LLMulator: Generalizable Cost Modeling for Dataflow Accelerators with Input-Adaptive Control Flow

TL;DR

LLMulator is a novel framework that uses large language models and reinforcement learning to accurately predict performance of dataflow accelerators across diverse architectures and input-dependent control flows.

Contribution

It introduces a hardware- and application-aware numeric modeling approach with input-adaptive calibration and data augmentation for improved generalization.

Findings

Reduces cycle prediction error by 9.7% with dynamic calibration.

Converges to 11.2% error after few iterations.

Significantly improves prediction accuracy across architectures.

Abstract

Accurate and fast performance prediction for dataflow-based accelerators is vital for efficient hardware design and design space exploration, yet existing methods struggle to generalize across architectures, applications, and input-dependent control flows. We present LLMulator, a progressive numeric modeling framework leveraging the program semantic knowledge of pre-trained large language models (LLMs) for robust, hardware- and application-aware prediction. Our numeric model treats performance values as categorical token sequences, enabling range-agnostic estimates and confidence-aware predictions for unseen applications. To handle input-dependent control flows, we introduce a reinforcement learning-based dynamic calibration method, reducing cycle prediction error by 9.7% over static models and converging to 11.2% error after a few iterations. For cross-hardware generalization, we develop a progressive data augmentation strategy that generates diverse datasets covering multi-level dataflow structures, memory parameters, and loop mapping primitives, significantly boosting prediction accuracy across architectures and configurations.