According to a news release, Purdue's Anvil supercomputer, funded through a $22 million grant from the National Science Foundation, is the largest-capacity computational system the university has ever developed, and the most technically diverse. The tool is now being used by researchers at Purdue, as well as 20 other universities.
“Its various components are all integrated in one place with GPUs (graphics processing units) and large memory nodes complementing a cluster of 1,000 compute nodes. As Anvil grows, it is also able to take on much more heterogeneous workflows that are more common in the research happening today at Purdue,” Carol Song, the principal investigator for the Anvil project, explained in a public statement.
Song told Government Technology that the supercomputer was assembled and developed last year for early user testing, which began in November 2021 before Purdue researchers met with the National Science Foundation’s review panel in January 2022 to begin production operations in February.
She said the Anvil supercomputer is now available to science and engineering researchers across the country for large-scale simulations and sharing research for replication across institutions and cohorts, among other uses.
“We have developed specialty computing capabilities, such as large memory nodes for applications that need to load in a lot of data into memory at once,” she said. “We also have a subsystem of GPUs for things like data-driven analytics, machine learning and AI applications. Along with all these computing elements, we have a very large storage system.
“Users are beginning to embrace these systems because as they look at problems in finer resolutions and in larger scales, a lot of [this research] requires a lot of computational power, as well as an ability to analyze data as part of their workflow with simulations and modeling.”
According to the university, Jonathan Poggie, a professor of aeronautics and astronautics, is exploring how to use Anvil for research into predicting and controlling aerodynamic heat from flight at high-Mach speeds. In addition, the university said a research group led by Purdue computer science professor Daniel Aliaga is looking into how Anvil’s graphics-processing power might “facilitate the creation of ‘what-if’ design tools in digital city planning.”
Aside from the ways Anvil’s capabilities can accelerate research such as this, Song said it allows researchers to share their work with one another more quickly than before.
“After they do their computations, researchers very often want to share their results. They need to let other people reproduce their work to repeat what they did and to confirm that their method works. For all of that, they need to share their code in a way that other people can easily get up and run and repeat their experiments,” she said. “Without these systems, researchers would have to hire people or have this kind of software expertise in their group to do these things. This simplifies how they share their applications, software data codes and data sets.”
The university said the supercomputer has been used in health research, such as work at the University of Kansas, where assistant professor Yinglong Miao is using Anvil for biomolecular simulations to discover remedies for heart disease.
“Using CPUs, this work could take months, if not years,” Miao said in a public statement. “With the GPUs on Anvil, we can run these simulations much faster. Instead of needing months, we just need a couple of weeks.”
Song said Miao’s experience using Anvil for his research echoes that of others who have tested the supercomputer for their research. She said Anvil’s hardware, in terms of processors and its high-performance cloud system, has been “tremendously helpful in accelerating people’s work.”
“We got comments like, ‘This is worth the wait. Now I’m getting double the speed of what I had on other systems.’ It’s great to hear that we’re achieving the goals we set out to achieve,” she said.
Now with the supercomputer open to researchers, Song said it’s time to familiarize others with the technology and its potential.
“Going forward, we have a couple of challenges. One is to get more people on here, especially those who could benefit from a large system like this, but they traditionally are not familiar with these systems,” she said. “With those tools, we’re hoping to help people ramp up more quickly and get people who haven’t used these systems to leverage them.”
