Adelia: A 4nm LLM Accelerator with Streamlined Dataflow and Dual-Mode Parallelization for Efficient Generative AI Inference

This paper presents Adelia, an efficient inference chip for large language models (LLMs) featuring a streamlined data-flow and dual-mode parallelization. The streamlined dataflow directly connects the external memory to Adelia's LLM-optimized compute engine with matched bandwidth, achieving an effective memory bandwidth usage of up to 91%. The systolic path between multiple engines facilitates data reuse to enhance computational power without compromising efficiency. Adelia dynamically transitions between context mode, which distributes the long context of a single request to optimize latency, and batch mode, which processes inputs from different requests to prioritize throughput based on the runtime status. Adelia is fabricated in 4nm technology and occupies $5.28\text{mm}^{2}$ in die area. Compared to the H100 GPU, it achieves 1.59x and 2.51x greater memory bandwidth efficiency and throughput efficiency, respectively, across various models.