Datanami
EnterpriseAI
HPCwire Japan
QCwire
HPC & AI Wall StreetThe advent of low-power mobile processors and cloud delivery models is changing the economics of computing. But just as an economy car is good at different things than a full size truck, an HPC workload still has certain computing demands that neither the fastest smartphone nor the most elastic cloud cluster can fulfill.
While some supercomputers today use low-power processors that were originally designed to power smartphones, that does not mean that a smartphone is a supercomputer. It should go without saying, but people really love their phones today and imagine they posses super powers.
“I often hear the saying, ‘My cellphone is a supercomputer because it’s just as fast, or faster, than a supercomputer was 30 years ago,'” says Barry Bolding, Cray’s vice president of storage and data management and corporate marketing, in a blog post on the Cray website.
Not so fast, Bolding says. “Believe it or not, you still cannot predict a tornado on a cellphone; it just won’t work. You need a supercomputer for that type of scientific prognostication. While today’s smartphones are, well, smart, they’re not that smart.”
The same holds true for cloud clusters, Bolding says. While today’s new cloud clusters can amass much more computing power than an iPhone–and perhaps deliver more CPU cycles than some supercomputers–they still don’t work well for HPC workloads. It all comes down to finding the best tool for the HPC job, Bolding says.
The Cray exec says that HPC systems like the Cray XC30, Cray XC30-AC, and Cray CS300 are targeted at heavy-duty scientific and engineering workloads. “What makes them supercomputers is that they can do work that cloud-clusters either cannot do, or are clumsy and inefficient for,” Bolding says.
“Weather prediction, turbulent airflow in aircraft engines, and high resolution seismic modeling are supercomputer workloads,” says Bolding, who has been with Cray for 21 years and holds Ph.D in chemical physics from Stanford University. “Grinding out the most accurate forecast every few hours, simulating accurate combustion and investing in the right oil/gas drilling site are all critical, production workloads. This is where Cray’s computing solutions shine.”
Bolding uses a car analogy to illustrate the differences. A ZipCar may be great for somebody who only occasionally needs a car, but it doesn’t make sense for somebody who needs a car everyday. If you need to move lots of people but still don’t want to buy a car, a bus (i.e. cloud cluster) provides lots of capacity at low cost, he says.
For somebody who needs their own car everyday, neither the bus nor the ZipCar make much sense. Once one decided they need a car, there are many to choose from. A Toyota Prius, for example, excels at economical transportation of people, while a Toyota Tundra excels at transporting lots of stuff. Supercomputers are similar.
“Fit the right computing model to the right application and you will be happy. Try to fit a production supercomputing application into a cloud-cluster and you will be sorely disappointed,” Bolding concludes. “Remember, the supercomputer of 30 years ago is a terrible cellphone and the cloud-cluster of today is not a supercomputer.”
Related Articles
TACC Longhorn Takes On Natural Language Processing
