Feature
- Jensen Huang believes fast computing combined with generative AI is a game-changing innovation accelerator.
- The alliance seeks to speed growth through infrastructure, software, and services that improve energy efficiency and lower costs.
- One of the first research groups to use PaxML with NVIDIA GPUs for exploratory research is Google DeepMind.
The announcement of a partnership between Google Cloud and NVIDIA aims to serve better clients who want to build and deploy large-scale generative AI models while accelerating data science workloads by offering improved AI infrastructure and software. It now offers advanced NVIDIA capabilities, enhancing its machine-learning environment for global AI leaders due to this collaboration. The following is a breakdown of their joint projects:
Bringing AI Supercomputers to Google Cloud
- At Google Cloud Next, CEOs Thomas Kurian (Google Cloud) and Jensen Huang (NVIDIA) shared their commitment to AI innovation. The collaboration makes it simple to deploy AI supercomputers using NVIDIA technology on Google Cloud.
- For over two years, Google’s DeepMind and research teams have utilized NVIDIA tech, contributing to new hardware and software integrations.
Driving Innovation with Accelerated Computing and Generative AI
- Jensen Huang believes fast computing combined with generative AI is a game-changing innovation accelerator.
- The alliance seeks to speed growth through infrastructure, software, and services that improve energy efficiency and lower costs.
- Thomas Kurian acknowledged Google Cloud’s AI history and NVIDIA GPU usage in various Google products.
Read more:- Fortitude Re and IBM and Partnership: Transforming Life Insurance Policy Servicing with AI
NVIDIA Integration for AI and Data Science Development
- For NVIDIA accelerated computing, Google’s PaxML framework, which is used to build large language models (LLMs), has been tuned.
- PaxML, which was first designed for Google TPU accelerator slices, now makes it possible to leverage NVIDIA H100 and A100 Tensor Core GPUs. A PaxML container that is GPU-optimized and accessible through the NVIDIA NGC software library facilitates enhanced experimentation and scaling.
- PaxML makes use of JAX, which the OpenXLA compiler has GPU-optimized.
- One of the first research groups to use PaxML with NVIDIA GPUs for exploratory research is Google DeepMind.
Advancing Data Science with NVIDIA GPUs
Google has integrated serverless Spark with NVIDIA GPUs in Dataproc for streamlined AI data prep in Apache Spark workloads.
Key Collaborative Initiatives
Google Cloud A3 VMs Powered by NVIDIA H100 GPUs: Soon, NVIDIA H100 GPU-equipped, purpose-built Google Cloud A3 VMs will be widely accessible. Compared to their predecessors, these VMs provide three times quicker training with much more networking bandwidth.
NVIDIA H100 GPUs on Vertex AI Platform: H100 GPUs will be available on Google Cloud’s Vertex AI platform in the following weeks, allowing for quick construction of generative AI LLMs.
NVIDIA DGX GH200 AI Supercomputer on Google Cloud: It will receive advance access to the NVIDIA DGX GH200 AI supercomputer, powered by the NVIDIA Grace Hopper Superchip. This partnership intends to investigate the supercomputer’s potential for generative AI workloads.
Google Cloud on NVIDIA DGX Cloud AI Supercomputing: Direct web browser access to the NVIDIA DGX Cloud AI supercomputing and software will provide advanced training workloads with speed and scalability.
NVIDIA AI Enterprise on Google Cloud Marketplace: Access to NVIDIA AI Enterprise, a safe cloud-native software platform that makes it easier to create and deploy enterprise-ready applications in generative AI, spoken AI, computer vision, and other areas, is provided to users.
NVIDIA L4 GPUs on Google Cloud: NVIDIA L4 Tensor Core GPUs were made available via Google Cloud via the G2 VM. Switching to L4 GPUs from CPUs for AI video operations yields a 120x speed boost and 99% efficiency gain. For creating images and text, VDI, and AI-accelerated audio/video transcoding, L4 GPUs are essential.
Read more:- WatsonX Introduced by IBM: Transforming Business with AI
Conclusion
Expectations are that the collaboration between Google Cloud and NVIDIA will transform data science and artificial intelligence, opening the door for cutting-edge innovation, efficient processes, and hitherto unheard-of capabilities in generative AI.
