Emerging applications for Artificial Intelligence (AI) are driving innovations in the technologies that are used to develop server- and edge-side solutions. Based on the industry-proven SuperFlash® technology, our memBrain™ neuromorphic memory product stores the neural net weights and uses an Analog Compute In-Memory capability to perform analog Vector Matrix Multiplication (VMM) for neural network computation. This process significantly reduces latency, power consumption, and cost over a digital, CPU/DRAM-based approach.
- Very low-power solution
- Reduces system latency
- Significant cost advantages over digital implementations