Skip to main content

A memory profiler for data batch processing applications.

Project description

The Fil memory profiler for Python

Your code reads some data, processes it, and uses too much memory. In order to reduce memory usage, you need to figure out:

  1. Where peak memory usage is, also known as the high-water mark.
  2. What code was responsible for allocating the memory that was present at that peak moment.

That's exactly what Fil will help you find. Fil an open source memory profiler designed for data processing applications written in Python, and includes native support for Jupyter.

At the moment it only runs on Linux and macOS, and while it supports threading, it does not yet support multiprocessing or multiple processes in general.

"Within minutes of using your tool, I was able to identify a major memory bottleneck that I never would have thought existed. The ability to track memory allocated via the Python interface and also C allocation is awesome, especially for my NumPy / Pandas programs."

—Derrick Kondo

For more information, including an example of the output, see https://pythonspeed.com/products/filmemoryprofiler/

Fil vs. other Python memory tools

There are two distinct patterns of Python usage, each with its own source of memory problems.

In a long-running server, memory usage can grow indefinitely due to memory leaks. That is, some memory is not being freed.

  • If the issue is in Python code, tools like tracemalloc and Pympler can tell you which objects are leaking and what is preventing them from being leaked.
  • If you're leaking memory in C code, you can use tools like Valgrind.

Fil, however, is not aimed at memory leaks, but at the other use case: data processing applications. These applications load in data, process it somehow, and then finish running.

The problem with these applications is that they can, on purpose or by mistake, allocate huge amounts of memory. It might get freed soon after, but if you allocate 16GB RAM and only have 8GB in your computer, the lack of leaks doesn't help you.

Fil will therefore tell you, in an easy to understand way:

  1. Where peak memory usage is, also known as the high-water mark.
  2. What code was responsible for allocating the memory that was present at that peak moment.
  3. This includes C/Fortran/C++/whatever extensions that don't use Python's memory allocation API (tracemalloc only does Python memory APIs).

This allows you to optimize that code in a variety of ways.

Installation

Assuming you're on macOS or Linux, and are using Python 3.6 or later, you can use either Conda or pip (or any tool that is pip-compatible and can install manylinux2010 wheels).

Conda

To install on Conda:

$ conda install -c conda-forge filprofiler

Pip

To install the latest version of Fil you'll need Pip 19 or newer. You can check like this:

$ pip --version
pip 19.3.0

If you're using something older than v19, you can upgrade by doing:

$ pip install --upgrade pip

If that doesn't work, try running that in a virtualenv.

Assuming you have a new enough version of pip:

$ pip install filprofiler

Using Fil

Measuring peak (high-water mark) memory usage in Jupyter

To measure memory usage of some code in Jupyter you need to do three things:

  1. Use an alternative kernel, "Python 3 with Fil". You can choose this kernel when you create a new notebook, or you can switch an existing notebook in the Kernel menu; there should be a "Change Kernel" option in there in both Jupyter Notebook and JupyterLab.
  2. Load the extension by doing %load_ext filprofiler.
  3. Add the %%filprofile magic to the top of the cell with the code you wish to profile.

Screenshot of JupyterLab

Measuring peak (high-water mark) memory usage for Python scripts

Instead of doing:

$ python yourscript.py --input-file=yourfile

Just do:

$ fil-profile run yourscript.py --input-file=yourfile

And it will generate a report.

As of version 0.11, you can also run it like this:

$ python -m filprofiler run yourscript.py --input-file=yourfile

Debugging out-of-memory crashes

First, run free to figure out how much memory is available—in this case about 6.3GB—and then set a corresponding limit on virtual memory with ulimit:

$ free -h
       total   used   free  shared  buff/cache  available
Mem:   7.7Gi  1.1Gi  6.3Gi    50Mi       334Mi      6.3Gi
Swap:  3.9Gi  3.0Gi  871Mi
$ ulimit -Sv 6300000

Then, run your program under Fil, and it will generate a SVG at the point in time when memory runs out:

$ fil-profile run oom.py 
...
=fil-profile= Wrote memory usage flamegraph to fil-result/2020-06-15T12:37:13.033/out-of-memory.svg

Reducing memory usage in your code

You've found where memory usage is coming from—now what?

If you're using data processing or scientific computing libraries, I have written a relevant guide to reducing memory usage.

How Fil works

Fil uses the LD_PRELOAD/DYLD_INSERT_LIBRARIES mechanism to preload a shared library at process startup. This shared library captures all memory allocations and deallocations and keeps track of them.

At the same time, the Python tracing infrastructure (used e.g. by cProfile and coverage.py) to figure out which Python callstack/backtrace is responsible for each allocation.

For performance reasons, only the largest allocations are reported, with a minimum of 99% of allocated memory reported. The remaining <1% is highly unlikely to be relevant when trying to reduce usage; it's effectively noise.

