Fast Python for GPUs and multi-core with NumbaPro and MKL Optimizations
Accelerate is an add-on to Continuum’s free enterprise Python distribution, Anaconda. It opens up the full capabilities of your GPU or multi-core processor to Python. Accelerate includes two packages that can be added to your Python installation: NumbaPro and MKL Optimizations. MKL Optimizations makes linear algebra, random number generation, Fourier transforms, and many other operations run faster and in parallel. NumbaPro builds fast GPU and multi-core machine code from easy-to-read Python and NumPy code with a Python-to-GPU compiler.
Install Anaconda Accelerate behind your firewall, and get tools to manage the deployment of Python, R, and internal packages with Anaconda Server.
- NumbaPro compiler targets multi-core CPU and GPUs directly from simple Python syntax
- Easily move vectorized NumPy functions to the GPU
- Multiple CUDA device support
- Bindings for CUDA libraries, including cuBlas, cuRand, cuSparse, and cuFFT
- Support for array slicing and fast array math
- Use multiple threads without worrying about the GIL
- Supported on NVIDIA CUDA-enabled GPUs with compute capability 2.0 or above on Intel/AMD (x86) processors.
List of CUDA GPUs.
Get Started with the Anaconda Accelerate Quick Start Guide [pdf]
MKL Optimizations Features
- Speed-boosted NumPy, SciPy, scikit-learn, and NumExpr
- The packaging of MKL with redistributable binaries in Anaconda for easy access to the MKL runtime library.
- Python bindings to the low level MKL service functions, which allow for the modification of the number of threads being used during runtime.
- Fast Fourier Transformations (FFT) in NumPy linked to MKL for increased speed (exclusively in Anaconda Accelerate)
MKL Optimizations Benchmarks
View the code used to create these benchmark plots.
In addition to receiving all of the above valuable features, your purchase of Accelerate ssupports the development of Numba and many other open-source projects. Learn more.
Faster than NumPy?
NumPy is the fundamental array library of choice for scientists and engineers. It has popularized array-oriented, vector programming techniques to a wide variety of Python programmers. Scientists and engineers regularly use NumPy's fast vectorized operations for processing and analyzing vast datasets.
Numba, our open source compiler, allows scientists and engineers to easily compile NumPy expressions to efficient parallel machine code, without needing explicit code-generation and recompilation steps. NumbaPro, the premium Numba extension in Accelerate, adds high-level API for user to build CUDA functions quickly, and has bindings to CUDA libraries: cuBLAS, cuSPARSE, cuRAND, cuFFT. Write your code with arrays, and execute it on GPUs and multi-core by adding just a single decorator.
NVIDIA CUDA-enabled GPU with compute capability 2.0 or above. List of CUDA GPUs.
For $129.00, you will receive one Anaconda Accelerate license, which includes the libraries MKL Optimizations and NumbaPro. You are entitled to any updates issued to these packages for a period of one year.
Although this license is persistent and permanent, access to package upgrades will expire after one year. You may choose to renew your license in order to access subsequent upgrades, or you may continue to use your existing copy of Anaconda Accelerate without upgrading.
This license entitles the licensee (a singular user) to use Accelerate on as many machines as are personally used by the licensee and up to 4 nodes within a cluster. (For example, you CAN use Accelerate on your home laptop and your office desktop.)
If Anaconda Accelerate is to be used on a cluster larger than 4 nodes, or used by/available for more than one user, please contact us and we will help determine the right solution for you.
Anaconda Accelerate End User License Agreement
Yes, keep me up to date with the latest news and information about Continuum's products and services.