Read how NVIDIA cuNumeric delivers accelerated computing to the NumPy ecosystem via a drop-in library replacement, scaling applications to over 1000 GPUs.
Today NVIDIA announced the availability of a public Alpha version of cuNumeric. This drop-in replacement library for NumPy, brings distributed and accelerated computing on the NVIDIA platform to the large and growing Python community and PyData ecosystem.
Python has become the most widely used language for data science, machine learning, and productive numerical computing. NumPy is the de facto standard library, providing a simple and easy-to-use programming model. The interface corresponds closely to the mathematical demands of scientific applications, making it the foundation upon which many of the most widely used data science and machine learning programming environments are constructed.
As datasets and programs continue to increase in size and complexity, there’s a growing need to harness computational resources, far beyond what a single CPU-only node can provide. cuNumeric brings GPU-accelerated supercomputing to the NumPy ecosystem. The graph below shows effortless scaling to over 1,000 GPUs.
Key benefits of the NVIDIA cuNumeric library:
- Transparently accelerates and scales existing NumPy workflows.
- Provides seamless, import drop-in NumPy replacement.
- Provides automatic parallelism and acceleration for multiple nodes across CPUs and GPUs.
- Scales to up to thousands of GPUs optimally.
- Requires zero code changes to ensure developer productivity.
- Is freely available through GitHub and Conda.
More about cuNumeric
Learn more about cuNumeric and the Legate scaling technology, and install the free alpha release by visiting our website.
Watch these brand new NVIDIA GTC sessions:
- GTC Keynote
- GTC Session A31168 – Legate: Scaling the Python Ecosystem
- GTC Session A31138 – Accelerate Computing with CUDA Python