Microsoft has released details of its latest virtual machine for AI workloads, as the company continues to develop and launch AI products after a significant investment in OpenAI, the creator of ChatGPT.
In an article published on his blog, the company highlights the important role its GPU-accelerated virtual machines play in supporting many advances in generative AI, both its own and other companies, including its own Azure OpenAI service, which puts AI writers as writers of artificial intelligence in the hands of its customers. GPT-3.5, Codex and DALL-E 2.
With its new ND H100 v5 virtual machine, Microsoft promises “significantly faster performance” than the previous generation ND A100 v4 virtual machines.
Microsoft Azure VM for Artificial Intelligence
The updated virtual machine uses eight NVIDIA H100 Tensor Core GPUs with the ability to scale up to thousands of units. It also features 4th Generation Intel Xeon Scalable processors and 16 channels of 4800 MHz DDR5 DIMMs.
With this move, Microsoft aims to bring some of the latest, highest-performing technologies at a more affordable price point to AI startups, many of which will compare their stories to the recent success of OpenAI. That company’s co-founder, Greg Brockman, said:
more interesting reading
“Co-engineering the supercomputers with Azure was instrumental in scaling our demanding AI training needs, enabling our research work and alignment on systems like ChatGPT.”
Looking to the future, ND H100 v5 virtual machines will need to be able to handle large amounts of training for significant periods of time as the world strives for AI to be as reliable and accurate as possible.
The new H100 virtual machines are now available for preview as the tech giant looks to make them part of its standard offering in the Azure portfolio.