / scalene Public * * Notifications * Fork 255 * Star 7.8k Scalene: a high-performance, high-precision CPU, GPU, and memory profiler for Python with AI-powered optimization proposals License Apache-2.0 license 7.8k stars 255 forks Star Notifications * Code * Issues 138 * Pull requests 5 * Discussions * Actions * Projects 0 * Security * Insights More * Code * Issues * Pull requests * Discussions * Actions * Projects * Security * Insights plasma-umass/scalene This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository. master Switch branches/tags [ ] Branches Tags Could not load branches Nothing to show {{ refName }} default View all branches Could not load tags Nothing to show {{ refName }} default View all tags Name already in use A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch? Cancel Create 53 branches 44 tags Code * Local * Codespaces * Clone HTTPS GitHub CLI [https://github.com/p] Use Git or checkout with SVN using the web URL. [gh repo clone plasma] Work fast with our official CLI. Learn more about the CLI. * Open with GitHub Desktop * Download ZIP Sign In Required Please sign in to use Codespaces. Launching GitHub Desktop If nothing happens, download GitHub Desktop and try again. Launching GitHub Desktop If nothing happens, download GitHub Desktop and try again. Launching Xcode If nothing happens, download Xcode and try again. Launching Visual Studio Code Your codespace will open once ready. There was a problem preparing your codespace, please try again. Latest commit @emeryberger emeryberger More details re: Windows. ... 4e8bd19 Jun 14, 2023 More details re: Windows. 4e8bd19 Git stats * 2,967 commits Files Permalink Failed to load latest commit information. Type Name Latest commit message Commit time .github Added wrapper script for some smoketests that checks for presence and... June 5, 2023 17:33 .vscode Added Makefile config. November 24, 2021 14:57 benchmarks Added updated expensive benchmarks December 11, 2022 21:01 docs Removed. October 31, 2022 14:46 scalene Fixes issue (#623) - couldn't reproduce but better safe than sorry. June 8, 2023 15:55 src Disabled possibly spurious assert. June 4, 2023 18:18 test Fix smoketests (#621) June 6, 2023 13:00 .gitignore Avoid including .vscode & generated libraries. November 16, 2021 19:19 CITATION.cff Updated an OrcID. July 28, 2022 17:26 CODE_OF_CONDUCT.md Create CODE_OF_CONDUCT.md July 24, 2020 00:22 GNUmakefile - adjusting build platform strings for building on MacOS >= 11; June 7, 2023 12:09 LICENSE Initial check-in. December 17, 2019 15:57 MANIFEST.in Added crdp. December 8, 2022 13:57 Makefile Added isort. August 3, 2022 16:01 Pipfile Restored wheel. November 15, 2022 11:53 Pipfile.lock Updated. February 23, 2023 16:18 README.md More details re: Windows. June 14, 2023 16:46 mypy.ini Disable unreachable warning, for now. February 28, 2021 13:17 pyrightconfig.json Added. December 4, 2021 16:27 pytest.ini Add pytest.ini to run tests successfuly February 15, 2020 23:56 requirements.txt Fix smoketests (#621) June 6, 2023 13:00 setup.py Updated cloudpickle in setupe June 7, 2023 18:31 View code [ ] Scalene: a Python CPU+GPU+memory profiler with AI-powered optimization proposals About Scalene Quick Start Installing Scalene: Using Scalene: Web-based GUI Scalene Overview Scalene talk (PyCon US 2021) Fast and Accurate CPU profiling GPU profiling Memory profiling Other features Comparison to Other Profilers Performance and Features Output Scalene Scalene with Jupyter Installation Asked Questions Technical Information Success Stories Acknowledgements README.md scalene Scalene: a Python CPU+GPU+memory profiler with AI-powered optimization proposals by Emery Berger, Sam Stern, and Juan Altmayer Pizzorno. Scalene community SlackScalene community Slack PyPI Latest ReleaseAnaconda-Server BadgeDownloads Downloads Python versions License Ozsvald tweet (tweet from Ian Ozsvald, author of High Performance Python) Semantic Scholar success story About Scalene Scalene is a high-performance CPU, GPU and memory profiler for Python that does a number of things that other Python profilers do not and cannot do. It runs orders of magnitude faster than many other profilers while delivering far more detailed information. It is also the first profiler ever to incorporate AI-powered proposed optimizations. To enable these, you need to enter an OpenAI key: Scalene advanced options Once a valid key is entered, click on the lightning bolt ([?]) beside any line or the explosion () for an entire region of code to generate a proposed optimization. Click on a proposed optimization to copy it to the clipboard. example proposed optimization You can click as many times as you