Semiconductor startup Tachyum Inc. announced today that it has joined the OpenBMC Project as a contributor to the community seeking to define standards for a standard Baseboard Management Controller (BMC) firmware stack that will work across heterogeneous systems, including enterprise, HPC, telco and cloud-scale data centers.
BMCs are specialized controllers that monitor the state of a computer or hardware, including aspects such as system health, log events for failure analysis, and a range of remote management capabilities. Tachyum will be among the contributors from across the industry helping to define and create the OpenBMC stack to ensure the greatest access and control to those involved in the management of remotely deployed server systems.
“Though the BMC technology is not a new one, proprietary approaches to hardware and software has made its advancement more difficult as a whole across heterogeneous systems and computing environments today,” said Dr. Radoslav Danilak, Tachyum founder and CEO. “With Tachyum developing its Prodigy® Universal Processor to serve as a replacement for the majority of existing chips provisioned in hyperscale data centers, we felt it was important to be a part of the community developing standards for an OpenBMC stack that will be adopted by the majority of the industry. We look forward to contributing with our fellow participants on this significant project.”
Tachyum is working to provide its OEM/ODM and system integrators with complete software and firmware for motherboard reference designs, including UEFI and BMC firmware for the Tachyum Prodigy Universal Processor. Enabling fast and wide deployment of the Tachyum technology, which recently achieved a verified physical design of more than 90 percent of the design silicon area, is the key focus of the company.
Prodigy, the company’s 64-core flagship product, is scheduled for high-rate production in 2021. It outperforms the fastest Xeon processors at 10x lower power (core vs. core) on data center workloads, as well as outperforming NVIDIA’s fastest GPU on neural net AI training and inference. Due to its high computational density and I/O bandwidth, networks of Prodigy processors comprising just 125 HPC racks, can deliver an ExaFLOPS (a billion, billion floating point operations per second) of capacity. Prodigy’s 3X lower cost per MIPS compared to other CPU competition, coupled with its 10X processor power savings, translates to a 4X reduction in Data Center TCO (Annual Total Cost of Ownership: CAPEX + OPEX). Even at 50 percent Prodigy attach rates, this translates to billions of dollars per year in real savings for hyperscalers such as Google, Facebook, and Amazon.
Since Prodigy can seamlessly and dynamically switch from data center workloads to AI or HPC workloads, unused servers can be powered up, on demand, as ad hoc AI or HPC networks – CAPEX free, since the servers themselves are already purchased. Every Prodigy-provisioned data center, by definition, becomes a low-cost AI center of excellence, and a low-cost HPC system.