AI Accelerator for Embedded Devices
BrainChip Holdings has announced the AKD1500 co-processor, capable of delivering up to 800 giga operations per second (GOPS) while operating below 300 milliwatts.
This provides a power-efficient solution for edge AI tasks, making it suitable for use in battery-powered or thermally constrained systems such as wearable electronics, smart sensors, and compact embedded devices.
- Delivers up to 800 GOPS
- Operates below 300 milliwatts
- Connects to x86, ARM, and RISC-V platforms via PCIe or serial interfaces
The co-processor allows system developers to add neuromorphic processing capabilities to existing SoCs or embedded microcontrollers without requiring full hardware redesign.
No direct quote available in the text.
Author's summary: BrainChip's AKD1500 co-processor enables efficient AI processing.
More News
- Orange-Colored Lichens Helping Paleontologists Discover Dinosaur Fossils | Sci.News
- Prop 50 Could Affect California and Future Elections – What Do You Think? | News Channel 3-12
- AI accelerator supports on-chip learning for embedded devices
- Federal Budget Signals Support for CBSA Customs Services at Hamilton Port
- It’s the Specification, Stupid! – Communications of the ACM
- AR glasses get cheaper as the next platform race begins