Shulou(Shulou.com)11/24 Report--

CTOnews.com, April 14, according to the official release of Microsoft Azure, Microsoft has applied its supercomputing experience over the past decade and its experience in supporting very large AI training workloads to building a large-scale and high-performance AI infrastructure. Microsoft Azure smart cloud, especially the virtual machine (VM) accelerated by graphics processing unit (GPU), has laid the foundation for the generative AI development of Microsoft and its customers.

Microsoft is introducing ND H100 v5 VM, a family of AI virtual machines that Azure has hitherto been more powerful and highly scalable. The virtual machine supports up to 8 to thousands of NVIDIA H100 GPU interconnected through Quantum-2 InfiniBand networks on demand, significantly improving the performance of the AI model. Compared to the previous generation of ND A100 v4 VM, this virtual machine includes the following innovative technologies:

8 NVIDIA H100 Tensor Core GPU interconnected through next-generation NV Switch and NV Link 4.0.

In the non-blocking fat tree network, each GPU carries 400 Gb / s NVIDIA Quantum-2 CX7 InfiniBand and performance per VM 3.2Tb / s.

8 pieces of local GPUs in each VM are interconnected by NV Switch and NV Link 4.0 with split bandwidth through 3.8Tb / s pairs.

The fourth generation Intel Xeon scalable processor.

PCIE Generation 5 host-to-GPU interconnects with each GPU bandwidth of 64Gb / s.

4800 MHz DDR5 memory with 16 channels.

CTOnews.com Note: NVIDIA Quantum-2 uses the seventh-generation NVIDIA InfiniBand architecture to provide AI developers and scientific researchers with superior network performance and rich features to help them solve challenging problems. NVIDIA Quantum-2 powers advanced supercomputing data centers with software-defined networking, network computing, performance isolation, advanced acceleration engines, remote direct memory access (RDMA), and ultra-fast speeds up to 400 Gb / s.

Microsoft said that large-scale AI has been built into Azure's DNA. Initial investment in large language model research, such as Turing, and milestones such as building the first AI supercomputer in the cloud are well prepared for the advent of generative AI. Azure services such as Azure machine learning enable customers to access AI supercomputers for model training. Azure OpenAI services enable customers to take advantage of the power of large-scale generative AI models. "scale" has always been one of Azure's goals for optimizing AI infrastructure. Now, Microsoft is bringing supercomputing capabilities that don't require a lot of physical hardware or software investment for startups and enterprises of all sizes.

Now, the ND H100 v5 preview release will become a standard service in the Azure portfolio.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.