There was a time when supercomputers were only available to a handful of organizations, mostly governments, public research facilities, and scientific bodies. The rise of cloud computing and the widespread availability of sophisticated cloud workload management (CWM) tools have significantly reduced the barrier to entry.
Just last week, YellowDog, a CWM specialist based in Bristol, UK, assembled a super virtual computer using his proprietary platform and at its peak, which took about 10 minutes, he had assembled an army of over 3.2 million vCPUs.
Even though it wasn’t anywhere as powerful as Fugaku, that was enough to propel it into the top 10 fastest supercomputers in the world, at least for a few minutes.
(PSA: by the way, we are going to update our Black Friday web hosting deals and Black Friday Website Builder Deals page at least once a day until Cyber Monday)
Of short time
The procurement, which was done on behalf of a pharmaceutical company, allowed a popular drug discovery application to run as a single cluster. Behind the envelope calculation, the gross cost of the project is approximately $ 65,000.
That’s 33,333 instances of the AWS 96-core c5.24xlarge instance, one of the instances used during runtime (basically similar to baremetal servers or dedicated servers) at $ 1.6013 per hour. So that’s $ 53,376 an hour or $ 57,824 to account for the entire duration of the session (65 minutes in all).
“With access to this on-demand supercomputer, researchers were able to analyze and filter 337 million compounds in 7 hours. To replicate that, using their on-premise systems would have taken two months, ”said Colin Bridger of AWS.
What’s amazing is that this kind of firepower is available to anyone who can afford it. And it’s based on the kind of hardware that runs our cloud computing world: web hosting, website builders, cloud storage, email services, among others.
CWM platforms have evolved over the years to develop algorithms and machine learning capabilities to choose the best compute source, regardless of its origin or type.
For example, a cloud provider might have the cheapest point-in-time compute, but that wouldn’t be chosen if it wasn’t available in the customer-defined territory, or if there wasn’t actually a number. sufficient servers of the required instance type available in that cloud provider, in which case another compute source would be chosen.