Microsoft launches ND H100 v5 virtual machine: uses Nvidia H100 GPU to meet enterprise ChatGPT generation AI requirements
CTOnews.com, March 14 (Xinhua)-- Microsoft announced today that it will strengthen its cooperation with Nvidia to upgrade GPU from A100 to H100 and launch a new ND H100 v5 virtual machine specifically for artificial intelligence.
Microsoft said that the virtual machine is mainly for enterprises that need to integrate or develop artificial intelligence, and the configuration of the virtual machine attached to CTOnews.com is as follows:
8 Nvidia H100 Tensor Core GPU interconnected via next-generation NVSwitch and NVLink 4.0
Under the non-blocking (non-blocking) fat tree (Fat-Tree) architecture network, the NVIDIA Quantum-2 CX7 InfiniBand performance of each GPU can reach 400 Gb / s, and each virtual machine can reach 3.2Tb / s. With NVSwitch and NVLink 4.0, the bisection bandwidth between the eight local GPU within each virtual machine is 3.6TB / s.
Using Intel 4th generation Xeon Scalable processor
Connect to GPU using PCIe Gen5 with each GPU bandwidth of 64GB / s
16-channel 4800MHz DDR5 DIMM
Microsoft previously announced ChatGPT-related services in its Azure OpenAI service, and this virtual machine complements the service, which allows third parties to access chatbot technology through Azure.