Intensivate Debuts IntenScale™ Clustered Computing Accelerator

BERKELEY, Calif. - Sept. 24, 2019 - PRLog -- Silicon Valley based startup Intensivate Inc. debuts its clustered computing accelerator technology for data center and edge deployments. The accelerator card requires no changes to a user's software and expands the data center by 12x for all cluster-ready applications - those that distribute their workload across a number of servers.

"Big data workloads are very important in the cloud. Intensivate's focus on accelerating these workloads without code change promises to deliver an exciting level of efficiency not possible with other accelerator technologies and products," said Mike McGrath, Director & Technologist at Microsoft Azure.

Applications running on clusters typically require high memory bandwidth and utilize as many CPUs as can be made available. But existing general purpose computing has been architected for traditional software and high single threaded performance, resulting in massive inefficiencies for cluster-ready applications. The Intensivate micro-architecture, IntenCore™, was designed for these workload characteristics and optimizes performance, power and cost simultaneously, resulting in 12 times more computation at the same total IT budget.

Intensivate was awarded a grant from the National Science Foundation to commercialize initial scientific research work. The company secured their first round of funding in 2018 and currently raising an additional financing round to complete chip tapeout and deliver the final product to market.

"For these kinds of applications, software development cost is king. The chip was designed from the start to boot Fedora and be a rackmount server with equivalent performance, but far lower power, lower cost and smaller footprint, in order to fit more within the same TCO," said Dr. Sean Halle, CEO / CTO and Co-Founder of Intensivate. "We succeeded, with customer code running on our implementation, unchanged.  The key is focusing on memory behavior and taking advantage of the many threads available, through starting more worker instances on the same node."

Intensivate's product consists of a PCIe card equipped with 21 high performance, ultra-low power processing chips, each accompanied by 32GB of memory. The card totals 672GB of memory and plugs into a standard 1U server. Applications view the card as a sub-net of 21 servers, with the Host acting as a router.

Rick Anderson, CIO at Inuvo Inc. explains what this means for them. "Mobile is a major growth area we'd like to move into, but adding this real-time processing would increase our compute load from 185,000 transactions per second to 700,000. Using conventional servers alone, our data center costs would increase substantially. We're working with Intensivate because we believe their technology is the only path to handling this much higher load without increasing costs."

A key technology choice for Intensivate was leveraging the open source RISC-V ecosystem. Intensivate's patented microarchitecture can be placed behind any instruction set, but the RISC-V phenomenon allows for a stunning reduction in development cost and time to market.

First Silicon is scheduled for early 4Q20. PCIe accelerator cards for first customer shipment is early 2021.

About Intensivate
Intensivate is a developer of high performance, low power server acceleration products for applications running on clusters. Intensivate's accelerator card provides a 12x boost in computational power with no added data center costs and no required software changes for the user. For more information please visit www.intensivate.com