# Collaborative Research: DESC: Type II: Enhancing Data Center Efficiency through Intelligent Tiered Memory Management

> **NSF 01002526DB NSF RESEARCH & RELATED ACTIVIT** · Regents of the University of Michigan - Ann Arbor (MI) · $1,000,000

## Abstract

Modern data centers rely on large-scale memory systems to support everything from search engines to scientific computing. However, the increasing demand for memory leads to significant resource and energy use, particularly when older hardware is discarded before its useful life ends. This project addresses that challenge by developing MemWise, an intelligent system that efficiently manages memory across a range of hardware types. The project’s novelties are a programmable memory controller that coordinates how data is moved and stored across different generations of memory; a technique for reusing older memory modules alongside newer ones to reduce cost; and new fault-tolerant methods to ensure reliability even as memory components age. The project's broader significance and importance are in demonstrating that thoughtful system design can extend hardware lifespans and reduce waste, without sacrificing performance. MemWise is tested in a range of real-world environments, including public cloud services, private infrastructures, and direct-to-hardware computing setups, highlighting its adaptability and impact across the data center ecosystem.

Specifically, the project designs a tiered memory architecture connected via a flexible hardware interface called Compute Express Link (CXL). A programmable controller dynamically moves data among memory types based on usage patterns, age, and reliability. To guide decision-making, the research introduces a new metric that jointly con

## Key facts

- **NSF award ID:** 2450085
- **Awardee organization:** Regents of the University of Michigan - Ann Arbor (MI)
- **SAM.gov UEI:** GNJ7BBP73WE9
- **PI:** Mosharaf Chowdhury
- **Primary program:** 01002526DB NSF RESEARCH & RELATED ACTIVIT
- **All programs:** —
- **Estimated total:** $1,000,000
- **Funds obligated:** $1,000,000
- **Transaction type:** Standard Grant
- **Period:** 09/01/2025 → 08/31/2029

## Primary source

NSF Award Search: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2450085

## Citation

> US National Science Foundation, Award 2450085, Collaborative Research: DESC: Type II: Enhancing Data Center Efficiency through Intelligent Tiered Memory Management. Retrieved via AI Analytics 2026-06-08 from https://api.ai-analytics.org/grant/nsf/2450085. Licensed CC0.

---

*[NSF Awards dataset](/datasets/nsf-awards) · CC0 1.0*