Proposed Memory Solution Can Run At Up To 600°C, Fueling Potential Adoption In Servers

Proposed Memory Solution Can Run At Up To 600°C, Fueling Potential Adoption In Servers

With ever-increasing demands on compute power, temperature, especially on-board memory, must be kept in line, and research offers a unique solution.

Next-gen server memory could feature layers of modified materials, replacing traditional silicon carbide and operating at extreme temperatures.

According to research published on Nature ElectronicsFlash drives based on ferroelectric aluminum scandium nitride (AlScN) could be the next-generation standard for the markets, not only because of how cool the name sounds but also because of the temperature durability of the material. University of Pennsylvania researchers Deep Jariwala and Roy Olson have demonstrated a memory prototype that remains active at temperatures up to 600°C, which is said to be twice that of current market alternatives.

Now, the big question is why is such a memory solution needed in the first place? Well, the answer is simple. Integrated with heavy components, the massive compute system generates enormous amounts of thermal energy, which is dissipated on board by the cooling solution, liquid or air. In memory devices, however, due to their volatile nature, out-of-range temperatures can lead to data loss and performance degradation, which is intolerable for large-scale systems. Continuous operation of memory on board becomes critical, even at high temperatures.

Image credit: PennToday

Jariwala says their solutions target large-scale AI systems because they allow them to operate even under harsh conditions. In addition to temperature advantages, AlScN-based memory devices reportedly have an edge over conventional products in eliminating inefficiencies in data transfer between the central processing unit and memory.

While silicon carbide technology is great, it doesn't come close to the processing power of silicon processors, so advanced processing and data-heavy computing like AI can't really be done at high temperatures or in any harsh environments.

Our memory device consolidation may allow memory integration and processing to be more closely coupled, increasing speed, complexity, and computing efficiency. We call this 'Memory-Enhanced Compute' and are working with new teams to set the stage for AI in new environments.

– Deep Jariwala, University of Pennsylvania

Research suggests that future memory standards may evolve rapidly, with fundamental changes used in the underlying material layers, and in this case, AlScN may seem like a no-brainer, but it's too early to say. It is premature.

Share this story.



About the Author

Leave a Reply