AMD extends the support to the RDNA 3 GPUs | 3 components of computer hardware | Hardware computer list | Hardware restaurant | Turtles AI
AMD has recently expanded the support for the development of AI applications on the GPUs of the RDNA 3 series, making the calculation accessible to even ordinary users. This evolution represents a significant step towards the integration of AI in consumer solutions.
Key points:
- Support for AI/ml development extended to RDNA 3 GPUs via Rocm 6.1.3.
- The latest Radeon GPUs offer superiors and greater memory capacity.
- Integration with popular tools such as Pytorch and Tensorflow.
- Most convenient local solutions compared to cloud services for calculating AI.
AMD has undertaken an important initiative to make the calculation of AI accessible not only to professionals in the sector, but also to consumers who use the Radeon graphic cards. With the update of the Rocm framework, the company has extended the support for the workloads of Machine Learning to the RDNA 3 architecture, which include the GPUs of the Radeon RX 7000 series and the Workstation solutions of the Radeon W7000 series. This move responds to a growing demand for economic solutions for the development and training of applications based on AI. Traditionally, Machine Learning’s workloads have been associated with data center infrastructures, but now the consumer GPUs can provide comparable performance, thus reducing dependence on cloud services.
The new Rocm 6.1.3 For Linux now allows researchers and developers to use tools widely adopted such as Pytorch, Tensorflow and ONNX Runtime, optimizing the use of the performance of modern AMD GPU. These devices, equipped with up to 48 GB of memory, offer a significantly higher calculation power, with a processing capacity for calculation units that exceeds that of previous generations by more than double. This availability of memory is crucial to manage increasingly complex machine learning models.
The new AMD GPUs, thanks to the presence of a considerable number of accelerators AI and an optimized architecture, are placed as a valid alternative to more expensive and complex solutions. In addition, the possibility of having a local system eliminates some of the limits associated with the use of the cloud, such as latency and management of sensitive data. The update made to the AMD software stack therefore represents an important progress for the adoption of Machine Learning technologies, allowing facilitated access to tools and resources until recently exclusively to more professional systems.
AMD is positioning itself as a key actor in the panorama of calculation AI, promoting the use of its RDNA 3 architecture to meet the emerging needs of the Machine Learning developers.