Ambient Scientific Inc. (Santa Clara, Calif.) has announced the availability of its Digant.ai development platform for its family of AI-edge processors.
The Digant.ai platform is a hardware and software solution that is supported by the GPX-10 processor and the DigAn AI (digital-analog) computing architecture. DigAn computing includes a matrix-computer, 3D memories and asynchronous clocking schemes.
The Digant.ai platform includes a sensor fusion data path that includes an ADC that can collect up to a million samples per second from multiple sensors simultaneously yet consumes only 5-microwatt of power with a rate of 20,000 samples per second.
The DigAn architecture can scale from five cores to 8000 cores and from 40nm technology to 7nm or even 5nm process technology. The products in the family, the GPX-10 and GPX-5 are available. The GPX-10 includes 10 DigAn AI cores, 8 simultaneous sensor inputs, ultra-low-power ADC, custom memories, ARM M4F processor and numerous peripherals.
The GPX-10 can be dynamically programmed to operate with operand resolutions from 4-bits to 32-bits, to balance power and accuracy. Based on application requirements, Digant.AI optimizations enable inference operations with power consumption as low as 80-microwatt for voice and sensor-fusion applications.
“Intelligent edge IoT applications are growing rapidly and looking for out-of-box solutions for battery operated devices, the growth limited only by lack of sufficiently low power consumption,” said GP Singh, founder and CEO at Ambient Scientific, in a statement. “Digant.AI, with AI processors like GPX-10 and GPX-5, combined with our advanced Software Development Kit (SDK) and Evaluation Kit (EVK), enable lowest-power AI for voice, computer vision, and Sensor Fusion applications.” Singh added.
Ambient Scientific is a fabless chip company develeoping AI processors for edge and on-device AI applications and founded in 2017.
Related links and articles:
News articles:
Ambarella shows transformers on edge AI chip
CES 2024: Etron enables cutting-edge AI-based technologies
Infineon selects Polyn analog neural network for TPMS
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
eeNews on Google News