like on the lightning bolt or explosion, and it will generate different suggested optimizations. Your mileage may vary, but in some cases, the suggestions are quite impressive (e.g., order-of-magnitude improvements). Quick Start Installing Scalene: python3 -m pip install -U scalene or conda install -c conda-forge scalene Using Scalene: Commonly used options: scalene your_prog.py # full profile (outputs to web interface) python3 -m scalene your_prog.py # equivalent alternative scalene --cli your_prog.py # use the command-line only (no web interface) scalene --cpu your_prog.py # only profile CPU scalene --cpu --gpu your_prog.py # only profile CPU and GPU scalene --cpu --gpu --memory your_prog.py # profile everything (same as no options) scalene --reduced-profile your_prog.py # only profile lines with significant usage scalene --profile-interval 5.0 your_prog.py # output a new profile every five seconds scalene (Scalene options) --- your_prog.py (...) # use --- to tell Scalene to ignore options after that point scalene --help # lists all options Using Scalene programmatically in your code: Invoke using scalene as above and then: from scalene import scalene_profiler # Turn profiling on scalene_profiler.start() # Turn profiling off scalene_profiler.stop() Using Scalene to profile only specific functions via @profile: Just preface any functions you want to profile with the @profile decorator and run it with Scalene: # do not import profile! @profile def slow_function(): import time time.sleep(3) Web-based GUI Scalene has both a CLI and a web-based GUI (demo here). By default, once Scalene has profiled your program, it will open a tab in a web browser with an interactive user interface (all processing is done locally). Hover over bars to see breakdowns of CPU and memory consumption, and click on underlined column headers to sort the columns. The generated file profile.html is self-contained and can be saved for later use. Scalene web GUI Scalene Overview Scalene talk (PyCon US 2021) This talk presented at PyCon 2021 walks through Scalene's advantages and how to use it to debug the performance of an application (and provides some technical details on its internals). We highly recommend watching this video! Scalene presentation at PyCon 2021 Fast and Accurate * Scalene is fast. It uses sampling instead of instrumentation or relying on Python's tracing facilities. Its overhead is typically no more than 10-20% (and often less). * Scalene is accurate. We tested CPU profiler accuracy and found that Scalene is among the most accurate profilers, correctly measuring time taken. Profiler accuracy * Scalene performs profiling at the line level and per function, pointing to the functions and the specific lines of code responsible for the execution time in your program. CPU profiling * Scalene separates out time spent in Python from time in native code (including libraries). Most Python programmers aren't going to optimize the performance of native code (which is usually either in the Python implementation or external libraries), so this helps developers focus their optimization efforts on the code they can actually improve. * Scalene highlights hotspots (code accounting for significant percentages of CPU time or memory allocation) in red, making them even easier to spot. * Scalene also separates out system time, making it easy to find I/ O bottlenecks. GPU profiling * Scalene reports GPU time (currently limited to NVIDIA-based systems). Memory profiling * Scalene profiles memory usage. In addition to tracking CPU usage, Scalene also points to the specific lines of code responsible for memory growth. It accomplishes this via an included specialized memory allocator. * Scalene separates out the percentage of memory consumed by Python code vs. native code. * Scalene produces per-line memory profiles. * Scalene identifies lines with likely memory leaks. * Scalene profiles copying volume, making it easy to spot inadvertent copying, especially due to crossing Python/library boundaries (e.g., accidentally converting numpy arrays into Python arrays, and vice versa). Other features * Scalene can produce reduced profiles (via --reduced-profile) that only report lines that consume more than 1% of CPU or perform at least 100 allocations. * Scalene supports @profile decorators to profile only specific functions. * When Scalene is profiling a program launched in the background (via &), you can suspend and resume profiling. Comparison to Other Profilers Performance and Features Below is a table comparing the performance and features of various profilers to Scalene. Performance and feature comparison * Slowdown: the slowdown when