By Carolyn Doyle
One of the most powerful GPU servers on the IU Indianapolis campus is poised to take research at the Luddy School to new heights, thanks to hard work by many members of the Luddy Indianapolis community, says Sarath Janga, Ph.D.
“The Kailash server represents a transformative leap for biomedical research, placing IU Indianapolis at the forefront of biomedical innovation,” says Janga, associate professor of biomedical engineering and informatics at IU’s Luddy School of Informatics, Computing, and Engineering in Indianapolis.
A graphics processing unit speeds up lengthy tasks in graphics programming, high performance computing, deep learning, and other research tasks.
Ensuring high-speed data transfer and connectivity, the special configuration of the server from Kailash Technology makes it the most powerful GPU server on the IU Indianapolis campus, says Janga, except for the University Information Technology Services supercomputers.
An impressive tool for research
“This new state-of-the-art GPU cluster marks an important step towards increasing our research capabilities,” says Shiaofen Fang, associate dean for research, “empowering students and faculty to tackle complex computational challenges in AI, data science, and other research areas.”
Fang, professor of computer science at Luddy Indianapolis, adds, “The new cluster opens doors to new potentials in innovative discoveries and solidifies Luddy’s position at the forefront of academic excellence and technological advancement.”
The server was made possible by a major contribution from Janga using a National Institutes of Health grant’s equipment budget as a base fund, he explained, which led to a matching from IU’s internal Research Equipment Fund.
Several others also contributed to the cost of this server, Janga says, including professors and Ph.Ds. Mohamed Abdel-Mottaleb, Rakesh Shiradkar, and Lu Zhang. Faculty including Fang; Luddy School Executive Associate Dean Davide Bolchini; Hyeju Jang; Hee-Tae Jung; Juexin Wang; and Bioinformatics Program Director Jingwen Yan, all Ph.Ds., wrote letters in support of the project.
Total funding for this project is $328,665.
“The Kailash server will be available to all Luddy faculty and staff, and their students/teams for research-oriented projects,” Janga notes, with those who contributed financially to the project being given some scheduling priority.
Doing more in less time
The new server empowers researchers to model complex biological systems, train deep learning algorithms for disease diagnosis, and analyze massive genomic datasets with unprecedented speed, says Janga, who leads the Janga Lab of Genomics and Systems Biology. Additional research areas where this server could make a significant difference include epidemiology and public health modeling, structural biology, medical imaging and AI-assisted diagnostics.
“Its unparalleled computational power, driven by 8 NVIDIA H100 GPUs and 4TB of memory, will accelerate discoveries in drug development, precision medicine, and large-scale biological data analysis, while its 240TB high-speed storage and cutting-edge networking can facilitate real-time collaboration and high-throughput data processing,” he says.
“With Kailash’s advanced capabilities,” Janga says, “IU Indianapolis researchers can lead groundbreaking biomedical discoveries, positioning the institution as a powerhouse for computational biosciences and AI-driven health care innovation.”
Promoting interdisciplinary achievement
With its focus on interdisciplinary research, Luddy Indianapolis is uniquely placed to use this new resource for projects across many fields. Janga and other faculty envision the potential for the server to enable breakthroughs in AI-driven health care, genomics, precision medicine, and molecular dynamics simulations.
Drug discovery is an area where the new GPU server can make a real difference at Luddy Indianapolis, says Juexin Wang, assistant professor of bioinformatics.
“This new GPU server helps to unlock the unmet desire for computational power performing cutting-edge research in bioinformatical foundation models,” Wang says.
“This new investment enables Luddy researchers and students applying, fine-tuning, and even training state-of-the-art models, to tackle several prominent questions in AI for science fields, including drug discovery and complex system modeling.”
Hyeju Jang, assistant professor in the Luddy School’s Department of Computer Science, notes, “The server will be very useful for my natural language processing projects, which use large language models and also large vision-language models.
“We are working on human language processing for various applications, like dementia detection and investigating social media data for health care issues.
Jang and her students also are doing projects where they work to detect and understand visual metaphors. “This server will help students do many experiments for these projects using large language models, and also large vision-language models,” she says.
The new server also will be a valuable asset in biomarker research, notes Rakesh Shiradkar, assistant professor of biomedical engineering and informatics at the Luddy School.
“The Kailash server will enable discovery and development of biomarkers from multi-modal medical imaging datasets,” Shiradkar says, “including radiology and digital pathology, along with genomics and electronic health records.
“These biomarkers can ultimately enable development and translation of AI tools into the clinic for improving healthcare including cancer. Foundation models are the next frontier of AI, and this resource will help IU Indianapolis establish as one of the top destinations for health care AI research.”
Media Contact
Joanne Lovrinic
jebehele@iu.edu
317-278-9208