Today at its developer conference, Google I/O, Google showed the cloud engine directing 10,000 individual computing cores toward a gene-sequencing app, reducing the 10-minute time to complete a task to mere seconds.
However, for apps that require even more computational power, and don't require much interaction with the data through input and output, Google Compute Engine can work overtime. In the onstage demo, Google said it was able to direct as many as 771,886 computing cores at a specific task -- the total number of cores Google has available for the service, which continually increases.
Using 600,000 of those cores, the gene-sequencing app completed several of those 10-minute tasks every second.
Google Compute Engine is similar to a service that Amazon offers, which employs unused cores in its EC2 cloud (a.k.a. Amazon Web Services) as a supercomputing service that businesses can buy time on. Amazon's EC2 is actually one of the top 500 fastest supercomputers in the world.