Fil and threading, with notes on NumPy and Zarr {#threading}

In general, Fil will track allocations in threads correctly.

First, if you start a thread via Python, running Python code, that thread will get its own callstack for tracking who is responsible for a memory allocation.

Second, if you start a C thread, the calling Python code is considered responsible for any memory allocations in that thread. This works fine... except for thread pools. If you start a pool of threads that are not Python threads, the Python code that created those threads will be responsible for all allocations created during the thread pool's lifetime.

Therefore, in order to ensure correct memory tracking, Fil disables thread pools in BLAS (used by NumPy), BLOSC (used e.g. by Zarr), OpenMP, and numexpr. They are all set to use 1 thread, so calls should run in the calling Python thread and everything should be tracked correctly.

This has some costs:

  1. This can reduce performance in some cases, since you're doing computation with one CPU instead of many.
  2. Insofar as these libraries allocate memory proportional to number of threads, the measured memory usage might be wrong.

Fil does this for the whole program when using fil-profile run. When using the Jupyter kernel, anything run with the %%filprofile magic will have thread pools disabled, but other code should run normally.

What Fil tracks

Fil will track memory allocated by:

  • Normal Python code.
  • C code using malloc()/calloc()/realloc()/posix_memalign().
  • C++ code using new (including via aligned_alloc()).
  • Anonymous mmap()s.
  • Fortran 90 explicitly allocated memory (tested with gcc's gfortran).

Still not supported, but planned:

  • mremap() (resizing of mmap()).
  • File-backed mmap(). The semantics are somewhat different than normal allocations or anonymous mmap(), since the OS can swap it in or out from disk transparently, so supporting this will involve a different kind of resource usage and reporting.
  • Other forms of shared memory, need to investigate if any of them allow sufficient allocation.
  • Anonymous mmap()s created via /dev/zero (not common, since it's not cross-platform, e.g. macOS doesn't support this).
  • memfd_create(), a Linux-only mechanism for creating in-memory files.
  • Possibly memalign, valloc(), pvalloc(), reallocarray(). These are all rarely used, as far as I can tell.

License

Copyright 2020 Hyphenated Enterprises LLC

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at

 http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distributions

No source distribution files available for this release.See tutorial on generating distribution archives.

Built Distributions

filprofiler-0.13.0-cp38-cp38-manylinux2010_x86_64.whl (1.8 MB view details)

Uploaded CPython 3.8manylinux: glibc 2.12+ x86-64

filprofiler-0.13.0-cp38-cp38-macosx_10_14_x86_64.whl (311.5 kB view details)

Uploaded CPython 3.8macOS 10.14+ x86-64

filprofiler-0.13.0-cp37-cp37m-manylinux2010_x86_64.whl (1.8 MB view details)

Uploaded CPython 3.7mmanylinux: glibc 2.12+ x86-64

filprofiler-0.13.0-cp37-cp37m-macosx_10_14_x86_64.whl (311.5 kB view details)

Uploaded CPython 3.7mmacOS 10.14+ x86-64

filprofiler-0.13.0-cp36-cp36m-manylinux2010_x86_64.whl (1.8 MB view details)

Uploaded CPython 3.6mmanylinux: glibc 2.12+ x86-64

filprofiler-0.13.0-cp36-cp36m-macosx_10_14_x86_64.whl (311.5 kB view details)

Uploaded CPython 3.6mmacOS 10.14+ x86-64

File details

Details for the file filprofiler-0.13.0-cp38-cp38-manylinux2010_x86_64.whl.

File metadata

  • Download URL: filprofiler-0.13.0-cp38-cp38-manylinux2010_x86_64.whl
  • Upload date:
  • Size: 1.8 MB
  • Tags: CPython 3.8, manylinux: glibc 2.12+ x86-64
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.2.0 pkginfo/1.6.1 requests/2.25.0 setuptools/49.2.1 requests-toolbelt/0.9.1 tqdm/4.54.0 CPython/3.8.6

File hashes

Hashes for filprofiler-0.13.0-cp38-cp38-manylinux2010_x86_64.whl
Algorithm Hash digest
SHA256 f9fc190ba1df112fd60c69db705ac52e6ad82eee827fc2cfaa1bb0e1b2d29756
MD5 6a6b5bbe2ea07a1a918c677747baf74b
BLAKE2b-256 66d34279a42314f457726b9820457135e8032e75f7758c42db759fac8ba4e3ae

See more details on using hashes here.

File details

Details for the file filprofiler-0.13.0-cp38-cp38-macosx_10_14_x86_64.whl.

