ENLIGHT™ is a deep learning accelerator IP technology that delivers unrivaled compute density and energy efficiency. ENLIGHT™ NPU IP accelerates inferencing computation with excellent efficiency in terms of power/performance/area. It targets +10x compute density (performance/area) and +10x power efficiency (performance / power) vs comparable GPU
Key Advantages
• Mixed Precision Processing
. Scalable Performance
. Min to 4x compute density & low-power consumption with 4-bit MAC (configurable)
•High Effective MAC Utilization
. NN optimization techniques result in sparse neural network (NN) architecture
. Optimized for efficient sparse NN handling
•Low Power Consumption
. +10x power efficiency (performance / power) vs comparable GPU
•Programmable DNN Accelerator Engine
. Toward higher computational efficiency while improving prediction accuracy
. Support most recent NN development for SR
