Amazon EC2 instances powered by Trainium2 chips now available

Dec. 4 (Portaltic/EP) –

Amazon Web Services (AWS) has announced the launch of Amazon Elastic Compute Cloud (Amazon EC2) Trn2 compute instances, equipped with Trainium2 chips and designed specifically for generative Artificial Intelligence (AI) applications.

The company presented Trainium2 at the previous edition of AWS re:Invent, where it announced that it was continuing to work to offer customers more opportunities to run any app or workload with both this chip and Graviton4.

AWS CEO Matt Garman announced at the 2024 annual conference, which started this Monday, the general availability of instances of its web service with the capacity to scalable on-demand computing Amazon EC2 powered by AWS Trainium2.

In this sense, has remembered The new Amazon EC2 Trn2 instances are designed specifically for high-performance deep learning training of AI models, including large language models (LLM) and latent diffusion models.

He has also pointed out that Trn2 is the instance of Amazon EC2 higher performance for deep learning and generative AI, offering 30 to 40 percent better price-performance than the current generation of graphical processing unit (GPU)-based instances .

This is because a single instance of Trn2 combines 16 new Trainium2 chips interconnected with ultra-fast NeuronLink high-bandwidth, as well as a low-latency chip-to-chip interconnect, with 20.8 petaflops of peak processing.

AWS has also noted that for larger models that require more compute, it has worked on the Trn2 UltraServers, virtual servers that allow software developers to have resizable processing capacity for any workload.

With these new systems, customers can scale their workloads to improve real-time inference performance for billion-parameter models in production. In this sense, the firm has indicated that it is building a UltraCluster EC2 of Trn2 Ultraservers together with Anthropic.

This project is called Project Rainierwhich will expand distributed model training through Trainium2 chips interconnected with EFA networks, that is, network interfaces for Amazon EC2 instances that allow customers to run applications that require high levels of communication between nodes at AWS scale.

The Trn2 compute instances now available in the AWS region of a part of the United States – specifically, Ohio – and will soon be extended to other regions. The Trn2 UltraServers, for their part, are in a preliminary version for testing.

TRAINIUM3

The company also announced at this meeting Trainium3, its next-generation AI chip, which will allow customers to create different larger models more quickly, as well as offer superior real-time performance in their implementation.

These will be the first chips of AWS manufactured with a 3 nanometer (3nm) process node, which will set a new standard for performance, energy efficiency and density.

Thus, the cloud servers used to train AI programs (UltraServers) with Trainium3 technology are expected to have four times the performance of the Trn2 UltraServers. Thanks to this, customers They will be able to iterate even faster by creating models and delivering superior real-time performance in your deployment.

Amazon Web Services has finally announced that the first Trainium3 chips are expected to be available late next year.