KAPLA: Pragmatic Representation and Fast Solving of Scalable NN Accelerator Dataflow
This work addresses the challenge of fast design exploration for dataflow in NN accelerators, which is crucial for hardware efficiency, though it appears incremental in improving existing methods.
The paper tackles the problem of representing and optimizing dataflow scheduling for scalable neural network accelerators, proposing a solver called KAPLA that achieves within 2.2% and 7.7% energy overheads compared to optimal schemes for training and inference, respectively, with orders of magnitude faster search speed.
Dataflow scheduling decisions are of vital importance to neural network (NN) accelerators. Recent scalable NN accelerators support a rich set of advanced dataflow techniques. The problems of comprehensively representing and quickly finding optimized dataflow schemes thus become significantly more complicated and challenging. In this work, we first propose comprehensive and pragmatic dataflow representations for temporal and spatial scheduling on scalable multi-node NN architectures. An informal hierarchical taxonomy highlights the tight coupling across different levels of the dataflow space as the major difficulty for fast design exploration. A set of formal tensor-centric directives accurately express various inter-layer and intra-layer schemes, and allow for quickly determining their validity and efficiency. We then build a generic, optimized, and fast dataflow solver, KAPLA, which makes use of the pragmatic directives to explore the design space with effective validity check and efficiency estimation. KAPLA decouples the upper inter-layer level for fast pruning, and solves the lower intra-layer schemes with a novel bottom-up cost descending method. KAPLA achieves within only 2.2% and 7.7% energy overheads on the result dataflow for training and inference, respectively, compared to the exhaustively searched optimal schemes. It also outperforms random and machine-learning-based approaches, with more optimized results and orders of magnitude faster search speedup.