MALT is a memory tool to find where you allocate your memory. It also provides you some statistics about memory usage and help to find memory leaks.
It is done to be used on laguages : C, C++, Fortran, Rust and Python.
You will find all the details about using MALT into the Your will find the dependences into the official documentation.
Your will find the dependences into the official documentation.
MALT use CMake for the build system but provide a simple configure wrapper for users familiar with autotools packaging so you can install by following the procedure :
mkdir build
cd build
../configure --prefix={YOUR_PREFIX}
make
make test
make installYour will find the dependences into the official documentation.
MALT currently provides a dynamic library you need to preload in your application to wrap the default memory allocator. It provides two basic instrumentation modes.
By default MALT use backtrace to reconstruct you stack on malloc/free/... calls :
malt {YOUR_PROGRAM} [OPTIONS]Your will find the dependences into the official documentation.
You can use the webview by calling command malt-webview as :
malt-webview [-p PORT] [--no-auth] malt-YOUR_PROGRAM-1234.jsonNote: This is currently fresh released, there are for sure still some issues.
First you need to build MALT by enabling python support : you will need to have
the python headers (package python3-dev or libpython3-dev or python3-devel) on your plateform.
In practice MALT after being built will be able to run over various versions of python without beeing rebuilt as long as they follow the standard API which is currently stable. It should also work on the python delivered by Anaconda.
Supported version are currently python from version 11.
Due to large number of memory allocations in python MALT currently have a large overhead over python. There is in consequence several way to instrument your app which I sort in overhead increasing order.
# Use default mode (python-only)
malt-python ./script.py
# profile without stacks
malt-python --profile python-no-stack ./script.py
# Get C and Python stack as two distinct domains (not mixed)
malt-python --profile python-domains ./script.py
# An approximativ method by sampling instead of tracking each stack (faster but not exact)
malt-python --profile python-sampling ./script.py
# Similar but with less samples
malt-python --profile python-sampling-10M ./script.py
# Similar but with less and less samples
malt-python --profile python-sampling-20M ./script.py
# profile considering only python stacks (C is mapped under python)
malt-python --profile python-only ./script.py
# Full instrumentation of Python + C
malt-python --profile python-full ./script.pyNote: The malt-python is just a wrapper over malt command profiding a different default
profile sepcific for python, you can also use directly the malt command.
Note: The malt-python command is a temporary workaround, it might disapear in future.
If you search similar tools all over the web you might find:
- Heaptrack: A Heap Memory Profiler for Linux: KDE/heaptrack: https://github.com/KDE/heaptrack
- Memoro: A detailed Heap Profiler : https://epfl-vlsc.github.io/memoro/
- Memtrail: https://github.com/jrfonseca/memtrail
- MTuner: https://milostosic.github.io/MTuner/
- Profiler provided with google allocator: Google Heap Profiler
- Valgrind memcheck
- Valgrind massif: Valgrind massif with Massif visualizer
- Dr. Memory
- Commercial tool, Parasoft Insure++
- Commercial tool, Unicom PurifyPlus (previously IBM)
- Tau is more a communication profiling tool for HPC apps, but it offers a memory module
- Similar approach than MALT for the backend: IgProf
- A debug malloc library: Dmalloc
- Profiling and leak detection: MemProf
- Malloc count
- mpatrol
- Tracing tool for parallel programs: EZTrace
- Find Obsolete Memory: FOM Tools
- Memray: A memory profiler support python. https://bloomberg.github.io/memray/
- Scalene: A perf and memory profiler for C & python : https://pypi.org/project/scalene/
If ever I missed new ones, you can also look on the repos of this person keeping an up-to-date list: https://github.com/MattPD/cpplinks/blob/master/performance.tools.md
If you search some parallel memory allocators, you can find those one on the net:
- Jemalloc (facebook, firefox)
- TCMalloc (google)
- Hoard
- Lockless allocator
- MPC memory allocator from Multi-Processor Computing framework.
- mimalloc
MALT is distributed under CeCILL-C license (LGPL compatible).
If you publish about MALT, you cite this research paper as reference :
Sébastien Valat, Andres S. Charif-Rubial, and William Jalby. 2017. MALT: a Malloc tracker.
In Proceedings of the 4th ACM SIGPLAN International Workshop on Software Engineering for
Parallel Systems (SEPS 2017). Association for Computing Machinery, New York, NY, USA,
1–10. https://doi.org/10.1145/3141865.3141867
If you have questions of remarks, you can also send me a mail at memtt@progranet.ovh.