June 1, 2021 — Tachyum today announced it has successfully improved performance, density and power consumption of its Prodigy Universal Processor by a process (die) shrink to 5nm (nanometer) feature size from 7nm. Achieving 5nm enables Prodigy to scale at the speed of transistors, not the wire traces, and reach much higher frequency than previously announced speeds of 4GHz on 7nm. It also enables Tachyum to instantiate 128 cores in a low-cost monolithic die.
The effort was accomplished with the critical support of Synopsys, a global leader in electronic design automation (EDA) and semiconductor IP, which provided fundamental IP along with optimized 5nm logic libraries, according to Tachyum’s requirements. In addition to the benefits of the process shrink, Tachyum was able to integrate the Synopsys DesignWare IP quickly because Tachyum engaged in Customer Owned Tooling (COT) flow.
“Scaling the Prodigy Universal Processor to the 5nm process node brings industry-leading performance and power efficiency to the next generation of data center, AI, and HPC workloads,” said Dr. Radoslav Danilak, Tachyum founder and CEO. “The high-quality Synopsys DesignWare IP, including their broad portfolio of Logic Libraries, was essential to reducing our risk of moving to a new process technology, while meeting our performance, power, area and time-to-market goals.”
The successful process shrink to 5nm is another milestone in Tachyum’s effort to build the world’s fastest AI supercomputer by 2022, and to facilitate the entry of humanity into an era of human brain-scale AI and beyond.
Tachyum’s Prodigy can run HPC applications, convolutional AI, explainable AI, general AI, bio AI, and spiking neural networks, plus normal data center workloads, on a single homogeneous processor platform using existing and standard programming models. Without Prodigy, data center customers must use a combination of CPUs, GPUs, TPUs and other accelerators for these different workloads, creating inefficiency, expense and the complexity of maintaining separate hardware infrastructures. Using specific hardware dedicated to each type of workload (e.g., data center, AI, HPC) results in the significant underutilization of hardware resources and more challenging programming, support and maintenance environments. Prodigy’s ability to seamlessly switch among these various workloads dramatically changes the competitive landscape and drastically improves data center economics.
Tachyum is disrupting data center, HPC, and AI markets by providing the world’s first Universal Processor, with industry leading performance, cost and power, across all three computational domains, while, at the same time, enabling data centers to exceed the capacity of the human brain. Tachyum, Co-founded by Dr. Radoslav Danilak, and its flagship product Prodigy, begins high-rate production in 2021, with software emulations and an FPGA-based emulator available to early adopters. It is targeting a $50B market, growing at 20% per year. With data centers currently consuming over 3% of the planet’s electricity, predicted to be 10% by 2025, the ultra-low power Prodigy Universal Processor is critical, if we want to continue doubling worldwide data center capacity every four years. Tachyum is one of the founding members of I4DI (Innovations for Digital Infrastructure), which will build the world’s fastest AI supercomputer in Slovakia showcasing Prodigy. Tachyum has offices in the USA and Slovakia, EU. For more information, visit https://www.tachyum.com/.