Technology

Nvidia launches its Hopper GPU

Nvidia chief Jensen Huang opened the tech giant’s conference call by announcing that the Hopper graphics processing unit (GPU) will begin shipping in systems from partners such as Dell, Hewlett Packard and Cisco Systems next month. GPU-equipped Nvidia systems will be available in the first quarter of next year, according to the tech giant’s CEO.

The Hopper processor, also known as the “H100”, is designed for data center tasks such as artificial intelligence. Nvidia says the H100 can “significantly” reduce the cost of deploying AI programs. For example, earlier high-performance 320 GPUs, the A100, can run with as few as 64 H100 chips, requiring only one-fifth the number of server computers and reducing power consumption by 3.5 times. Recall that Hopper was first introduced by Nvidia in March.

Along with Hopper’s availability, Jensen Huang talked about Nvidia’s new NeMo LLM “as a service” cloud offering for “large language models” that will allow customers to easily deploy very large processing models. like OpenAI GPT-3 and Megatron-Turing NLG 530B, a language model developed by Nvidia in conjunction with Microsoft.

Nvidia BioNeMo

Available in “early access” next month, the NeMo service provides GPT-3 and other pre-trained models and can be customized by a developer using a method Nvidia calls “operational learning,” adapted from a method introduced by Google scientists last year.

The NeMo version will be hosted specifically for the biomedical use of large language models such as drug discovery and will be referred to as the “Nvidia BioNeMo” service.

The Nvidia presentation also included a series of computing platforms targeting many industries, including healthcare, and OmniVerse, Nvidia’s software platform for Metaverse. These platforms include IGX for medical applications, IGX for factories, a new version of OmniVerse development’s OVX computing systems, and a new version of Nvidia’s DRIVE automotive computing platform called DRIVE Thor, which will combine GPU Hopper with Nvidia’s upcoming technology. CPU “Grace”.

Source: .com

Back to top button

Adblock Detected

Please consider supporting us by disabling your ad blocker.