Gazing Into the Future of Insight & Discovery: 5 Top Trends at SC24

There isn't a conference as inspiring in the tech sector as Supercomputing (SC24), the annual gathering of the top technical computing centers on the planet, and the industry that supports them with the latest advancements in large scale tech.

While AI has taken center stage in every technology conversation, it's important to remember that the massively parallel computing delivered by HPC clusters are both the precursors of AI training clusters and the machines that have opened up calculations of some of the boldest scientific projects in history.

Here are just a few examples of things HPC clusters are studying today and poised to to advance tomorrow:

– Complex systems represented by climate and Earth's evolution 

– Mapping of weather patterns 

– Peering into the outer reaches of space

–The future of disease management

At TechArena, we are plotting a full lineup of interviews from the leading researchers in the field today, and will be publishing throughout the week on how these institutes are pushing scientific discovery forward with massive scale computing as their tool. We'll seek out insights from national labs across the globe as well as university centers engaged in collaborative studies to propel academic research forward. And, of course, we'll peer into the industry innovation that underpins advancement of supercomputing clusters. What are we looking for? Introducing our top five top trends we're tracking at SC24:

• The AI Infiltration of Supercomputing: It's easy to say that AI has invaded the world of supercomputing clusters. After all, much of the compute infrastructure sought by large AI model developers are also tapped by HPC. What we're seeking here is the integration of AI into supercomputing problem solving, and this is a question of mathematical models and precision. AI, long known as a low precision science, allows for massive acceleration of modeling vs. the tight precision of traditional HPC algorithms (i.e. the hard science). How do these come together? DeepMind's research into protein discovery gives us a map. Imagine the ability to tap AI to narrow down solutions from millions of options to a virtual handful. This allows scientists to narrow their focus for traditional HPC models, accelerating actual discovery by traditional high precision models to the point of making certain discoveries uniquely in reach. We'll be looking to see the applicability of AI integration across scientific spheres to see if further advancement in this space is on the table in the coming half decade, as well as seeing the most innovative integration being driven by scientific computing centers today.
• Network Innovation: We've written about it a lot on TechArena, but the next major advancement in compute cluster performance is removal of the chip-to-chip and node-to-node bottlenecks through further advancement of scale up networks, scale out networks and cabling. The moment of optical integration is upon us, and we're excited to see what the optical industry is delivering towards true integrated optics on platforms. We're also excited to see how the rest of the industry is faring in competing with NVIDIA's strong grip on networks with NVLink and InfiniBand solutions. Earlier this year, AMD released the first Ultra Ethernet adaptor, signaling the rapid delivery of this industry standard alternative for scale out connectivity, and we expect to see more in this space. Will we see highlights from other solutions at SC'24? Watch this space to learn more.
• Keeping it Cool: The liquid cooling industry is coming home to roost at the only conference that's always loved it. Supercomputers have embraced liquid cooling alternatives for years as they feature compute density requiring liquid technology. We expect to see a galvanized liquid cooling ecosystem on hand working to differentiate their solutions within their core market, bolstered by the acquisitions and broader investment that they are enjoying in the AI era. We'll be tracking the news from these players as they seek to further differentiate their offerings in a landscape that is getting more competitive and profitable.
• Efficient Data Management at Scale: We love threes, but we couldn't let this bonus trend sit on the cutting room floor. There's no bigger "big data" challenges than supercomputer calculations, and we expect to see advancements in data management from the major players delivered in part because of infusion of urgency from the AI arena. How will this AI momentum deliver new approaches to HPC data oversight? How will underlying advancements in storage media, most notably the introduction of 122 TB drives from Solidigm signaling the latest wave of storage density advancement, change the game for what's possible? We'll be reporting out this week on this storyline.

Following TechArena: Buckle up and set your eyes on the stars, as TechArena's platform is about to be inundated with inspirational stories of what our sector does best. If you enjoy this coverage, please give us a follow on your favorite social or content platform and subscribe to our newsletter below!