File metadata

  • Download URL: filprofiler-0.13.0-cp38-cp38-macosx_10_14_x86_64.whl
  • Upload date:
  • Size: 311.5 kB
  • Tags: CPython 3.8, macOS 10.14+ x86-64
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.2.0 pkginfo/1.6.1 requests/2.25.0 setuptools/49.2.1 requests-toolbelt/0.9.1 tqdm/4.54.0 CPython/3.8.6

File hashes

Hashes for filprofiler-0.13.0-cp38-cp38-macosx_10_14_x86_64.whl
Algorithm Hash digest
SHA256 95e29180bf81fa38ffcc7517d6cc4f64ae251b29bdb177d7a69447074b109a06
MD5 9e95acda0e8404cddfda85232dfbfb28
BLAKE2b-256 1b49a96f114bd4602cc335e5ea85a1b79c2e81a73494cf153057dfa1271ed506

See more details on using hashes here.

File details

Details for the file filprofiler-0.13.0-cp37-cp37m-manylinux2010_x86_64.whl.

File metadata

  • Download URL: filprofiler-0.13.0-cp37-cp37m-manylinux2010_x86_64.whl
  • Upload date:
  • Size: 1.8 MB
  • Tags: CPython 3.7m, manylinux: glibc 2.12+ x86-64
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.2.0 pkginfo/1.6.1 requests/2.25.0 setuptools/49.2.1 requests-toolbelt/0.9.1 tqdm/4.54.0 CPython/3.8.6

File hashes

Hashes for filprofiler-0.13.0-cp37-cp37m-manylinux2010_x86_64.whl
Algorithm Hash digest
SHA256 dfaf584511b0561171f1ac029926168218ff64ede9c8ade9281a7131dd2eefad
MD5 4c0361c73d9f8597ce96b18cf42be311
BLAKE2b-256 4b3ca9f50049154ad3b9857690e7129243442628ab0b3eb149495fb69c6592b5

See more details on using hashes here.

File details

Details for the file filprofiler-0.13.0-cp37-cp37m-macosx_10_14_x86_64.whl.

File metadata

  • Download URL: filprofiler-0.13.0-cp37-cp37m-macosx_10_14_x86_64.whl
  • Upload date:
  • Size: 311.5 kB
  • Tags: CPython 3.7m, macOS 10.14+ x86-64
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.2.0 pkginfo/1.6.1 requests/2.25.0 setuptools/47.1.0 requests-toolbelt/0.9.1 tqdm/4.54.0 CPython/3.7.9

File hashes

Hashes for filprofiler-0.13.0-cp37-cp37m-macosx_10_14_x86_64.whl
Algorithm Hash digest
SHA256 eebf66ccefe643a84ac7176f5d666faf48fe2fe4887bb0a1bf22edcceb4d7045
MD5 2bea38bdbda9b443a2b34db395934e96
BLAKE2b-256 68a2acf380c8cd7610e4e9251274f774d677bbd609f6d6aef8da4778593bd419

See more details on using hashes here.

File details

Details for the file filprofiler-0.13.0-cp36-cp36m-manylinux2010_x86_64.whl.

File metadata

  • Download URL: filprofiler-0.13.0-cp36-cp36m-manylinux2010_x86_64.whl
  • Upload date:
  • Size: 1.8 MB
  • Tags: CPython 3.6m, manylinux: glibc 2.12+ x86-64
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.2.0 pkginfo/1.6.1 requests/2.25.0 setuptools/49.2.1 requests-toolbelt/0.9.1 tqdm/4.54.0 CPython/3.8.6

File hashes

Hashes for filprofiler-0.13.0-cp36-cp36m-manylinux2010_x86_64.whl
Algorithm Hash digest
SHA256 e07066d0bac6694ce1059d6e7df725d388b67c1ede61dc2121d5f26f1f1ebf23
MD5 58712ad212e36b09079cda6441acef6f
BLAKE2b-256 e54fe321a261427e025e0fb255e761c1853c66cbcb22f8eb4a5317493ae55577

See more details on using hashes here.

File details

Details for the file filprofiler-0.13.0-cp36-cp36m-macosx_10_14_x86_64.whl.

File metadata

  • Download URL: filprofiler-0.13.0-cp36-cp36m-macosx_10_14_x86_64.whl
  • Upload date:
  • Size: 311.5 kB
  • Tags: CPython 3.6m, macOS 10.14+ x86-64
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.2.0 pkginfo/1.6.1 requests/2.25.0 setuptools/40.6.2 requests-toolbelt/0.9.1 tqdm/4.54.0 CPython/3.6.12

File hashes

Hashes for filprofiler-0.13.0-cp36-cp36m-macosx_10_14_x86_64.whl
Algorithm Hash digest
SHA256 20d366835fbdbfad030258d4f771e85a8a251532a0df699c3e4c5eec9bedeb07
MD5 af3a707007318de852caf50d6d353c7d
BLAKE2b-256 66256d25694a83ce86431390b1e2ada009dd4133758ef7375c7e22107fa0c8e3

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page