murmurhash VS py-spy

If you have very large dicts, you might find this hash table I wrote for spaCy helpful: https://github.com/explosion/preshed . You need to key the data with 64-bit keys. We use this wrapper around murmurhash for it: https://github.com/explosion/murmurhash

There's no docs so obviously this might not be for you. But the software does work, and is efficient. It's been executed many many millions of times now.

Some notes: We don't need Sha256 and not evey base64; nothing will happen if keys will not distribute very equally. we could take MMH3; googling "python murmurhash" gives 2 interesting results; and since both use the same cpp code, let's take the one with most stars Other options would be to simply do (% NUM_SHARDS) or even shift right (however must have shards count == power of 2).

For CPU cycles, py-spy[0] is getting more and more used. For RAM, I would like to known too...

[0] -- https://github.com/benfred/py-spy

Theres also Py Spy, a profiling tool that can generate flame charts containing a mix of python and C (or C++) calls.

https://github.com/benfred/py-spy

It's worked really well for my needs

You should profile your consumer to check the bottlenecks. You can use the excellent py-spy(written in Rust). IMO a few usage of Numba there and there should solve your performance issues.

So as a first step you'll want to profile your program to figure out where it's slow, and hopefully that'll also tell you why it's slow. I'm the (biased) author of the Sciagraph profiler which is designed for this sort of application (https://sciagraph.com) but you can also try py-spy, which isn't as well designed for data processing/analysis applications (e.g. it won't visualize parallelism at all) but can still be informative (https://github.com/benfred/py-spy). Both are written in Rust ;)

Any advantages/disadvantages compared to py-spy [1]?

[1]: https://github.com/benfred/py-spy

Python profiling is enabled primarily through cprofile, and can be visualized with help of tools like snakeviz (output flame graph can look like this). There are also memory profilers like memray which does in-depth traces, or sampling profilers like py-spy.

A good profiler would be https://github.com/benfred/py-spy . If you run your app/benchmark with it, it should be able to draw a flamegraph telling you where the majority of time is spent. The info here is quite fine grained so it would already tell you where the bottleneck is. Without a full-fledged profiler you can also measure the timings in various parts of the code to understand where the bottleneck is.

Might be worth raising an issue on py-spy (a python profiler written in rust which "supports profiling native python extensions written in languages like C/C++ or Cython" to see if that can close the loop.