Lennart Bamberg

Lennart Bamberg, Ardalan Najafi, Alberto Garcia-Ortiz

Specialized compute blocks have been developed for efficient DNN execution. However, due to the vast amount of data and parameter movements, the interconnects and on-chip memories form another bottleneck, impairing power and performance. This work addresses this bottleneck by contributing a low-power technique for edge-AI inference engines that combines overhead-free coding with a statistical analysis of the data and parameters of neural networks. Our approach reduces the interconnect and memory power consumption by up to 80% for state-of-the-art benchmarks while providing additional power savings for the compute blocks by up to 39%. These power improvements are achieved with no loss of accuracy and negligible hardware cost.

Lennart Bamberg, Filippo Minnella, Roberto Bosio et al.

Neural Processing Units (NPUs) are key to enabling efficient AI inference in resource-constrained edge environments. While peak tera operations per second (TOPS) is often used to gauge performance, it poorly reflects real-world performance and typically rather correlates with higher silicon cost. To address this, architects must focus on maximizing compute utilization, without sacrificing flexibility. This paper presents the eIQ Neutron efficient-NPU, integrated into a commercial flagship MPU, alongside co-designed compiler algorithms. The architecture employs a flexible, data-driven design, while the compiler uses a constrained programming approach to optimize compute and data movement based on workload characteristics. Compared to the leading embedded NPU and compiler stack, our solution achieves an average speedup of 1.8x (4x peak) at equal TOPS and memory resources across standard AI-benchmarks. Even against NPUs with double the compute and memory resources, Neutron delivers up to 3.3x higher performance.

Lion Mueller, Alberto Garcia-Ortiz, Ardalan Najafi et al.

Integer AI inference significantly reduces computational complexity in embedded systems. Quantization-aware training (QAT) helps mitigate accuracy degradation associated with post-training quantization but still overlooks the impact of integer rescaling during inference, which is a hardware costly operation in integer-only AI inference. This work shows that rescaling cost can be dramatically reduced post-training, by applying a stronger quantization to the rescale multiplicands at no model-quality loss. Furthermore, we introduce Rescale-Aware Training, a fine tuning method for ultra-low bit-width rescaling multiplicands. Experiments show that even with 8x reduced rescaler widths, the full accuracy is preserved through minimal incremental retraining. This enables more energy-efficient and cost-efficient AI inference for resource-constrained embedded systems.

Shereef Helal, Alberto Garcia-Ortiz, Lennart Bamberg

Leveraging high degrees of unstructured sparsity is a promising approach to enhance the efficiency of deep neural network DNN accelerators - particularly important for emerging Edge-AI applications. We introduce VUSA, a systolic-array architecture that virtually grows based on the present sparsity to perform larger matrix multiplications with the same number of physical multiply-accumulate MAC units. The proposed architecture achieves saving by 37% and 68% in area and power efficiency, respectively, at the same peak-performance, compared to a baseline systolic array architecture in a commercial 16-nm technology. Still, the proposed architecture supports acceleration for any DNN with any sparsity - even no sparsity at all. Thus, the proposed architecture is application-independent, making it viable for general-purpose AI acceleration.

Lennart Bamberg, Arash Pourtaherian, Luc Waeijen et al.

Manufacturing-viable neuromorphic chips require novel computer architectures to achieve the massively parallel and efficient information processing the brain supports so effortlessly. Emerging event-based architectures are making this dream a reality. However, the large memory requirements for synaptic connectivity are a showstopper for the execution of modern convolutional neural networks (CNNs) on massively parallel, event-based (spiking) architectures. This work overcomes this roadblock by contributing a lightweight hardware scheme to compress the synaptic memory requirements by several thousand times, enabling the execution of complex CNNs on a single chip of small form factor. A silicon implementation in a 12-nm technology shows that the technique increases the system's implementation cost by only 2%, despite achieving a total memory-footprint reduction of up to 374x compared to the best previously published technique.