Apache SINGA vs MegEngine

Apache SINGA is a free distributed deep-learning library for training deep-learning and machine-learning models. Its listed training mode is local, with data-parallel training across multiple GPUs on one node or across multiple nodes. GPU packages use CUDA and cuDNN, and Docker images are available for Nvidia GPUs. Installation options include pip, Docker, Conda, and source builds; pip documentation covers Python 3.9, 3.10, and 3.11. SINGA supports Python and C++, loads ONNX models, and can save models defined through its APIs in ONNX format. Optimizers listed include momentum-based stochastic gradient descent, Adam, RMSProp, and AdaGrad. Models can be queried in an RDBMS. The repository and Google Colab provide domain-specific models, including healthcare and science examples. It runs on Linux, macOS, Windows, and self-hosted environments, with on-premises deployment and Apache License 2.0 licensing.

MegEngine is a free, open-source deep learning framework for developing and deploying models, with automatic differentiation. It uses one model for training and inference, including quantization and dynamic shapes. The project says enabling DTR can reduce GPU memory use to one-third of the original. Inference is supported across x86, Arm, CUDA, and ROCm; GPU use requires compatible device drivers. Python packages are listed for 64-bit Linux and Windows, macOS 10.14 or later, and Android 7 or later, though macOS and Android installations are CPU-only. The installation guide specifies Python 3.6–3.9. MegEngine Lite supplies C/C++, Rust, and Python runtimes for deployment. MgeConvert handles conversion between MegEngine and third-party model formats, MegFile provides Python file interfaces for S3, HTTP, and local files, and MegFlow is a streaming computation framework for AI applications. Tutorials address beginners and advanced developers. Support options include GitHub issues, a forum, a QQ group, and email.