In many ways, high performance computing is still more of an industry concept than a real strategy among Canadian SMBs.
Yet high performance computing (HPC), data mining and big data are only going to play a bigger role in businesses in the next few years, according to a panel at the Ontario Centres of Excellence (OCE) Discovery conference Monday afternoon.
For Abe Heifets, CEO of Chematria Inc., his company uses HPC to try to develop drugs more quickly and cheaply than what’s seen in mainstream pharmaceutical companies. In the past, he said, pharmaceutical giants had to build several types of prototypes and test them all out. While that was obviously inefficient, it was necessary because computers weren’t advanced enough to model the complex statistics required in creating drugs.
But now, computers have reached the threshold where they can make drugs much more simply through algorithms and data sets – an achievement that Chematria hopes to take advantage of in the future.
On the academic side, researcher Jennifer Flexman of the Sargent Laboratory at the University of Toronto is working with a group of 30 researchers to create commercially viable solar cells. While solar cells are touted as a great alternative energy source, they have been too expensive to compete with fossil fuels, she said. Yet with HPC, the hope is they will be able to build a solar cell that companies can monetize and use as a greener source of energy.
Marketing consultant Wayne Gudbranson of Branham Group Inc. said his clients have found there is so much unstructured data out there, and it has only been in the past few years that they’ve found ways to tap into its power. To even begin analyzing that data, they have needed massive amounts of computational power, he said.
Many of the challenges in implementing HPC are felt across the industry. Panelists agreed that one of the biggest problems of HPC is that it requires skilled workers who know how to mine data for solutions.
Another barrier, said Ted Mao of water treatment company Trojan UV, is that it may be difficult to ensure data is kept private and secure. Companies need to know how to set up reliable firewalls to prevent leaking their data when using HPC, he said.
Still, HPC will only become more and more relevant in the next five to seven years, said Gudbranson.
“Organizations will need to understand data, and then figure out response mechanisms to that data to achieve corporate goals,” he said. “Let me just say this – your [traditional] architectures, within five years, are going to be gone.”