Compute.ai has patented and patent pending technologies including AI/ML driven algorithms that target the above problems. For example, our Just-In-Time (JIT) Spill to disk algorithms do dynamic demand paging of arbitrary sized data pages to lower tiers of memory in fractions of milliseconds as a part of its SQL CBO (Cost Based Optimizer) strategiess
Compute.AI has targeted some of the pressing technological needs of the day. These include increasing the compute efficiency while not sacrificing performance, opening the floodgates on concurrency, and increasing performance by over 10x for complex compute. In other words–we make compute abundant and infinitely scalable.
Our technologies help analyze fast paced data with very high efficiency, fast time to value, and at a low cost to make business solutions viable. We do not imagine 7 billion humans typing ad hoc SQL or running batch jobs, but a future where machines driven by AI/ML will be emitting SQL and complex compute. Gone also are the days of BI dashboards built by humans but instead powered by autonomous semantic layer creation alongside the massive low-code and no-code movements.
THE RIGHT TIME
The industry’s love for in-memory architectures has served us well up until now and allowed us to scale horizontally with elastic clusters. However, memory hungry apps come with their tradeoffs. Scaling with memory also means more CPU cores that come along with the memory in cloud instances.
Utilizing these cores is not trivial as consistency and concurrency fight each other and many relational algorithms, including data skew management, are not naturally parallelizable. These challenges along with the cost of node to node network latencies results in roughly 7 out of 10 cores being idle for SQL based real-world workloads. Additionally, the highly ephemeral and time sensitive nature of data processing has exacerbated the core to memory ratios of CPUs resulting in high memory latency (starved CPU cores writing for data to be available in memory) for analytics workloads.
Already today, compute infrastructure struggles under the load of relational SQL from BI workloads. AI is poised to drive orders of magnitude more complex compute demand at unprecedented levels of concurrency driven by AI Powered BI, Autonomous Semantic Layer creation, and Low-Code/No-Code applications.
Latest News & Articles
Joel Inman and Vikram Joshi discuss generative AI and the role of compute with John Furrier, CEO of Silicon Angle Media
A deep dive into database architecture, opportunities, challenges, shortcomings and where Compute.AI fits in to solve these challenges
Presentation to the Deep Tech Showcase Forum on Compute.AI’s Innovative Technology
Joel Inman, CEO at Compute.AI joins theCUBE for a discussion of AI and the Semantic Layer at Supercloud 3