Search papers, labs, and topics across Lattice.
This paper introduces ATLAS, a full-stack simulation framework for 3D-DRAM-based LLM accelerators, addressing the lack of publicly available performance analysis tools for this emerging hardware. ATLAS provides unified abstractions for 3D-Accelerator architecture and programming primitives, validated against real silicon with high accuracy (≤8.57% simulation error). Design space exploration using ATLAS reveals key insights for optimizing both 3D-DRAM memory systems and accelerator microarchitectures across diverse LLM inference scenarios.
Open-source ATLAS unlocks rapid, accurate co-design of 3D-DRAM memory systems and LLM accelerators, previously hindered by closed-source tools and customized designs.
Large language models (LLMs) exhibit memory-intensive behavior during decoding, making it a key bottleneck in LLM inference. To accelerate decoding execution, hybrid-bonding-based 3D-DRAM has been adopted in LLM accelerators. While this emerging technology provides strong performance gains over existing hardware, current 3D-DRAM accelerators (3D-Accelerators) rely on closed-source evaluation tools, limiting access to publicly available performance analysis methods. Moreover, existing designs are highly customized for specific scenarios, lacking a general and reusable full-stack modeling for 3D-Accelerators across diverse usecases. To bridge this fundamental gap, we present ATLAS, the first silicon-proven Architectural Three-dimesional-DRAM-based LLM Accelerator Simulation framework. Built on commercially deployed multi-layer 3D-DRAM technology, ATLAS introduces unified abstractions for both 3D-Accelerator system architecture and programming primitives to support arbitrary LLM inference scenarios. Validation against real silicon shows that ATLAS achieves $\le$8.57% simulation error and 97.26-99.96\% correlation with measured performance. Through design space exploration with ATLAS, we demonstrate its ability to guide architecture design and distill key takeaways for both 3D-DRAM memory system and 3D-Accelerator microarchitecture across scenarios. ATLAS will be open-sourced upon publication, enabling further research on 3D-Accelerators.
