When we announced the preview of our new NCv3 virtual machines back in November, we knew they’d be very popular with our customers. NCv3 brings NVIDIA’s latest GPU – the Tesla V100 – to our best-in-class HPC, machine learning, and AI products to bring huge amounts of value across a variety of industries. One preview customer told us their speech recognition models trained in less than 20 minutes, instead of the 1-2 hours that previous generation GPUs required. Another customer told us about the 40-50% performance boost they saw on their reservoir simulations.
With these fantastic customer success stories, we are ecstatic to announce that the NCv3 virtual machines are now generally available in the US East region. We'll be adding NCv3 to EU West and US South Central later this month. We'll add AP Southeast in April and UK South and IN Central in May.
Read more about it in the Azure blog.