When you think about scientific breakthroughs, you probably picture massive telescopes, particle accelerators or sophisticated AI models. But there’s an unsung hero enabling these advances: memory and storage technology. At this year’s Supercomputing Conference (SC24), Micron demonstrated how innovations in memory and storage are fundamentally changing what’s possible in scientific discovery.
HPC Around the World Map
Addressing the memory and storage bottleneck in particle physics
While the Large Hadron Collider (LHC) at CERN usually collides protons, once a year or so it switches to colliding “heavy ions” for several weeks. Recently, the LHC switched to this “data taking mode,” and once again began colliding lead ions, which each contain 208 nucleons. These collisions create a quark-gluon plasma, a hot dense soup of particles, that mimics the conditions of the very early universe.
The Compact Muon Solenoid (CMS) experiment at the LHC measures the particles created by the collisions. During normal proton-proton collisions, the CMS typically collects data at around 10 GB/s. But during heavy ion collisions, data collection increases to a sustained 30 GB/s. To assist in this high-speed data collection, CERN has recently deployed an array of Micron 7500 SSDs.
In parallel, CERN is exploring the use of Micron CXL-based memory expansion modules to support more advanced event-selection algorithms, enable new detector diagnostics, luminosity measurements, and the study of otherwise inaccessible physics signatures.
The collaboration between Micron and the CMS experiment is facilitated by CERN openlab. CERN openlab is a unique public-private partnership, forging collaborations between leading ICT companies and research centerr worldwide, uniting them with the forefront of scientific innovation at CERN.
SC24 Micron booth
Breaking down barriers in computational chemistry
While particle physics demands speed, computational chemistry requires both speed and precision. Traditionally, running complex molecular simulations required access to multimillion-dollar supercomputers, putting this research out of reach for many scientists. Through a groundbreaking collaboration with Pacific Northwest National Laboratory (PNNL) and Microsoft, Micron is rewriting these rules.
By combining Micron’s latest memory architectures with Microsoft’s cloud infrastructure and PNNL’s scientific expertise, researchers can conduct sophisticated molecular modeling and chemical analyses on standard cloud computing platforms. This democratization of computational chemistry could accelerate drug discovery, materials science and clean energy research — fields where simulation costs have traditionally limited innovation.
Helping researchers and developers navigate the AI landscape
Managing the AI landscape is difficult. The environment is changing so rapidly it’s difficult for even those focused specifically on AI to keep up. Imagine the challenge for researchers who have their own areas of investigation underway yet want to leverage the benefits of AI. A group of researchers at the Barcelona Supercomputing Center recognized the problem and took the challenge to heart.
Partnering with Micron, they developed an open-source platform, FAiNDER, that is designed to transform how researchers and developers navigate the rapidly evolving AI landscape. It provides centralized, up-to-date information on the key system requirements of all major AI models to facilitate exploration and optimize hardware choices, especially regarding the memory resources needed for their efficient execution.
Reshaping scientific computing through revolutionary memory technologies
At SC24, Micron showcased two groundbreaking memory innovations that promise to reshape scientific computing.
First, the powerful combination of CXL™ (Compute Express Link) technology and Micron’s open-source fabric-attached memory file system (FAM-FS) represents a fundamental shift in how scientific computing systems handle memory. This breakthrough allows researchers to dynamically scale memory resources independently of computing power — similar to how cloud computing revolutionized processing power allocation. Early testing shows this approach could reduce infrastructure costs while enabling previously impossible analyses of massive scientific datasets. For fields like genomics and climate modeling, where memory constraints often force compromises in research scope, this technology opens new possibilities for comprehensive, large-scale studies.
Second, our award-winning MRDIMM technology achieved a landmark in computational fluid dynamics (CFD) simulation. The demonstration showed unprecedented capability to model complex turbulent flows with billions of mesh points in real time. This breakthrough has immediate applications in aerospace design, weather prediction and clean energy research, where understanding fluid dynamics at this scale could accelerate innovation and improve prediction accuracy by orders of magnitude, and at level of detail that typically requires weeks of processing on traditional systems.
Shaping the future of scientific discovery
The advances showcased at SC24 underscore a crucial reality: Memory and storage technology isn’t just supporting scientific progress — it’s actively shaping what’s possible in modern research. From enabling real-time analysis of particle physics data to democratizing access to computational chemistry, these innovations are breaking down long-standing barriers in scientific discovery.
As we look to the future, the partnership between memory and storage technologies and scientific research will become even more critical. The next wave of breakthroughs — whether in climate modeling, drug discovery or artificial intelligence — will depend not just on brilliant scientists and powerful algorithms but also on continued innovation in how we store, access and process the massive datasets that modern science generates.
The silent revolution in memory and storage technology may not grab headlines like quantum computers or fusion reactors, but it’s fundamentally changing how we explore and understand our world. As these technologies evolve, they promise to unlock new frontiers in scientific discovery, making the impossible possible and the expensive accessible. The future of science isn’t just about generating new data — it’s about transforming that data into knowledge that can help solve humanity’s greatest challenges.
