Accelerating AI with Computational Storage
Eideticom brings the first standards-based solution to market
From an observation to a guideline, ‘Moore’s Law’ has been a driving force for technological change, productivity, and economic growth for over half a century. We’ve held fast to the notion that computer processing power increases exponentially every two years, but experts now agree that we’ll reach the physical limits of Moore’s Law within the next ten years. This is a worrying concept during this data-driven digital age.
Over the past several years, applications have become increasingly data-heavy, as AI has proliferated through IoT smart devices and mobile apps, as well as deep tech in the areas of health, mobility, smart cities, commerce and beyond. Fortunately, smart folks like Roger Bertschmann (Inovia Advisor for 10 years) and Stephen Bates, Co-Founders of Calgary and San Francisco-based Eideticom, have been preparing for the winter of Moore’s Law for years. As serial founders and semiconductor veterans from PMC Sierra and Seagate, Eideticom’s core team has the foresight and technical skills to bring cutting edge storage technologies to market. When Inovia Capital community members come along who have deep domain expertise, and who are tackling fundamental problems, we love to roll up our sleeves and get involved. As such, today we’re proud to announce our investment in Eideticom alongside Molex Ventures to help Roger, Stephen, and the team scale capacity and accelerate deployment of end solutions.
This year, Eideticom became the first — and so far, only acceleration platform to implement a standards-based computational storage architecture. They have produced the first acceleration card which leverages the NVMe protocol that can be dropped into an existing system. This puts enterprise ops and cloud managers in a position to provide additional benefits, at a lower cost, to their end users.
Behind the Tech
Eideticom’s new NoLoad® Computational Storage Platform (CSP) disaggregates compute and storage into independently scalable resources, thus providing greater scalability, performance and cost efficiency. By shifting data processing closer to the storage unit and onto an FPGA (avoiding the CPU all together), Eideticom’s solution increases performance while reducing power consumption by a factor of up to 10x. In addition, NoLoad® CSP enables real-time data protection and analysis, reduces energy consumption and improves application performance through having reduced input/output (I/O) bottlenecks. It accelerates storage workloads, including erasure coding, deduplication, compression, and encryption. NoLoad® CSP is compatible with standard server architectures, integrating directly into software stacks like Hadoop, RocksDB, ZFS, and others, giving it significant base operational capabilities from monitoring to hot-swapping.
Eideticom is solving the root cause of a valuable problem. As the team and their solutions scale, they will enable advancements down the line, impacting the entire ecosystem. In theory, machine learning applications could process more, faster, and at lower costs — and this is great news for us all. As a long-time Advisor to Inovia, we know and believe in the operational mindset of Co-Founder Roger Bertschmann, and are excited to help the team bring their vision to life.
Learn more and become part of the early adopter program at Eideticom.com.
