The latest version provides full support for running deep learning, automotive and scientific analysis on NVIDIA’s Ampere GPUs.
Overview
NVIDIA has announced the availability of MATLAB R2021a on the NGC catalog, which is optimized for deep learning and scientific computing on Ampere GPUs. This version includes support for NVIDIA TensorRT, TensorFlow model import, and new features for managing deep learning experiments.
What You'll Learn
1
How to generate optimized CUDA code using NVIDIA TensorRT in GPU Coder
2
How to import TensorFlow 2 models directly into MATLAB
3
How to use the new Experiment Manager to design and run training experiments
4
How to add dynamic controls and plots to live scripts without writing code
Key Questions Answered
What new features does MATLAB R2021a offer for deep learning on Ampere GPUs?
MATLAB R2021a introduces several features for deep learning on Ampere GPUs, including support for NVIDIA TensorRT 7.2.x for generating optimized CUDA code, the ability to import TensorFlow 2 models directly, and a new Experiment Manager for managing training experiments. Additionally, it offers dynamic controls in live scripts and tools for adding plots without coding.
How can users download the MATLAB R2021a container?
Users can download the MATLAB R2021a container from the NGC catalog at the provided link. This container is specifically designed to leverage the capabilities of NVIDIA's Ampere GPUs for deep learning and scientific computing.
Technologies & Tools
Software
Matlab
Used for deep learning, automotive, and scientific analysis on NVIDIA's Ampere GPUs.
Software
Nvidia Tensorrt
Used for generating optimized CUDA code in GPU Coder.
Software
Tensorflow 2
Models can be imported directly into MATLAB for further development.
Key Actionable Insights
1Utilizing the Experiment Manager can significantly streamline the process of training deep learning models.This tool allows users to efficiently design, run, and compare multiple experiments, which is crucial for optimizing model performance.
2Importing TensorFlow 2 models into MATLAB can enhance workflow integration for users familiar with TensorFlow.This feature allows for leveraging existing models and transitioning smoothly into MATLAB's environment, thus saving time and resources.
3Using dynamic controls in live scripts can improve interactivity and user engagement.This feature is particularly useful for presentations or educational purposes, allowing users to manipulate parameters in real-time.