running a benchmark from the Pyperformance suite. Green means less than 2x overhead. Scalene's overhead is just a 35% slowdown. Scalene has all of the following features, many of which only Scalene supports: * Lines or functions: does the profiler report information only for entire functions, or for every line -- Scalene does both. * Unmodified Code: works on unmodified code. * Threads: supports Python threads. * Multiprocessing: supports use of the multiprocessing library -- Scalene only * Python vs. C time: breaks out time spent in Python vs. native code (e.g., libraries) -- Scalene only * System time: breaks out system time (e.g., sleeping or performing I/O) -- Scalene only * Profiles memory: reports memory consumption per line / function * GPU: reports time spent on an NVIDIA GPU (if present) -- Scalene only * Memory trends: reports memory use over time per line / function -- Scalene only * Copy volume: reports megabytes being copied per second -- Scalene only * Detects leaks: automatically pinpoints lines responsible for likely memory leaks -- Scalene only Output If you include the --cli option, Scalene prints annotated source code for the program being profiled (as text, JSON (--json), or HTML (--html)) and any modules it uses in the same directory or subdirectories (you can optionally have it --profile-all and only include files with at least a --cpu-percent-threshold of time). Here is a snippet from pystone.py. Example profile * Memory usage at the top: Visualized by "sparklines", memory consumption over the runtime of the profiled code. * "Time Python": How much time was spent in Python code. * "native": How much time was spent in non-Python code (e.g., libraries written in C/C++). * "system": How much time was spent in the system (e.g., I/O). * "GPU": (not shown here) How much time spent on the GPU, if your system has an NVIDIA GPU installed. * "Memory Python": How much of the memory allocation happened on the Python side of the code, as opposed to in non-Python code (e.g., libraries written in C/C++). * "net": Positive net memory numbers indicate total memory allocation in megabytes; negative net memory numbers indicate memory reclamation. * "timeline / %": Visualized by "sparklines", memory consumption generated by this line over the program runtime, and the percentages of total memory activity this line represents. * "Copy (MB/s)": The amount of megabytes being copied per second (see "About Scalene"). Scalene The following command runs Scalene on a provided example program. scalene test/testme.py Click to see all Scalene's options (available by running with --help) % scalene --help usage: scalene [-h] [--outfile OUTFILE] [--html] [--reduced-profile] [--profile-interval PROFILE_INTERVAL] [--cpu-only] [--profile-all] [--profile-only PROFILE_ONLY] [--use-virtual-time] [--cpu-percent-threshold CPU_PERCENT_THRESHOLD] [--cpu-sampling-rate CPU_SAMPLING_RATE] [--malloc-threshold MALLOC_THRESHOLD] Scalene: a high-precision CPU and memory profiler. https://github.com/plasma-umass/scalene command-line: % scalene [options] yourprogram.py or % python3 -m scalene [options] yourprogram.py in Jupyter, line mode: %scrun [options] statement in Jupyter, cell mode: %%scalene [options] code... code... optional arguments: -h, --help show this help message and exit --outfile OUTFILE file to hold profiler output (default: stdout) --html output as HTML (default: text) --reduced-profile generate a reduced profile, with non-zero lines only (default: False) --profile-interval PROFILE_INTERVAL output profiles every so many seconds (default: inf) --cpu-only only profile CPU time (default: profile CPU, memory, and copying) --profile-all profile all executed code, not just the target program (default: only the target program) --profile-only PROFILE_ONLY profile only code in filenames that contain the given strings, separated by commas (default: no restrictions) --use-virtual-time measure only CPU time, not time spent in I/O or blocking (default: False) --cpu-percent-threshold CPU_PERCENT_THRESHOLD only report profiles with at least this percent of CPU time (default: 1%) --cpu-sampling-rate CPU_SAMPLING_RATE CPU sampling rate (default: every 0.01s) --malloc-threshold MALLOC_THRESHOLD only report profiles with at least this many allocations (default: 100) When running Scalene in the background, you can suspend/resume profiling for the process ID that Scalene reports. For example: % python3 -m scalene [options] yourprogram.py & Scalene now profiling process 12345 to suspend profiling: python3 -m scalene.profile --off --pid 12345 to resume profiling: python3 -m scalene.profile --on --pid 12345 Scalene with Jupyter Instructions