Developers and enthusiasts interested in learning more about Machine Learning frameworks may be interested in a new framework interoperability series created by the team at NVIDIA. In the first part you will learn about distinct memory layouts and how the Apache Arrow format can significantly speed up data transfers across distinct data science and machine learning frameworks such as TensorFlow, PyTorch, MXNet, cuDF, CuPy, Numba, and JAX. As well as how asynchronous memory allocation facilitated by memory pools is crucial to avoid overheads as big as 90% of the overall runtime of your pipeline.
“Efficient pipeline design is crucial for data scientists. When composing complex end-to-end workflows, you may choose from a wide variety of building blocks, each of them specialized for a dedicated task. Unfortunately, repeatedly converting between data formats is an error-prone and performance-degrading endeavor. Let’s change that!”
In this Machine Learning frameworks series, NVIDIA discuss different aspects of efficient framework interoperability:
- We start with this post discussing pros and cons of distinct memory layouts as well as memory pools for asynchronous memory allocation to enable zero-copy functionality.
- In the second post, we highlight bottlenecks occurring during data loading/transfers and how to mitigate them using Remote Direct Memory Access (RDMA) technology.
- In the third post, we dive into the implementation of an end-to-end pipeline demonstrating the discussed techniques for optimal data transfer across data science frameworks.
“Zero-copy functionality is a crucial technique to efficiently copy data across GPU-accelerated data science frameworks: TensorFlow, PyTorch, MXNet, cuDF, CuPy, Numba, and JAX (see Figure 2). In the following, we will show you how to achieve that in a systematic manner. If you are only here to look up the commands on how to transfer data from one framework to another, you might want to have a look at this conversion table.”
Source : NVIDIA
Latest Geeky Gadgets Deals
Disclosure: Some of our articles include affiliate links. If you buy something through one of these links, Geeky Gadgets may earn an affiliate commission. Learn about our Disclosure Policy.
