LoopTree: Exploring the Fused-Layer Dataflow Accelerator Design Space

Latency and energy consumption are key metrics in the performance of deep neural network (DNN) accelerators. A significant factor contributing to latency and energy is data transfers. One method to reduce transfers or data is reusing data when multiple operations use the same data. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Fused-layer accelerators</i> reuse data across operations in different layers by retaining intermediate data in on-chip buffers, which has been shown to reduce energy consumption and latency. Moreover, the intermediate data is often tiled (<italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">i.e.</i>, broken into chunks) to reduce the on-chip buffer capacity required to reuse the data. Because on-chip buffer capacity is frequently more limited than computation units, fused-layer dataflow accelerators may also recompute certain parts of the intermediate data instead of retaining them in a buffer. Achieving efficient trade-offs between on-chip buffer capacity, off-chip transfers, and recomputation requires systematic exploration of the fused-layer dataflow design space. However, prior work only explored a subset of the design space, and more efficient designs are left unexplored. In this work, we propose (1) a more extensive design space that has more choices in terms of tiling, data retention, recomputation and, importantly, allows us to explore them <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">in combination</i>, (2) a taxonomy to systematically specify designs, and (3) a model, LoopTree, to evaluate the latency, energy consumption, buffer capacity requirements, and off-chip transfers of designs in this design space. We validate our model against a representative set of prior architectures, achieving a worst-case 4% error. Finally, we present case studies that show how exploring this larger space results in more efficient designs (<italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">e.g.</i>, up to a 10<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\times$</tex-math></inline-formula> buffer capacity reduction to achieve the same off-chip transfers).