danbass512 / mm-toolbox

toolbox of fast mm-related funcs

mm-toolbox

toolbox of fast mm-related funcs

Installation

To install mm_toolbox, follow these steps:

1. Optional: If you are familiar with virtual environments, create one now and activate it. If not, this step is not necessary:

$ virtualenv venv
$ source venv/bin/activate

2. Clone the repository or download the source code to your local machine.

3. With your virtual environment activated, navigate to the root directory of mm_toolbox (where setup.py is located) and run:

   python setup.py install

This will install mm_toolbox and its dependencies into your virtual environment.

Usage

After installing mm_toolbox, you can start using it in your projects by importing the necessary modules and functions. Here's an example:

from mm_toolbox.orderbook.orderbook import Orderbook

# Example usage of the orderbook from mm_toolbox
base_orderbook = Orderbook(size=500)

Contributing

Please create issues to flag bugs or suggest new features and feel free to create a pull request with any improvements.

License

mm_toolbox is licensed under the MIT License. See the LICENSE file in the repository for more details.

Performance tricks for Numba

1. Look at https://numba.pydata.org/numba-doc/dev/reference/envvars.html

   • Set NUMBA_OPT: max
   • Set NUMBA_ENABLE_AVX: 1

2. Read https://tbetcke.github.io/hpc_lecture_notes/simd.html, TLDR:

   • Ex. Doing 2.0 rather than 2 in a long calculation
   • @njit(error_model="numpy") when working with div
   • Examine ASM/LLVM generation for autovectorization debugging
   • @njit(fastmath=True) in for loops if execution order doesn't matter

Contact

If you have any questions/suggestions regarding the repository, or just want to have a chat, my handles are below 👇🏼

Twitter: @beatzXBT | Discord: gamingbeatz