Breakthrough AI-powered Edge Processor Outperforms 32-Bit Microcontrollers by a Factor of 100 in Performance
Ambient Scientific Unveils GPX10 Pro: Revolutionary AI-Native SoC
Ambient Scientific, a leading innovator in AI-native silicon technology, has announced the launch of their latest system-on-chip (SoC), the GPX10 Pro. This highly integrated SoC promises up to 100x improvements in power, performance, and area compared to conventional 32-bit microcontrollers, making it ideal for edge and endpoint devices, even those powered by a single coin cell battery.
The GPX10 Pro is built on Ambient Scientific's proprietary DigAn silicon architecture, which enables local AI inference in these devices. The SoC features an Arm Cortex-M4F CPU core for classic control functions, along with integrated analog functionality such as an ultra-low power ADC, enhanced I2S logic, and interfaces for up to eight simultaneous analog and 20 digital sensors.
The AI processing in the GPX10 Pro is performed in two sets of five MX8 AI cores, each capable of up to 2,560 multiply-accumulate (MAC) operations per cycle. This high performance is demonstrated at microwatts of power, setting a new standard for AI performance in battery-powered devices.
One of the key advantages of the GPX10 Pro is its on-chip SRAM, which is a massive 2MB, ten times more than in the existing GPX10. This increased memory capacity allows for more complex AI models to be run on the device, expanding its capabilities significantly.
Ambient Scientific also provides the Nebula AI enablement toolchain for the GPX10 and GPX10 Pro, compatible with leading model training frameworks including TensorFlow, Keras, and ONNX. This toolchain allows the MX8 cores in the GPX10 Pro to be programmed, giving designers flexibility to adapt to evolving AI model types and topologies.
The GPX10 Pro performs common edge AI functions such as voice recognition, keyword spotting, low-frequency computer vision, and intelligent sensing faster and at lower power than today's MCUs, NPUs, or GPUs. GP Singh, CEO of Ambient Scientific, compared the performance of today's MCUs and NPUs in running AI models to hitting a baseball with a tennis racket, suggesting they are not well-suited for the task.
At booth 5.E31 at Electronica (17-19 September, Bangalore, India), Ambient Scientific is demonstrating various on-device AI applications running on coin cell batteries, showcasing what is possible when a system is built natively for AI. The GPX10 Pro truly represents a leap forward in AI-powered edge devices, offering unprecedented power, performance, and area efficiency.
Read also:
- Hematology specialist and anemia treatment: The role of a hematologist in managing anemia conditions
- Southeast Asia's initial AI-linked battery-swapping station was launched by U Power
- Artificial Intelligence with independent agency could potentially intervene in cybercrises.
- Mercedes unveils its new electric CLA model, equipped with an advanced AI assistant
