In April 2017, Simon Fraser University launched Canada’s most powerful academic supercomputer, the HPC system, named “Cedar,” after British Columbia’s official tree. The system, which debuted at number 86 on the TOP500 list of the world’s most powerful supercomputers, enables thousands of researchers across Canada to run a wide variety of scientific workloads, including those related to AI, machine learning, deep learning, personalized medicine and green energy technology.
SFU bioinformatics and genomics professor Fiona Brinkman, who leads the university’s Integrated Rapid Infectious Disease Analysis Project, uses the sophisticated and secure computer power of an HPC system from Dell EMC to understand disease outbreaks.
To understand more about the computing power behind the analysis, Digital Journal met with Thierry Pellegrino, vice president of High Performance Computing, Dell Technologies.
Digital Journal: How was Canada’s most powerful academic supercomputer – “Cedar” – developed?
Thierry Pellegrino: The road to Cedar, Canada’s most powerful academic supercomputer, began when Compute Canada, an independent not-for-profit corporation that provides research computing resources to researchers across the nation, put out a call to its member universities inviting them to apply to host a computing facility.
Simon Fraser University won the bid to host a new data center and a world-class general-purpose supercomputer, which would become known as Cedar. They then partnered with Scalar Decisions, a Canadian IT solutions provider, and Dell Technologies to design and build the machine. A lot of the evaluation criteria for building the supercomputer revolved around the question of whether the Scalar Decisions could come up with a design that suited the needs of this very diverse set of researchers, not just at Simon Fraser University but all across the country. They ultimately chose to build the computer using Dell EMC’s infrastructure.
This included:
902 Dell EMC™ PowerEdge™ C4130 and C6320 servers
27,696 Intel® Xeon® CPU cores for computation
584 NVIDIA® P100 Tesla® GPU accelerators
An Intel® Omni-Path Architecture (Intel® OPA)
next-generation fabric
11.4 petabytes of storage
DJ: How does the computer stand in the global list of supercomputers?
Pellegrino: Cedar is one of the top 100 supercomputers worldwide. It has more than 3.6 petaFLOPS of computing power and nearly 28,000 CPU cores, placing Cedar 86th on the TOP500 list of the world’s most powerful supercomputers. Cedar and its network will allow thousands of researchers from British Columbia and across Canada to collect, analyze, share and store immense volumes of data. The supercomputer builds on Simon Fraser’s growing strength in big data.
DJ: What is an HPC system?
Pellegrino: High Performance Computing (HPC) most generally refers to aggregating computer power in a way that delivers much higher performance than one could get out of a typical desktop computer. A helpful way to understand what high-performance computers are is to think about what is in them. All the elements of a desktop computer— processors, memory, disk, operating system — are part of the supercomputer but there are more of them. HPC systems are really clusters of computers, often referred to as supercomputers because they run advanced application programs efficiently, reliably and quickly. This most notably helps solve large problems in science, engineering or business.
DJ: What will the supercomputer contribute to the development of AI?
Pellegrino: Supercomputers are able to run a wide variety of scientific workloads. They enable researchers like Simon Fraser’s bioinformatics and genomics professor Fiona Brinkman, who leads the university’s Integrated Rapid Infectious Disease Analysis Project, to use the sophisticated and secure compute power of an HPC system from Dell EMC to understand disease outbreaks. That system, named “Cedar,” is designed to link different sets of data, including those related to AI, machine learning, deep learning, personalized medicine and green energy technology together. This not only helps Professor Brinkman’s team study DNA codes but also to further teach the computers what to look for.
Brinkman uses Cedar to track pattern recognition, understanding more and more about the data, linking the data through sophisticated methods of analysis and machine learning.
DJ: What will the supercomputer bring to personalized medicine?
Pellegrino: As HPC becomes more inclusive through hardware and cloud advancements, we are now at the point where personalized medicine can take the steps necessary to move from concept to reality. By combining the work of clinical researchers and hospitals with the biomedical community, we can help physicians embrace upcoming technologies to better diagnose and treat the patient.
Looking at what Simon Fraser University is doing alongside researchers across Canada, using HPC they can aggregate data, enabling them to pinpoint specific bacteria lifecycles. In the same way, physicians using HPC systems are also able to aggregate data enabling them to prescribe treatments tailored to an individual’s genetic makeup. While Simon Fraser is looking at disease outbreak, others are using Dell’s infrastructure in HPC systems to personalize medicine and healthcare practices.
DJ: How about the applications in terms of green energy?
Pellegrino: Cedar is an extremely energy-efficient system. Its power utilization efficiency (PUE) ratio is 1.07, meaning that very little power is wasted. Of all the energy that flows into Simon Fraser’s Burnaby data center, 93% goes to running the supercomputer and just 7% goes to cooling and other things that consume energy.
The remarkable PUE — far lower than that of the typical data center — stems largely from innovative approaches to cooling. Cedar is designed to use the environment on Burnaby Mountain as efficiently as possible, and the team at Simon Fraser accomplished this by using a method that’s called evaporative cooling. During cooler times of the year, heat exchangers outside the data center use the ambient air to cool glycol, the fluid used in the cooling system. The cool fluid is then piped through the data center. On hotter days, the cooling system sprays water on radiators, which evaporate the water. That process removes heat from the data center, much the way our bodies remove heat when we perspire.
These innovative approaches to cooling have helped the university earn accolades for sustainability. Cedar made its debut at number 13 on the GREEN500 list of the worlds’ most energy-efficient supercomputers.
DJ: Can the supercomputer contribute to reducing disease outbreaks?
Pellegrino: Simon Fraser University’s Integrated Rapid Infectious Disease Analysis Project is using the supercomputer to study the DNA code in infectious disease microbes and to understand how diseases are spreading and ways to better track them. In this case, Cedar is like a DNA detective for infectious diseases and must sort through the billions of base pairs, or nucleotide letters, in the DNA of bacteria and analyze that code of life. Researchers use this like a fingerprint to track how an infectious disease is changing in different locations, and how it is spreading over time.
