Bolt: Bridging the Gap between Auto-tuners and Hardware-native Performance

TL;DR

Bolt introduces a novel approach that combines auto-tuners and hardware-native libraries, leveraging modular vendor libraries to significantly improve tensor operation performance and auto-tuning efficiency on GPUs.

Contribution

Bolt presents a hardware-native templated search method that bridges the gap between auto-tuners and vendor libraries, enabling faster and more efficient tensor program optimization.

Findings

Achieves 2.5x faster inference speed on CNNs

Auto-tunes models within 20 minutes

Demonstrates effectiveness on NVIDIA GPUs in production environments

Abstract

Today's auto-tuners (e.g., AutoTVM, Ansor) generate efficient tensor programs by navigating a large search space to identify effective implementations, but they do so with opaque hardware details. Thus, their performance could fall behind that of hardware-native libraries (e.g., cuBLAS, cuDNN), which are hand-optimized by device vendors to extract high performance. On the other hand, these vendor libraries have a fixed set of supported functions and lack the customization and automation support afforded by auto-tuners. Bolt is based on the recent trend that vendor libraries are increasingly modularized and reconfigurable via declarative control (e.g., CUTLASS). It enables a novel approach that bridges this gap and achieves the best of both worlds, via hardware-native templated search. Bolt provides new opportunities to rethink end-to-end tensor optimizations at the graph, operator, and model levels. Bolt demonstrates this concept by prototyping on a popular auto-tuner in TVM and a class of widely-used platforms (i.e., NVIDIA GPUs) -- both in large deployment in our production environment. Bolt improves the inference speed of common convolutional neural networks by 2.5x on average over the state of the art, and it auto-tunes these models within 20 minutes.