Hwu's MulticoreWare ranks in top 110 fastest-growing private companies
Since Professor Wen-mei Hwu co-founded MulticoreWare six years ago, the company has grown from a small startup to the leading provider of GPU-accelerated software and GPU-computing infrastructure software, used by such tech giants as Amazon and Google.
MulticoreWare was recently recognized by Inc.com as one of the fastest-growing, privately held companies in America (ranked No. 110 out of 5,000). Inc.com takes growth data from nominated companies and ranks the top 5,000 according to the percentage growth of their annual revenue over a three-year period.
The top 500 companies were highlighted in the September issue of Inc. magazine. MulticoreWare saw a three-year growth of 3,322 percent, with revenue increasing from $224,601 in 2010 to $7.7 million in 2013.
“It’s really about how we make the best use of parallel hardware and increase the application capabilities and efficiency for customers,” said Hwu, about the company’s growth.
In 2008, Hwu teamed up with GDA-Technology CEO A.G. Karunakaran and Ageia Technologies co-founder Curtis Davis to start a company that took advantage of the fast-growing parallel computing world. The small team of people licensed three pieces of Hwu’s technology from his lab at Illinois and began MulticoreWare, which has now grown from four people to about 215 people in seven offices worldwide.
“We started with the technologies we had at Illinois and began to put commercial grade engineering work into them,” said Hwu, who is MulticoreWare’s chief technology officer and the AMD Jerry Sanders Chair in Electrical and Computer Engineering at Illinois. “We began working with semiconductor companies that are producing parallel hardware and were able to gain several fairly large contracts with the major industry companies in the beginning.”
MulticoreWare offers a variety of services including image processing in commercial printers, rapid image processing for semiconductor inspection, video processing for video service providers, and bio-informatics algorithms for genome research.
In addition to being a leader in the GPU-accelerated software world, in 2012, MulticoreWare became the leading provider of parallel software to video service providers, such as Amazon, Netflix, and Google, after successfully partnering with a video service provider for the 2012 Olympic Games to supply parallel transcoding software for video broadcasting.
While MulticoreWare now has two offices in China, as well as Taiwan, St. Louis, California, and India, the core of the technology still lies in Champaign. When MulticoreWare is offered a major project from a customer, the solution architects in Champaign develop a quick prototype that demonstrates the feasibility of the project and sets up a framework for a larger engineering team to fully develop.
MulticoreWare now employs several of Hwu’s former graduate students at its various offices, including Deepthi Nandakumar (MSEE '11), who is the technology lead at the India office.
“Research at the University of Illinois Electrical and Computer Engineering department and the Coordinated Science Laboratory and their graduates are a major factor of MulticoreWare’s growth,” Hwu said. “This is a testimony to the economic impact of our research program.”
According to Hwu, in the past three years, MulticoreWare has taken much of the technology created in Hwu’s university labs and developed it commercially. This process has allowed insights into how the technology is used in practice, what to adjust for in the future, and where the next generation of technology needs to be
“The technologies we’ve been working into the company are somewhat mature now, and we understand them,” Hwu said. “We see the industry going through another big transition in the next three to five years, so we will need to come up with another wave of technology to grow with that.”
Looking forward, Hwu is confident the company will continue to succeed and take the industry to the next level.
“We have the capability to support different types of applications that people want for parallel systems and we’re the leading provider of these critical components to all the semiconductor companies,” Hwu said. “We build the software on top of tools and compilers, so we can provide a level of efficiency that very few other people can. We all know there is still a long way to go and now we’re focusing on that next level.”