Intel's reference deep learning framework designed for high performance across CPUs, GPUs, and custom hardware.
neon is Intel's reference deep learning framework designed for high performance across all hardware types, including CPUs, GPUs, and custom Nervana hardware. It solves the need for a fast, extensible framework that allows developers to train neural networks efficiently without being locked into a specific hardware backend.
Deep learning researchers, data scientists, and developers who require high-performance model training and flexibility in hardware deployment, particularly those working with Intel architectures.
Developers choose neon for its benchmark-leading performance, swappable hardware backends that enable code portability, and strong optimization for Intel CPUs via MKL integration, making it ideal for rapid experimentation and production deployment.
Intel® Nervana™ reference deep learning framework committed to best performance on all hardware
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
Achieves 2x faster than cuDNNv4 on certain tasks, with examples like 2.5s per macrobatch on AlexNet, enabling rapid model iteration.
Swappable backends allow the same code to run on CPUs, GPUs, or Nervana hardware without modifications, providing flexibility in deployment.
Integrates Intel Math Kernel Library (MKL) automatically for enhanced performance on Intel Xeon and Xeon Phi systems, as highlighted in the README.
Offers pre-trained weights and scripts for state-of-the-art models including VGG, ResNets, and reinforcement learning, facilitating transfer learning.
Intel has ceased maintenance and support, with no bug fixes or updates, making it unsuitable for production environments that require reliability.
Requires manual configuration of environment variables like OMP_NUM_THREADS and KMP_AFFINITY to achieve optimal performance on Intel CPUs, adding setup complexity.
Compared to frameworks like TensorFlow or PyTorch, neon has a smaller community and fewer third-party tools, exacerbated by its discontinuation.