Ocient taps AMD EPYC CPUs for performance boost

Hyperscale analytics data warehouse supplier Ocient is collaborating with AMD to use its Gen 4 EPYC CPUs, achieving a 3.5x increase in processing power and more than doubling memory throughput.

Ocient supplies a massively parallel relational database system capable of analyzing petabytes of data, meaning trillions of rows of data in seconds to minutes. The Ocient Hyperscale Data Warehouse (OHDW) features Compute Adjacent Storage Architecture (CASA) to remove network bottlenecks and facilitate rapid data access. A feature of its Megalane tech increases throughput “by enabling an extreme number of parallel tasks.” Data reliability is provided by erasure coding.

Chris Gladwin, Ocient
Chris Gladwin

CEO Chris Gladwin stated: “AI and compute-intensive data analytics workloads are placing immense pressures on datacenters around the globe, meaning efficiencies delivered by hardware and software are essential for enterprise data growth, performance, and costs. We’re thrilled to work with AMD to drive increased customer data performance, cost savings, and operational and energy efficiency.”

Ocient already supports both Intel and AMD CPUs in the processor requirements for its hyperscale data warehouse. Now AMD takes precedence.

The company told us it “has chosen the AMD 9654 with 96 actual cores over the previous generation of Intel’s Icelake 6348 with 28 actual cores. The increased core count, along with the increased all-core clock speed, is what is delivering increased performance. Not only that, but on AMD, this is being done with similar cost, space and power on a per node basis, which in turn allows Ocient to deliver leading edge-performance solutions in a smaller system footprint and with the most cost-effective power requirements.” 

Ocient’s core-based licensing model prioritizes cost efficiency and the performance of compute-intensive workloads. By using AMD CPUs, Ocient says its customers will get:

  • 3x better performance for compute-intensive workloads
  • Reduced operational costs through improved power and energy efficiency, achieving a 3× reduction in power consumption per core
  • Sustained scalability for future AI and analytics needs

It says that, with global datacenter energy consumption projected to more than double by 2026 due to increasing AI and data analytics demands, enterprises are increasingly seeking ways to reduce energy consumption. This move to AMD will help with that.

Kumaran Siva, Corporate VP for Strategic Business Development at AMD, said: “We’re pleased to collaborate with Ocient to bring the performance of AMD EPYC processors together with Ocient’s efficient software solutions, enabling enterprises to better handle their most demanding AI and analytics workloads with ease and efficiency.”

Click here for more info.