Gemmini: Enabling Systematic Deep-Learning Architecture Evaluation via Full-Stack Integration
This addresses the problem of inaccurate performance assessment for DNN accelerator designers by enabling systematic, cross-stack evaluation, though it is incremental in integrating existing components into a full-stack framework.
The paper tackles the challenge of evaluating DNN accelerators in isolation by presenting Gemmini, an open-source full-stack accelerator generator that captures system-level effects like resource contention and OS overheads, resulting in fabricated accelerators achieving up to three orders-of-magnitude speedups over CPUs on DNN benchmarks.
DNN accelerators are often developed and evaluated in isolation without considering the cross-stack, system-level effects in real-world environments. This makes it difficult to appreciate the impact of System-on-Chip (SoC) resource contention, OS overheads, and programming-stack inefficiencies on overall performance/energy-efficiency. To address this challenge, we present Gemmini, an open-source*, full-stack DNN accelerator generator. Gemmini generates a wide design-space of efficient ASIC accelerators from a flexible architectural template, together with flexible programming stacks and full SoCs with shared resources that capture system-level effects. Gemmini-generated accelerators have also been fabricated, delivering up to three orders-of-magnitude speedups over high-performance CPUs on various DNN benchmarks. * https://github.com/ucb-bar/gemmini