The Stillwater KPUTM is based on a different execution model than CPUs or GPUs. Perfectly aligned with VLSI technology and intelligent systems algorithms, the KPU can yield 50-100GFlops/Watt for machine learning applications.
The Stillwater OpenKL platform is your insurance against a changing hardware universe. With multi-core, many-core, DSP, VLIW, and FPGAs, you can rest assured your application does not need to change.
The Stillwater Domain Flow ArchitectureTM is designed to be tailored to the workload so that no extraneous hardware is wasting energy. By parameterizing the data path to the workload DFAs minimize power consumption.
Ranging from mWatts to millions of cores, from micro-scale to peta-scale, the Stillwater KPUTM was designed to create the highest performance per Watt to deliver intelligence at any scale.
Just shoot us an email and we'll be glad to give you a hand with anything you need. Or just say hi!
Leveraging new technology is always difficult. Be assured that the Stillwater team is here to assist you along the way. From software and application infrastructure, to system integration and quality assurance.