for installing and using Scalene with Jupyter notebooks This notebook illustrates the use of Scalene in Jupyter. Installation: !pip install scalene %load_ext scalene Line mode: %scrun [options] statement Cell mode: %%scalene [options] code... code... Installation Using pip (Mac OS X, Linux, Windows, and WSL2) Scalene is distributed as a pip package and works on Mac OS X, Linux (including Ubuntu in Windows WSL2) and (with limitations) Windows platforms. (Note: the Windows version isn't yet complete; it currently only supports CPU and GPU profiling, but not memory profiling.) You can install it as follows: % pip install -U scalene or % python3 -m pip install -U scalene You may need to install some packages first. See https://stackoverflow.com/a/19344978/4954434 for full instructions for all Linux flavors. For Ubuntu/Debian: % sudo apt install git python3-all-dev Using Homebrew (Mac OS X) As an alternative to pip, you can use Homebrew to install the current version of Scalene from this repository: % brew tap plasma-umass/scalene % brew install --head plasma-umass/scalene/scalene On ArchLinux You can install Scalene on Arch Linux via the AUR package. Use your favorite AUR helper, or manually download the PKGBUILD and run makepkg -cirs to build. Note that this will place libscalene.so in / usr/lib; modify the below usage instructions accordingly. Asked Questions Can I use Scalene with PyTest? A: Yes! You can run it as follows (for example): python3 -m scalene --- -m pytest your_test.py Is there any way to get shorter profiles or do more targeted profiling? A: Yes! There are several options: 1. Use --reduced-profile to include only lines and files with memory /CPU/GPU activity. 2. Use --profile-only to include only filenames containing specific strings (as in, --profile-only foo,bar,baz). 3. Decorate functions of interest with @profile to have Scalene report only those functions. 4. Turn profiling on and off programmatically by importing Scalene (import scalene) and then turning profiling on and off via scalene_profiler.start() and scalene_profiler.stop(). By default, Scalene runs with profiling on, so to delay profiling until desired, use the --off command-line option (python3 -m scalene --off yourprogram.py). How do I run Scalene in PyCharm? A: In PyCharm, you can run Scalene at the command line by opening the terminal at the bottom of the IDE and running a Scalene command (e.g., python -m scalene ). Use the options --cli, --html, and --outfile to generate an HTML file that you can then view in the IDE. How do I use Scalene with Django? A: Pass in the --noreload option (see #178). How do I use Scalene with PyTorch on the Mac? A: Scalene works with PyTorch version 1.5.1 on Mac OS X. There's a bug in newer versions of PyTorch (pytorch/pytorch#57185) that interferes with Scalene (discussion here: #110), but only on Macs. Technical Information For details about how Scalene works, please see the following paper, accepted to appear at OSDI 2023: Triangulating Python Performance Issues with Scalene. Note that this paper currently does not include information about the AI-driven proposed optimizations. Success Stories If you use Scalene to successfully debug a performance problem, please add a comment to this issue! Acknowledgements Logo created by Sophia Berger. This material is based upon work supported by the National Science Foundation under Grant No. 1955610. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation. About Scalene: a high-performance, high-precision CPU, GPU, and memory profiler for Python with AI-powered optimization proposals Topics python cpu profiler gpu performance-analysis memory-allocation profiling cpu-profiling memory-consumption gpu-programming python-profilers scalene profiles-memory performance-cpu Resources Readme License Apache-2.0 license Code of conduct Code of conduct Stars 7.8k stars Watchers 70 watching Forks 255 forks Report repository Releases 39 Maintenance Release: fixes MacOS wheel tags, updates cloudpickle Latest Jun 7, 2023 + 38 releases Sponsor this project * * Learn more about GitHub Sponsors Packages 0 No packages published Used by 452 * @sthagen * @rmaniego * @ryantam626 * @fwerf43klld * @devsetgo * @jacopok * @Cainuse * @parantapa + 444 Contributors 38 * @emeryberger * @jaltmayerpizzorno * @sternj * @dependabot[bot] * @Insutanto * @johanvergeer * @groutr * @SnoopJ * @cclauss * @RuRo + 27 contributors Languages * JavaScript 46.9% * Python 45.1% * C++ 6.5% * Other 1.5% Footer (c) 2023 GitHub, Inc. Footer navigation * Terms * Privacy * Security * Status * Docs * Contact GitHub * Pricing * API * Training * Blog * About You can't perform that action at this time.