Skip to main content

Google Cloud now offers preemptible GPUs

(Image credit: Image Credit: Ken Wolter / Shutterstock)

Google Cloud can now benefit from the use of preemptible GPUs to provide a major boost to their online platforms.

The new launch, detailed in a recent company announcement, shows that Google is following the same practice in terms of pricing, offering these GPUs with a 50 per cent discount. However, as one may expect from preemptible computing solutions, if Google decides to shut you down to use the services, it can do so at any given time, with just a 30-second warning.

It was also said that one may use any given preemptible GPU for up to 24 hours.

Google offers two GPUs for the services – nVidia K80 and nVidia P100. The former will set you back $0.22 per GPU hour, while the latter will cost $0.73. You also need to pay for the VM, although preemptible VM prices apply for these.

Announcing the new offering, Google said these GPUs are a good fit for any fault-tolerant machine learning workloads and similar tasks that run in batches. “Preemptible GPUs will be a particularly good fit for large-scale machine learning and other computational batch workloads as customers can harness the power of GPUs to run distributed batch workloads at predictably affordable prices,” it said.

A few months back, Google dropped the prices for non-preemptible GPUs. Also now, Google said the GPUs are available in the US-central1 region.

Image Credit: Ken Wolter / Shutterstock

Sead is a freelance journalist with more than 15 years of experience in writing various types of content, from blogs, whitepapers, and reviews to ebooks, and many more, across sites including Al Jazeera Balkans, TechRadar Pro, IT Pro Portal, and CryptoNews.