Azure and NVIDIA Unleash Next-Generation AI GPU Acceleration

Microsoft Azure and NVIDIA Partnership for Accelerated Computing

Microsoft Azure and NVIDIA have collaborated to integrate NVIDIA H100 Tensor Core GPUs with Azure ND H100 v5 virtual machines (VMs). This partnership allows Azure users to take advantage of the latest advancements in NVIDIA’s accelerated computing technology, specifically in the field of generative AI applications. Here’s a detailed overview of the collaboration:

NVIDIA H100 GPU and Azure ND H100 v5 VMs

• Azure ND H100 v5 VMs are now integrated with NVIDIA H100 Tensor Core GPUs and Quantum-2 InfiniBand networking.
• The integration enables seamless scaling of generative AI and high-performance computing applications on Azure.
• NVIDIA H100 GPUs offer supercomputing-class performance through architectural innovations like fourth-generation Tensor Cores, a new Transformer Engine for enhanced large language model (LLM) acceleration, and NVLink technology for faster inter-GPU communication.
• The integration of the NVIDIA Quantum-2 CX7 InfiniBand ensures flawless performance across GPUs, even at massive scales, making it comparable to the computational capabilities of advanced supercomputers.

Empowering LLMs and Computer Vision Models

• Azure ND H100 v5 VMs hold immense potential for training and inferring complex LLMs and computer vision models.
• These neural networks power various generative AI applications such as question answering, code generation, audio, video, image synthesis, and speech recognition.
• The ND H100 v5 VMs deliver up to a 2x speedup in LLM inference compared to previous generation instances, as demonstrated by the BLOOM 175B model.
• This performance boost optimizes AI applications and fuels innovation across industries.

AI Training and Inference Capabilities

• The synergy between NVIDIA H100 Tensor Core GPUs and Microsoft Azure provides enterprises with unparalleled AI training and inference capabilities.
• The collaboration streamlines the development and deployment of production AI, leveraging the NVIDIA AI Enterprise software suite and Azure Machine Learning for MLOps.
• Industry-standard MLPerf benchmarks validate the groundbreaking AI performance achieved through this partnership.
• The integration of the NVIDIA Omniverse platform with Azure extends the collaboration’s reach, offering users industrial digitalization and AI supercomputing capabilities.

Overall, the partnership between Microsoft Azure and NVIDIA brings advanced GPU acceleration to Azure’s cloud infrastructure, enabling developers and researchers to unlock the full potential of generative AI applications and accelerate innovation in various industries.