New NVIDIA computing platforms for large language models and generative AI workloads
The new platforms from NVIDIA are a great help for developers to quickly build specialized AI applications capable of delivering new services and insights.
The platforms combine the full NVIDIA computing software stack with the latest NVIDIA Ada, NVIDIA Hopper™ and NVIDIA Grace Hopper™ processors, including the new NVIDIA L4 Tensor Core GPU and NVIDIA H100 NVL GPU. Each platform is optimised for demanding workloads, including image generation, AI video, large language model deployment, and recommendation output.
Each platform includes an NVIDIA GPU optimized for specific generative AI workloads, as well as dedicated software. For example, the NVIDIA L4 for AI Video can deliver up to 120 times more AI video performance than CPUs, combined with 99 percent better power efficiency. And the NVIDIA L40 for Image Generation is optimized for graphics and AI-enabled 2D, video and 3D image generation. The NVIDIA H100 NVL for Large Language Model deployment is ideal for large-scale deployments of large LLMs such as ChatGPT. As for NVIDIA Grace Hopper for Recommendation Models, it is ideal for graph recommendation models, vector databases and graph neural networks.
Free practice labs to try out NVIDIA's inference platform for generative AI are available on NVIDIA LaunchPad. Sample labs include training and deploying a support chatbot, deploying an end-to-end AI workload, configuring and deploying a language model on the H100, and deploying a fraud detection model with NVIDIA Triton™.