Feng Chen
Indiana University, IN·Computer Science
Computers spend much of their time simply moving data between where it is stored and where it is processed, and that movement has become the main bottleneck for applications that handle enormous datasets. Both projects here attack the same underlying problem from different sides: instead of hauling data to the processor, bring the computation to where the data already lives, whether that is main memory or the storage device.
- Fields
Computer Science
Engineering
- Topics
computational storage
in-memory computing
near-data processing
- Grants
- 2 known grants, through Jun 2027
What they work on
- Develop hardware functionality and system-level memory management for in-memory computing over Compute Express Link (CXL), the interconnect standard that lets processors and memory devices share memory, to reduce fragmentation of memory resources.
- Add architectural support for memory sharing across CXL devices and close the gap between hardware behavior and the software model, so applications can use near-memory computation through general programming platforms.
- Design a service-oriented abstraction through which applications invoke computation on computational storage devices, making the device a cooperative part of the system rather than a standalone accelerator.
- Exploit the proximity of compute to data inside storage devices while mitigating contention for their limited on-device memory, under the devices' tight cost and power constraints.
- Adapt the core data structures and algorithms of representative data-intensive applications so work is divided effectively across heterogeneous host and in-device compute resources, and evaluate the gains on application case studies.
More about the research
The work spans hardware features, operating-system resource management, and programming interfaces, with the goal of making these ideas usable by real applications rather than only in prototypes.
Why it matters
Data-center workloads, from databases to analytics, are increasingly limited by the cost in time and energy of shuttling data between storage, memory and processors. Techniques developed here could let the same hardware process large datasets faster and with less power by computing inside memory or inside storage devices, benefiting operators of data-intensive services and the designers of CXL-based systems and computational storage drives.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Xian-He Sun
Illinois Institute of Technology, IL·Computer Science
Engineer IOWarp, a modular and scalable data management platform for AI-infused scientific workflows, using NVMe SSDs, CXL memory devices and CPU-GPU co-design to reduce data access times on HPC systems.
Mahmut Kandemir
Pennsylvania State University, PA·Computer Science
Characterize diverse HPC workloads on real and simulated computational storage devices to identify which computations benefit from running inside storage, and what performance tradeoffs result.
Simon Peter
University of Washington, WA·Computer Science
Design a programmable memory controller that moves data among memory tiers connected by Compute Express Link based on usage patterns, module age and reliability.
The school
Indiana University: 21 labs on PI Pages, $897M a year on research, 438 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.