Anaconda’s Accelerate package is a package that enables us to leverage the exceptional performance of Intel CPUs as well as Nvidia’s GPUs. It's an add-on to the anaconda package that from the documentation features:
- Binding to several CUDA libraries such as cuBLAS, cuFFT, cuSPARSE, cuRAND, and CUDA Sorting
- Speed-boosted linear algebra operations in NumPy, SciPy, scikit-learn, and NumExpr libraries using Intel's Math Kernel Library
- Accelerated variants of NumPy's built-in UFuncs
- Increased-speed Fast Fourier Transformations (FFT) in NumPy
Much like the other libraries in this chapter, it's designed specifically to cope with huge datasets.
The official documentation of Anaconda’s Accelerate can be found at https://docs.continuum.io/accelerate/# It's unfortunately quite a sparse documentation at the time...
