filprofiler
filprofiler copied to clipboard
A Python memory profiler for data processing and scientific computing applications
Right now there's One Big Lock, which will slow things down for large-scale computation.
You can have a tree of alternating call_id and line number. start_call()/finish_call()/set_line_number() are traversals of the tree, occasionally inserting new node if necessary. This should be O(siblings), so pretty fast....
If a library uses a C thread pool, allocations won't be assigned correctly: they'll be assigned to `import pylibrary` rather than the code that triggered the threadpool tasks. Solutions: 0....
Instead of just having a stacktrace like `run_some_code()` it should be `run_some_code() -> [Non-Python thread]` or something. Otherwise it can be confusing where that memory is being allocated, especially for...
Good bookkeeping practice.
Original text: "The Software may not be used in applications and services that are used for or aid in the exploration, extraction, refinement, processing, or transportation of fossil fuels. The...
The way to do this would be `nbconvert --execute`, except you need to use the Fil kernel... and there's no way to choose the kernel on the CLI. Options: 1....
Hoping for reproducer, but may have to make my own: 1. Python 3.6 (running on Debian) loads 200MB CSV. Result uses 1GB of RAM, this shows up fine in Fil....
And: ``` $ grep peak-memory.svg TB@@.. TB@@.. TB@@.. TB@@.. ... ``` It's truncated so getting rid of it will be a little tricky...
When detecting memory leaks, you don't need to track every allocation: the whole point is that some allocation is going to happen over and over. Sampling in this case is...