Research

Research focus: Energy Efficient hardware research for artificial Intelligence(AI) algorithm acceleration;

1) Lightweight Large language Model (transformer) through

• Algorithm advancement ( Fused operation, sparsity, numerical precision and quantization)

• Hardware advancement ( Optimize data movement between memory and logic, efficient data packing)

2) Custom hardware for AR/VR algorithm acceleration

• Neural Radiance Field (Nerf) hardware acceleration

3) Neuromorphic computing ( ex. Spiking Neural Network ) with new memory devices

• Apply Process-In-Memory with new memory devices to architect ultra low power system for edge AI products. 

Any research topic related to AI hardware can be considered as our future research area. Please feel free to contact us!